Archive
Archive |
|
Persuasive Design Techniques in the Attention Economy
User Awareness Theory and Ethics
by Devangi Vivrekar
A thesis submitted in partial fulfillment
of the requirements for the degree of
Master of Science
at Stanford University
June 6 2018
1
2018
Devangi Vivrekar
All Rights Reserved
2
To the Directors of the Program on Symbolic Systems I certify that I have read the thesis of
Devangi Vivrekar in its final form for submission and have found it to be satisfactory for the
degree of Master of Science
Signed Electronically
06062018
Professor James Landay Principal Advisor
Computer Science
To the Directors of the Program on Symbolic Systems I certify that I have read the thesis of
Devangi Vivrekar in its final form for submission and have found it to be satisfactory for the
degree of Master of Science
Signed Electronically
06062018
Professor Alia Crum Second Reader
Psychology
3
Acknowledgements
I would like to express my sincere gratitude to Professor James Landay my primary advisor for
overseeing this thesis over the course of my coterm and to Professor Alia Crum for being my second reader
I would also like to thank Grace Wu and Gobi Dasu who acted as coauthors on the study described in
Chapter 3 and Professor Michael Bernstein and Geza Kovacs for their advice and feedback on the work done
in Chapter 3 with the Chrome extension HabitLab My sincere thanks to Damon Horowitz and Mikey Siegel
instructors in my Symbolic Systems coursework who deeply influenced my thinking early on and Tristan
Harris for inspiring me to study this topic Finally I would like to thank my parents and sister for their
support throughout my academic journey
4
Table of Contents
Chapter 1 Introduction 5
Chapter 2 Related Work 6
21 The Attention Economy 6
22 Persuasive Technology 7
23 Cognitive Effects 8
24 Metaphysics of Related Modes of Influence 10
25 Persuasive Strategies 12
26 Persuasive Design Techniques 14
Chapter 3 Nudget System Design for User Awareness 16
31 Introduction 16
32 System 17
33 Evaluation 20
34 Results 20
35 Discussion 22
36 Summary 23
Chapter 4 Taxonomy of Persuasive Design Techniques 24
41 Motivation and Goals 24
42 Mapping Persuasive Design Techniques onto the Fogg Behavioral Grid 24
43 Design Space Dimensions 36
Chapter 5 Implications Ethics and the Philosophy of Technology 40
51 Structure and Agency 40
52 Ethical Persuasion 41
53 CounterArguments 45
54 Improvements 46
55 ValueDriven Design 49
56 Case Study Phenomenological Analysis of Google Maps 50
Chapter 6 Future Work and Conclusion 55
References 57
Appendix A Nudget Intervention Text 64
Appendix B Nudget Transfer Assessment 66
Appendix C LinkedIn Persuasive Design Techniques 68
5
1 Introduction
The systematic study of persuasion has captured researchers interest since the advent of mass
influence mechanisms such as radio television and advertising With the unprecedented growth of popular
social media applications that are ubiquitously accessible on smart devices consumers attention attitudes
and behaviors are constantly influenced by persuasive design techniques on platforms that profit by
maximizing users time spent on site We take three different approaches empirical theoretical and
philosophical to better understand the awareness mechanisms and ethicality of persuasive design
techniques in the modern attention economy
In Chapter 2 we review related work on the attention economy persuasive technology cognitive
foundations of persuasion the metaphysics of different modes of influence persuasive strategies in social
psychology and identifications of persuasive design techniques in the attention economy
Many users feel that these digital platforms draw them in and manipulate their time and behavior
but they lack a detailed awareness of persuasive design techniques in context of their use of the products In
Chapter 3 we discuss the creation and evaluation of a system that makes persuasive design techniques
visible on social media We compare the effectiveness of our system in improving user knowledge of such
techniques to traditional methods of educating users finding a significant improvement when our system is
used
Although there exists a rich social psychology literature on methods of persuasion and exploitable
cognitive biases we lack a mapping of the specific persuasive design techniques used by products like
Facebook or LinkedIn onto this persuasive space In Chapter 4 we take a theoretical approach to persuasive
design with the goals of showing how design techniques interrelate to function as influential systems and of
providing a more nuanced vocabulary with which to discuss them drawn from a range of relevant
disciplines We map our datasets of persuasive design techniques used on Facebook and LinkedIn onto
existing theoretical frameworks in behavioral design and also identify useful dimensions for future
taxonomies
Finally we find it important to include a detailed discussion of the ethical questions surrounding
persuasive technologies in the attention economy In Chapter 5 we discuss philosophical approaches to the
balance between persuasive structure and human agency and compile useful ethical heuristics to make
progress on the question of what constitutes ethical persuasive design We discuss the method of
valuedriven design as a way to create products designed for intentionfulfillment rather than
impulsefulfillment and we provide a sample phenomenological analysis of a modern product to showcase
how philosophy of technology can add value to our understanding of what technology adds and subtracts
from our lives In Chapter 6 we discuss opportunities for future work and present our conclusions
6
2 Related Work
We begin by reviewing related work on the emergence of the attention economy the foundations
of the field of study of persuasive technology the effects of digital technologies on cognition the distinctions
between modes of influence related to persuasion the social psychology account of persuasive strategies and
the identification of persuasive design techniques in digital products
21 The Attention Economy
The abundance and rapid growth in popularity of information technologies over the past two
decades has created a marketplace where consumers attention is a scarce and valuable resource Herb Simon
first identified this phenomenon now called the attention economy when he wrote about the need to
allocate that attention efficiently among the overabundance of information sources that might consume it
1 By using valuable intangibles in their design such as immediacy personalization accessibility and
findability popular digital applications constantly influence consumers attention attitudes and behaviors
2 By giving consumers the ability to share comment on and create their own content on social network
services SNS these technologies empower users to curate and reach their own audiences furthering
information abundance and constraining the resource of attention on not only an individual level but also
the scale of entire social populations 3 Persuasion scholar BJ Fogg has called the influential reach of
Facebook a social phenomenon of Mass Interpersonal Persuasion due to the culmination of persuasive
experience automated structure social distribution rapid cycle social graph and measured impact 4
These massive platforms exemplified by Facebook YouTube Snapchat Twitter and LinkedIn among
others often aim to maximize users time spent on their sites The desire to maximize the metric of time
spent on these platforms often stems from their bottom line profit from advertisements Advertisers pay the
platforms to host their ads on the sites and users views and clicks of the ads become a payment of time and
attention for their free use of the products The digital landscape then has become situated in an economy
where time and attention are the currencies The focus of this work is to analyze the design of mass
consumer digital products in the attention economy
Users around the world spend large amounts of time on these products 68 of American adults
across a wide range of ages and demographics use Facebook and 74 of them use the site daily 5 Among
1824 year old Americans 78 use Snapchat of which 71 visit the app multiple times a day 6 94 of
1824 year old Americans use YouTube 71 use Instagram and nearly half 45 use Twitter 6 The
median American uses at least three of the following eight products YouTube Facebook Instagram
Pinterest Snapchat LinkedIn Twitter and WhatsApp 6 Globally Facebook has over 2 billion monthly
active users and over 13 billion daily active users 8 For comparison this is just under the 23 billion
followers of Christianity the worlds largest religion 9 In addition to high frequencies of use these
technologies are characterized by long durations of use For example on mobile devices YouTube viewing
sessions last over 60 minutes on average and 70 of time spent on YouTube is driven by automated
recommendations according to its chief product officer 10
There are two broad categories of motivation due to which people engage with these products
instrumental and habitual Instrumental use occurs when a user approaches and engages the technology with
a specific purpose and habitual use occurs when a user engages with the technology less intentionally and
out of habit to merely pass time 11 Hiniker et al report that this dichotomy of styles of use could
7
contribute to the level of meaning drawn from experiences engaging with products Furthermore they note
that the same products can be used in both these modes of engagement in 39 of samples participants
described their Gmail use as passively getting information whereas Gmail is listed under the
Communication category on the Google Play Store a category that might suggest instrumental use 11
Snapchat is listed under Social apps which might suggest more habitual use but in 66 of samples
participants reported active engagement rather than passive social media consumption Almost all
participants in a study of meaningful smartphone use reported passively scrolling on social media as a
meaningless activity and many reported a loss of autonomy citing the habitual nature of their social media
checking habits 11 Even instrumental motivation has been found to degrade over the duration of use of
these apps leading to an erosion of intention by the very products that exist to help us intentionally
consume information and communicate 11
Technologist and design ethicist Tristan Harris describes the attention economy using the analogy of
a city whose urban planners are the large corporations eg Apple and Google that lay the infrastructure for
the rest of the applications in the market and function as monopolies that set the standard for what other
applications must do to remain competitive in harnessing user attention 12 When all applications even
meditation apps that intend to instill calm must compete with social media apps for morning attention they
join what Harris calls the arms race for attention 12 This proliferates techniques that race to the bottom
of the brainstem 12
The effects of such techniques can be seen indirectly in user sentiment and the general epidemic of
overconsumption and underfulfillment a phenomenon that has been present in many social trends
including obesity heart disease industrial pollution social isolation diabetes credit card debt and now
Internet addiction 13 Many users are dissatisfied with the time they spend using these apps especially on
their phones 11 In a study done by Google vice president of product management Sameer Samat reported
that Over 70 percent of users had told Google that they wanted help striking a balance between their digital
life and realworld interactions 14 This has recently prompted the addition of new features to its latest
operating system Android P that reflect design ethicist Harriss calls for ensuring that time spent on these
products is time well spent 14 While Facebook has not replicated this feature it has started changing its
algorithm to promote interpersonal interactions and downrank viral videos that elicit passive consumption
15 Nevertheless the problems that arise for users of these products as well as for the societies in which the
products are fundamentally embedded are far from solved Thus the attention economy has become an
important subject of research for persuasion scholars who study the design of the technology that has these
farreaching effects
22 Persuasive Technology
The subset of persuasive technology we focus on in this work is that of digital technologies in the
attention economy that intend to persuade Digital persuasive technologies differ from traditional persuasive
media in that they act as persuasive intermediaries between the persuader and the target of persuasion and
unlike billboards or static content they interact dynamically with the target 16 The technologies we
analyze are all adaptive persuasive technologies that use both endadaptive and meansadaptive strategies
meaning that they learn about the user over time to better tailor their persuasion Endadaptive strategies are
8
those where the knowledge gained of the user has minimal application to future persuasive attempts in other
domains whereas meansadaptive strategies may apply in other domains 1718
The intermediary role of these technologies is the subject of study of the field of captology an
abbreviation for computers as persuasive technologies In the spirit of captology we focus on the planned
persuasive effects of computer technologies and study both the attitudes and behavior changes intended by
designers where intent is endogenous as opposed to exogenous 19 However as Atkinson warns if the
program results in unforeseen induced behavioural changes it is appropriate for the discipline to assume
responsibility and to name such phenomena 20 The two broad kinds of persuasion conducted by digital
technologies include macrosuasion the overall persuasive intent of a product and microsuasion the
smaller persuasive elements to achieve an overall goal 19 Microsuasion is embodied through elements of
visual and interaction design such as dialogue boxes icons interaction patterns or reward strategies 19 It
is these more granular microsuasive techniques that are of interest in the majority of this work
23 Cognitive Effects
We now discuss prior work on the effects of products in the attention economy on cognition By
virtue of competing in the attention economy these products capture not only users time but also their
attention or focused mental engagements on particular sets of information 21 For example many of
these products use structured notifications to draw users in interrupting other behaviors This in some part
results in people unlocking their phones over 150 times a day 7 The distraction cost of receiving a phone
notification during attentiondemanding tasks has been shown to be comparable in magnitude to that of
actively using a mobile phone for texting or making a voice call 22 When the interruption is contextually
different from the task it causes increased stress alters work strategies and mental states and increases task
reorientation time 23 Interruption and distraction from digital technologies can have more serious
consequences as well Case studies have documented shallow breathing and temporary suspensions of breath
while checking email a phenomenon called email apnea 24 In safetycritical domains an illtimed
notification could alter response times in highstakes situations causing loss of life or catastrophic damage
25
The dualprocess model of persuasion describes two ways of processing persuasion systematically
and heuristically In systematic processing people differentiate strong and weak arguments and are not
affected by variables outside of the substance of the message such as message length or attractiveness of the
message source 26 Systematic processing is triggered by things like forewarning that a persuasive message
will be presented In heuristic processing people are susceptible to factors external to the substance of
message Heuristic processing is triggered by the presence of heuristic cues such as social proof which is a
technique commonly used by SNS 26 The design of technologies in the attention economy often includes
such stimuli that act as demandances which unlike the utilitybased concept of affordances clues about
how something should be used are psychological pulls to execute a specific action like an itch that drive
goaldirected behavior 27
These effects on users attention have an impact beyond control of momentary awareness they are
also related to goaldirected achievement and higherorder aspects of cognition relevant to selfidentity
These three levels of cognition have been described as the spotlight starlight and daylight of attention
28 The spotlight refers to perceptual attention towards tasksalient features of the environment and
9
interaction rules pertinent to those features Starlight refers to goals relevant to who a person wants to be
and how current tasks relate to those higher goals And daylight refers to metacognition or defining the
higherlevel goals in the first place By influencing our momentary attention affecting our daily attitudes
and changing our behaviors over time persuasive technologies affect all three of these cognitive modes For
example video advertisements on Instagram direct users perceptual attention towards sponsored content
while its aesthetically filtered photo feed leads to deeper attitudes like envy spiraling where users compare
themselves both subliminally and explicitly to glamorous images This in turn can lead to longerterm
changes in selfimage and by one users account negatively affect mental health and selfworth issues
1529 Persuasion ethicist James Williams similarly categorizes types of distraction in terms of the effect
they have on cognition functional which interferes with doing what you want to do and affects cognitive
processes like task or goal alignment and intention awareness existential which interferes with being who
you want to be and affects value and identity alignment workstyle regret and time management and
epistemic which interferes with wanting what you want to want and affects reflection willpower and
reasoning 30 The compounding effect of these products on higherlevel cognition can then be extended to
impact the very basis of our freedom wellbeing and even the integrity of the self which demonstrates how
these seemingly small design techniques can have large cognitive impacts when interacted with so frequently
and intimately 31
Persuasive design techniques often engage or exploit cognitive biases and psychological limits
misperceptions and fallacies 32 They trigger the same cognitive processes that for example cause people
to spend more on dinner when a restaurant is called Studio 97 instead of Studio 17 or to estimate higher
athlete performance for players with higher jersey numbers 33 These biases have been extensively studied
and documented in the field of social psychology which laid the groundwork for the field of behavioral
economics We now list some of these common biases and fallacies Biases that stem from lacking full
meaning in a situation include confabulation anthropomorphism authority bias bandwagon effect halo
effect reactive devaluation denomination effect illusion of transparency spotlight effect selfconsistency
bias hindsight bias timesaving bias and impact bias overestimating the length or intensity of future feeling
states 34 Biases stemming from the need to make decisions quickly include social desirability bias illusion
of control risk compensation actorobserver bias loss aversion endowment effect status quo bias
ambiguity bias decoy effect hyperbolic discounting irrational escalation of commitment appeal to novelty
identifiable victim effect social comparison decision fatigue ratio bias regret aversion Elaboration
Likelihood when people rely on peripheral rather than central processing and forecasting errors 3435
The overabundance of information can also trigger biases such as the availability heuristic bizarreness effect
anchoringframing priming confirmation bias bias blind spot choice overload ostrich effect
diversification bias and partitioningmental accounting 34 A final category of biases is due to memory
constraints the spacing effect fading affect bias suffix effect primacy effect peakend rule Google effect
and the nextinline effect 34
Users mental models of these platforms vary greatly In studying peoples stories about how the
Facebook algorithm ordered posts in their newsfeed Eslami et al found that people had a range of often
false folk theories that determined their mental model and interpretive stance towards the platform 36 If
users mental models do not align with the design or if technical constraints arise behind the scenes of the
product then designers of these platforms often employ some degree of what Adar et al call deception in the
user experience 32 For example when Netflixs servers are overwhelmed it seamlessly switches from a
personalized recommender system to a simpler one based on general popularity heuristics unbeknownst to
10
the user due to the lack of perceived change in the visual experience 32 Hiding or abstracting away critical
information is an important design technique that can be used both to improve users perception as well as
detract from it We now turn to what makes a particular design technique persuasive as opposed to
influential in another perhaps more negative way
24 Metaphysics of Related Modes of Influence
The study of influence is defined as the study of the cause of human change in belief attitude and
behavior 37 There exist several forms of symbolic transaction besides persuasion in which one tries to
influence others behaviors and attitudes for example manipulation and coercion Our definition of
persuasive technology ultimately casts a wide net across several of these related concepts However for
completeness we review here the ways they have been defined in the literature drawing from social
psychology political philosophy and philosophy of action 38
241 Compliance Historically compliance has been defined as inducing change in a behavior 39
Rhoads et al describe it as a quickfix solution to a social problem since it does not require user agreement
at a rational level and merely demands the performance of a behavior 39
242 Conviction Some distinguish techniques that achieve persuasion from those that achieve
conviction saying that persuasion relies more on symbolic strategies that trigger the emotions of intended
persuadees whereas conviction relies on strategies rooted in logical proof and reasoning and that appeal to
persuadees reason and intellect 26 In other words persuasion is said to cater more to irrationalities
whereas conviction caters to rationality
243 Education Education has been defined as inducing a change in beliefs and is similar in content and
technique to propaganda if we already believe in the target beliefs we tend to refer to it as education
whereas if we do not we call it propaganda 39
244 Deception One working definition of deception as it applies in humancomputer interaction is
when an explicit or implicit claim omission of information or system action occurs that is mediated by user
perception attention comprehension prior knowledge beliefs or other cognitive activity and creates a
belief about a system or one of its attributes that is demonstrably false or unsubstantiated as true where it is
likely that the belief will affect behavior or a substantial percentage of users 32 In philosophy deception
has been defined as outright lying to those manipulated including making false promises to them but also
misleading them without actually misrepresenting anything such as by encouraging false assumptions or
fostering selfdeception that is advantageous to the manipulators ends or getting the target to interpret the
situation in a light favorable to the manipulators purposes 38
245 Coercion Coercion is distinguished by its degree of constraint when a performed action is
characterized by the strong lack of an acceptable alternative a user is coerced or compelled or forced to
do it 38 However mere constraint in itself is different from coercion since constraint can exist even when
users have several acceptable options such as during automatic speed limit enforcement 40
11
246 Seduction Verbeek a philosopher of technology defines seductive technologies as those that do
not so much coerce people or persuade them to act in a certain way at the cognitive level but that simply
make some actions more attractive than others 40
247 Manipulation Manipulation has been characterized both as deceptive noncoercive influence and
as nonrational influence 38 Barnhill defines manipulation as directly influencing someones beliefs
desires or emotions such that she falls short of ideals for belief desire or emotion in ways typically not in
her selfinterest or likely not in her selfinterest in the present context 38 When influence leads to beliefs
or desires that are in someones selfinterest it is then not considered manipulative Manipulation can also be
indirect For instance evoking an emotion that does not directly induce a selfharming action but rather
evokes a state of mind that leads to decisions that are not directly in selfinterest is also manipulative if a
manipulator appeals to a cold competitive malice thats under his control then she does not manipulate
him because feeling controlled competitive malice and making decisions on the basis of it is likely to be in
his selfinterest However if she stokes a hot competitive malice thats not tightly controlled then she does
manipulate him since feeling and acting on nottightlycontrolled malice is typically not in ones
selfinterest 38 Barnhill further classifies types of manipulation as intricate vs blunt paternalistic vs
nonpaternalistic covert vs overt targeting beliefs vs emotions and changing a situation vs a person
These modes of influence are interrelated in interesting ways the use of some can amplify or lead to
the use of others and some can be seen as more intense applications of others Wood considers coercion and
manipulation to be on a continuous spectrum coercion destroys free choice by rendering all but one
option unacceptable whereas manipulation merely influences choice without removing it 38 However
some see the two as fundamentally different since a coercer typically alters the world or the product such
that seeing the altered worldview makes it rational to do what the coercer wants by the targets own lights
38 Manipulation by contrast aims to alter the targets viewpoints beliefs desires and values while
being indifferent to whether the alterations reflect what is true or desirable acting more as a puppeteer
messing with our heads 3841
Faden and Beauchamp contrast persuasion and manipulation by saying that persuasion improves
someones understanding of her situation but manipulation does not 38 They even define manipulation
in terms of a lack of the cognitive process of understanding as any intentional act that successfully influences
a person to belief or behavior by causing changes in mental processes other than those involved in
understanding 38 Mills claims that manipulation disguises itself as good persuasion by appearing to offer
valid reasons but in fact offers faulty ones even though the manipulator knows the reasons to be faulty He
defines manipulations as a persuasion manqu as an attempt at internally directed and nonphysicallybased
influence that deliberately falls short of the persuasive ideal 38
How then has persuasion been historically characterized Social psychology traditionally defines
persuasion as the process that changes attitude 37 Persuasion attempts to win the heart and mind of the
target which involves affective change it is also characterized by more longlasting effects than other forms
of influence such as compliance since the target accepts and internalizes the message to a greater degree 37
In humancomputer interaction persuasion has been described as a voluntary attempt to change attitudes or
behaviors or both whereas coercion implies force and deception implies misinformation 19 Persuasion
comes from someone elses conceptual framework 20 and involves specific intent from an outside agent
12
to cause the target to adopt previously untenable beliefs attitudes or behaviors that are foreign to their own
conceptual and behavioral repertoire 20 In this spirit we define persuasion as follows
Persuasion the mode of influence in which one agent for our purposes a digital technology acting as a
vehicle for the designer intentionally attempts to change another agents the users attitudes or behaviors
James Williams has identified even more modes of influence that persuasive technologies can have
over users which vary in degree of goal alignment and constraint these include technologies that demand
drive tempt guide invite suggest and direct 42 Moving forward we do not discard persuasive
techniques that fall under metaphysical categories besides persuasion and instead use the term persuasion as
a broad umbrella to capture techniques across these related modes of influence We default to the singular
term of persuasion because many of the other modes of influence have builtin normative connotations
and we aim to provide a descriptive and relatively nonjudgemental account of persuasive design especially
in Chapter 4 However we revisit the distinction between different kinds of influence in Chapter 5 where
we discuss the questions of what constitutes ethical persuasion and how to design for more ethically
permissible forms of persuasion
25 Persuasive Strategies
Designers of our physical world often employ persuasive strategies to encourage particular attitudes
or behaviors Although the approach of using interpersonal or nondigital persuasion to understand digital
persuasion has largely focused on computermediated communication rather than humancomputer
interaction 32 it is still insightful to review examples of realworld persuasion since their analogues are
present in digital design as well For example architectural elements in casinos such as mazelike paths the
illusion of small secluded spaces and lack of apparent exits persuade people to stay and continue spending
money 32 Visual cues indicating food portion size affect intake of items like Campbells soup and popcorn
by removing visual cues that indicate how much has been consumed or when to stop consuming people
engage more in activities that are considered mindless 43 Park benches with central armrests discourage
overnight occupation by the homeless 44 Placebo buttons such as those found in crosswalks elevators
and thermostats often provide the illusion of control without actual functionality 32 Dualbutton toilet
flush controls steer users to choose between two behaviors one that conserves water and one that does not
32
Underlying these nondigital persuasive design techniques are wellknown persuasive strategies that
have been studied and categorized in social psychology Kelton Rhoads conducted a review of landmark
compilations of persuasive strategies published since the topic gained prominence due to the advent of mass
media 39 What constitutes a persuasive strategy Marwell and Schmitt define it as the reduction of the
multitude of possible behaviors into meaningful clusters such that a strategy is a group of techniques
towards which potential actors tend to respond similarly 45 Rhoads found varying numbers of strategies
however these distinctions are often not rigid and quantitative but rather stem from the differences in level
of social psychology abstraction For example Marwell Schmitt identify several strategies in their 1967
taxonomy of the dimensions of compliancegaining behavior reward punishment positive expertise
negative expertise likingingratiation giftingpregiving debt aversive stimulation moral appeal positive
13
selffeeling negative selffeeling positive altercasting
negative altercasting positive esteem of others and
negative esteem of others 46 Sequentialrequest
compliance theory groups strategies into two kinds
footinthedoor FITD small initial request followed
by a larger but still reasonable second request and
doorintheface DITF first outrageously large
request followed by a smaller more moderate request
47 Levine Wheless identify 53 tactics in a 1990
review of previous taxonomies 48 listed in Figure 1
Robert Cialdini has taken a social psychology
and communications approach to persuasion breaking
down the psychological laws that govern persuasion
into presuasion the factors that establish trust and
set the tone for the persuasion and six psychological
principles reciprocation liking social proof authority
scarcity and consistency that represent certain
psychological universals of persuasion 49
The presuasion strategies include
1 Priming via context focusing recipients
initially on concepts that are aligned with
forthcoming information such as playing a
German song in a store to make people buy more German products
2 Associations irrelevant to goals using priming to elicit unsavory behavior such as exposing
subjects to violent language to increase the number of shocks administered
3 Mood contamination using earlier information to influence affective state relevant to later
questions or tasks such as how asking a question about happiness about social life influences
response to overall happiness levels
4 Agendasetting theory drawing on the fact that we assume earlier information points out what to
consider important in future information
5 Investigatory reflex using distinguishing factors and selfrelevance eg photos of self
unfinished arguments and unsolved problems to draw attention
The six key psychological factors in persuasion 49 include
1 Reciprocation our tendency to want to repay the favor
2 Consistency our need for internal selfunderstanding or our tendency to observe our own
behaviors to learn about ourselves which can trump rationality and is often enforced through
commitment
3 Social proof using the actions of others to decide on behavior for ourselves especially when we
view others as similar to us
4 Authority deferring to authority even in outrageous situations such as administering shocks to
other experimental participants due to the instructions of an experimenter
14
5 Scarcity the fear of losing out on access to a scarce resource
6 Liking attractiveness which often leads to a massive halo effect where associated aspects of an
attractive prospect seem more attractive For example aesthetically pleasing designs and faces
including online avatars elicit more positive response than others regardless of function 52
From an HCI perspective Adar categorizes types of deception by functional behavioral and mental
model strategies which respectively misrepresent performance information take advantage of psychological
or sensory limits of perception eg Fitts Law is applied when a dropdown menu bar that is programmed
not to roll back as the user moves a couple pixels outside of the box and use metaphors to mislead users into
thinking something works differently than it actually does 32 Once users develop a relationship with the
product they become more susceptible to truth bias believing the computer to be truthful which opens the
door to further deception this is because users are less adept at placing themselves in the mindset of a
technological system as they might be with another human 32
There is also empirical work on persuasion that tests the success of different methods on controlled
populations For example computergenerated graphics are more persuasive than text as a form of
computermediated communication 50 In environmental conservation settings focusing persuasive
messages on descriptive normative information increased undesired behavior whereas focusing messages on
injunctive normative information suppressed it 51 High source credibility has been shown to lead to more
favorable attitudes than low source credibility when people agree with the message or perceive it as having
stronger arguments without knowing the source but high source credibility has the opposite effect when
people initially disagree with the message 52
26 Persuasive Design Techniques
For better or worse products in the attention economy employ persuasive techniques in their design
that are similar to the techniques that are used in the nondigital world Digital products use these techniques
virtually everywhere from the font colors to the text that frames anything from profile completion cues to
promoted posts 53 These techniques have been called subtle and Pavlovian and include things like the
red badge on Facebooks app that signals a posted comment or message YouTube and Netflix videos that
automatically cue up to play one after the other and the Snapchat orange fireball emoji that signals a streak
of daily chats between friends 54 Now entire products exist to help websites optimize for where to place
their content to get the most clicks and shares 55
While these techniques add utility for users to some degree some techniques are considered more
insidious than others One collection of more unsavory techniques called dark patterns includes bait and
switch confirmshaming disguised ads forced continuity friendspam hidden costs misdirection
purposefully focusing your attention on one thing in order to distract your attention from another price
comparison prevention sneak into basket and trick questions 124 In fact LinkedIns old version of the
onboarding flow to add a contact spammed the users contacts without explicit permission which led to the
class action lawsuit Perkins v LinkedIn in 2015 and allowed affected users to submit a claim to get a payout of
around 10 each 124 But some techniques are less contested and even more subtle by making use of
justnoticeable differences JNDs developers can create the illusion that an action has or has not been
influenced 32 Slight variations of these techniques appear throughout all products in the attention
15
economy due to the nature of the attention economy itself For instance examining the Twitter home
screen reveals techniques that are present across all SNS Figure 2 Once one product starts for example
autoplaying videos it is more likely to increase users time spent on its platform and since attention and time
are limited resources competing products adopt successful design tricks from each other
Figure 2 The home screen of Twitter annotated with common persuasive design techniques that are also used in other
products 1 counts of unread messages 2 asymmetric dot buttons on the Home tab 3 artificially promoted trends at
the top of the trends list 4 text quantifying the amount of content to catch up on 5 animated buttons that fill with
color as users click them to share content 6 large ads formatted to look similar to other posts 7 customized
suggestions of more people to follow and 8 oneclick contact imports from other SNS
It is interesting to consider the question of how many persuasive design techniques there are We
gather data on what users of these products consider persuasive design techniques in Chapter 3 It is even
more interesting to consider how these techniques are related We address these questions in Chapter 4
16
3 Nudget System Design for User Awareness
Modern social media websites routinely employ persuasive design to orchestrate specific user
actions To raise Facebook users awareness about the behavioral design techniques and tricks used on their
cognitive resources we developed a system Nudget that makes such techniques visible during Facebook use
Nudget named loosely after the idea of nudging user awareness through microinterventions annotates
desktop Facebook feeds in realtime offering information about relevant persuasive theory at play We
conducted a 5day trial in which 15 participants used Nudget while 6 participants used traditional
stateoftheart print material to learn about persuasive design techniques We found that Nudget users
displayed a significant difference in postintervention scores of learning about the techniques compared to
the control p001 We discuss the implications of our system on the effort to educate social media users
about the web design calculated to guide their online behavior
31 Introduction
To set and accomplish goals we must have control over our own attention Modern social media
users lose agency the capacity to act completely independently and make free choices because social media
platforms implement persuasive design techniques to orchestrate target behaviors that improve the websites
bottom line In particular they employ bottomup visual and contentdriven techniques to drive up their
engagement metrics 8 It is important to recognize measure and think critically about such manipulations
and their effect on cognitive resources but we lack modern vocabularies user awareness and transparency
to do so This motivates us to design a system that elucidates the underlying social and psychological tricks
driving users interaction with social media By drawing back the curtain to show users how websites
conceive of the purpose of their design features we aim to educate users about such techniques to make them
more aware for future social media use
Past studies have shown human susceptibility to social media persuasion and a number of systems
exist to encourage users to take control of their time spent on social media MeTime was a system that spread
user awareness by offering users a glimpse at their pie chart of time spent on various SNS ie how big of a
slice of their pie each social media company owns 5758 The MeTime system showed that pure awareness
helped users better manage their time and grow stronger against social media manipulations Another system
LivingSmart showed that it is possible to use technology like focussupporting browser extensions and task
management applications to prevent digital attention manipulation even with participants with attention
deficit disorder 59 These systems demonstrate the viability of using technology to empower users against
the platforms that reduce their agency to complete their own tasks and achieve their personal goals
We extend this work by expanding the design space of our interventions beyond postuse reflection
of time spent on social media sites Current systems can tell people post hoc that they have been manipulated
and provide them with graphical displays of their time consumed by corporations Existing solutions can also
block entire sites or remove news feeds but this option is not feasible for many social media users These
systems are not able to tell users when and how they are being manipulated in real time even though it is not
the sites themselves but rather the attentiongrabbing tactics they use that lead to user attention
manipulation Our system is granular enough to highlight manipulation in real time and in visual and
17
temporal proximity to the very manifestations of persuasive theory that manipulate user attention By giving
users realtime relevant insight into the underlying persuasive theory behind the features they use we aim
to improve knowledge retention about what happens behind the scenes of what they see on the screen
Although we believe educating users about the inner workings of persuasive technology platforms is a
valuable end in itself such interventions can also provide us with intuition about which persuasive design
techniques users find most memorable and problematic
32 System
Nudget is a Google Chrome extension that annotates users Facebook feeds on their desktop
computers showing informational popup messages about persuasive theory at play insitu We discuss the
design space of our Nudget intervention messages the design principles we considered when building our
display and the implementation of our system
322 Persuasive Theory Design Space
We now describe the subset of persuasive theories that inform the text of our Nudget interventions
Instead of exposing issues about time distributions or mental health disorders on the site as covered in
previous work we focus on situating our interventions in the larger space of persuasive technology design
We draw upon Robert Cialdinis research on the effects of persuasive framing on behavior 49 as well as Nir
Eyals and BJ Foggs persuasive technology design loops 6061 Taken together these theories suggest that
all the manipulations we are interested in occur at some point along the following model that describes user
behavior motivation ability and prompt 64 Motivation in a digital context refers to sensation anticipation
and belonging emotional needs ability refers to reducing the barrier to access for the target behavior and
prompt refers to deadlines or cues to make the target behaviors urgent For example getting and giving likes
as well as the Show Comments feature is an instance of motivation the social and emotional factors
relating to the need to be heard and make themselves heard which motivates people to generate and share
content Autoplaying a video from a post is an example of lowering ability needed to complete the target
behavior of viewing the video because it reduces the number of clicks needed from 1 to 0 The scrollable
everupdating side feed is a prompt a deadline for reading information now before it is forever swept away
to disappear The Home and Notification buttons are landing points for prompts such as the bright red
numerical count of unread updates because they are affixed to the top of the page and always noticed The
action flow from motivation to ability to prompt gets people to repeat the behavior on a fixed schedule This
is described by the following cognitive flow acquisition using external prompt creating desire and affixing
the internal prompts in other words attention comprehension elaboration integration and enduring
attitude change For example people come to Facebook out of habit boredom or loneliness prompts
external to the site and after the login have desires created by the motivators present onsite creating
desireselaboration have these desires fulfilled by manipulations onsite integration which completes the
cycle of these created desires being affixed to internal prompts leading to enduring attitude change over
time
18
323 Design Principles
To teach users about these complex persuasive theories we broke down the interconnected webs of
manipulation into featurespecific instances of persuasion Please see Appendix A for our final list of
informational Nudget messages called Nudgets In designing our messaging we wanted to balance a
desired level of knowledge with short attention spans and the rapid scrolling behavior users display on
Facebook We also did not want to disrupt the natural Facebook experience by demanding extensive reading
to the extent that users would uninstall our extension Via informal usability testing we determined that
users would tend to read 3 Nudgets on one screen We also realized that the optimal text length of Nudgets
would be about Tweetsized 280 characters and we included a twoword tagline on each Nudget to allow
for rapid reading A drawback inherent in our design is that we too use a bottomup attentiongrabbing
technique when displaying these bright popup messages However we found in usability testing that
without the popup animation or the bright yellow color users would not even notice our messages due to
the variety of elements already on their Facebook feeds as well as their habit of focusing intensely on the
central feed itself Thus we employed our own subset of persuasive design techniques to even be noticed in
the vicinity of Facebooks display We anticipated that this could make Nudgets slightly disruptive for some
users even while educating them
324 Implementation
We built the Nudget system on top of the HabitLab project out of the Stanford HCI Group 62
HabitLab itself is a Chrome extension that helps users gain control of their browsing We implemented our
custom intervention in Javascript and jQuery packaged it into a fork of HabitLab and deployed it to the
Chrome store for our participants to install
Upon a user scroll event on the Facebook feed the system runs Nudget as long as Nudget has not
already run within the last 4 hours The system tracks divs of the various Facebook manipulations and
annotate those divs with 3 Nudget descriptions of the manipulative technique These bitesized chunks of
information appear with arrows pointing to the element on the screen to which the message refers The
annotations appear for 15 seconds one by one in the form of open popup boxes and then close up and move
to the top left corner of the screen where the arrows in the messages change into numbers to avoid spatial
mislabelling as users scroll Nudget annotations are clickable so that users can toggle an annotation between
open and closed states
Some examples of Nudgets are shown in Figure 3 and Figure 4 The first uses jQuery selectors to
find the notification button and then display the following text next to it Quantified FOMO Facebook
shows a number to tell you exactly how many important things youre missing out on What is actually
important Figure 3 The second uses selectors to find the small access point to settings in the top right of
the Facebook page and annotate the following text there Dont Leave Compare how easy it is to see your
feed vs deactivate your account in settings Facebook purposefully makes it hard for you to exit Figure 4
For a closeup of another contextual Nudget see Figure 5 To try Nudget on Google Chrome download it
here bitlynudget
19
20
33 Evaluation
To evaluate the effectiveness of the Nudgets as a system we conducted a 5day betweensubjects
study with 15 participants in the treatment group and 6 participants in the control group We recruited
participants through our personal networks and HCI class Both the treatment and control groups completed
two questionnaires at the beginning of the study The first questionnaire was a survey about their usage and
opinions of social media websites The second questionnaire see Appendix B was an openended quiz that
asked participants to identify manipulation techniques on three screenshots of LinkedIn a social networking
site that overlaps with Facebook in the design techniques used
After taking these surveys the treatment group used the Nudget system for a week on Facebook
while the control group read articles on persuasive techniques used by social media websites At the end of
the study both the treatment and control groups were given the same two questionnaires again We
hypothesized that if our system were successful it would increase the number of techniques identified by
people in the treatment group after the intervention significantly more than the number identified by people
in the control group
We used the first questionnaire to control for preexisting attitudes towards social media sites
explained further in the next section We used the second questionnaire to measure changes in knowledge
of persuasive techniques Specifically for each participant we calculated the difference in the number of
relevant persuasive techniques they identified across the screenshots of LinkedIn before and after the
intervention we used this difference to calculate two scores of awareness 1 the Nudgetspecific awareness
assessment score which we obtained by dividing the difference by 22 the size of the set of persuasive
techniques covered in Nudget interventions and 2 the general awareness assessment score which we
obtained by dividing the difference by 171 the size of the pool of all 171 unique persuasive design techniques
identified by the participants
In this way since the participants were instructed about the use of the techniques on Facebook and
tested on examples from LinkedIns our calculated scores parameterized transfer knowledge the
knowledge that transfers across platforms The
Nudgetspecific score allowed us to grade
participants in context of their learning with the
system and the general score allowed us to get a
sense of their performance when graded against a
more complete gold standard set of possibilities See
Appendix D for the full list of 171 techniques
34 Results
We now present both quantitative and
qualitative results measuring the effect Nudget had
on our participants awareness of design techniques
21
341 Quantitative Evaluation
We performed two statistical evaluations to determine whether condition was a significant predictor
of transfer knowledge scores
3411 Transfer Knowledge Evaluation I
We found that the condition participants were placed in experimental or control mattered
significantly in our assessment of transfer knowledge Namely a ttest yielded a significant difference
p001 between the postintervention transfer scores of the experimental group which was exposed to
Nudget and the control group Figure 6
We obtained the postintervention scores used in this test by grading against knowledge of
manipulations presented by Nudget not against the general answer pool While we also computed a
postintervention score graded against the general answer pool the system was not able to significantly
improve performance for the experimental group The ttest between postintervention general transfer
scores of the two conditions had a pvalue of 04
3412 Transfer Knowledge Evaluation II
We also performed a regression to predict participants postintervention Nudgetspecific awareness
assessment score using the following controls age how much users likeddisliked Facebook condition
experimental vs control preintervention general score and preintervention Nudgetspecific score We
found that condition was indeed a significant predictor p0023 Figure 7 However when we regressed
postintervention general awareness assessment scores we did not find that condition was a significant
predictor p0056 Figure 8
22
We note that in Figure 7 and Figure 8 the variable facebook_liking was split by our statistics software into
two variables one indicating whether the user liked Facebook or not before the intervention and the
other indicating whether or not the user had no feelings toward Facebook
342 Qualitative Evaluation
Participants had positive responses to the Nudget system When asked if they would rather learn
about the topics through Nudgets or reading articles 667 responded that they preferred the Nudgets
Many participants said the interactive and personal aspects of Nudgets made it more relevant and enjoyable
to learn about the topics They thought that seeing concrete examples of persuasive design techniques in a
product they used daily was more useful than reading an article One participant remarked in the
postquestionnaire Having the Nudgets appear on Facebook made the experience more interactive and it resonated
on a more personal level for me Sometimes when I read an article about these things I tend to distance those experiences
from my own experiences Seeing the Nudgets appear on my personal account allowed me to see just how affected I
personally am by Facebooks persuasive design techniques
However many participants also had suggestions for ways to improve the Nudgets Many said that
the Nudgets were repetitive and reported seeing the same Nudgets many times Others suggested that the
Nudgets should be more contextaware saying that sometimes the text of the Nudget did not correspond to
where and when the Nudgets showed up Finally some said that using the Nudgets for an extended period
would be a nuisance and would have diminishing returns after a few days of use because they had already
read most of the Nudgets before
Participants reported mixed results in regard to Nudgets changing their opinions and use of social
media sites Many people said that they already knew that social media sites used persuasive techniques but
Nudgets emphasized and enhanced that knowledge which is not surprising given that our participant pool
was highly trained and in several cases proficient in humancomputerinteraction Participants said that
Nudgets made them more aware of the pervasiveness of these design techniques and how intentional every
design choice is to maximize time spent on social media However most participants were unsure of how
this knowledge would translate to action 33 of participants in the treatment group said they would spend
less time on social media the rest said that the Nudgets prompted them to think more intentionally about
social media but they were unsure if they would actually spend less time on it
35 Discussion
We found that Nudget was able to teach users about the manipulations that it was designed to cover
with significant results from both a ttest and regression however Nudget did not facilitate statistically
significant generalizable awareness transfer learning The fact that condition was not a significant predictor
for generalizable knowledge transfer may have been because our system did not teach enough interventions
common to Facebook and LinkedIn effectively First perhaps the duration of our longitudinal study was not
long enough to repeat concepts enough times for users to gain higher level knowledge process
metaanalyze and retain The second reason may be that even though we coded our system to choose from
20 manipulations at random participants reported recalling only between 3 and 9 unique nudges Certain
divs like autoplaying videos or posts with comments open may not have showed up enough times to be
23
picked for annotation A third reason may be because such a system has to run on more than one platform
not just Facebook for participants to be able to generalize knowledge Finally the large size of the gold
standard answer pool size may have made increases in awareness a tiny effect size
Our results suggest that Nudget is most effective as an education tool that is used for a few days It is
clear that the Nudgets did increase awareness and knowledge of persuasive design techniques validating our
goal for creating Nudget Nudget effectively demonstrated persuasive design examples in real time on
Facebook a website that the participants regularly visited However after the first few days of the study
many participants complained that the Nudgets were annoying and repetitive for example one remarked
The Nudgets were super annoying because of where they were on the screen their color and their unavoidability
Additionally while Nudgets were an effective tool to increase awareness of persuasive design techniques this
did not translate into direct action Many participants said that they did not foresee themselves spending less
time on social media as a direct result of Nudgets For instance one noted Its hard to change such behaviors I
just feel more guilty about it now Maybe thats a start Therefore more work can be done to investigate how
increased awareness can inspire action We discuss additional future work to be done in Chapter 6
36 Summary
In summary we built and evaluated a system that makes persuasive design techniques visible on
Facebook We found that Nudget users displayed a significant difference in transfer learning scores
compared to those who used traditional print and video materials Given that Nudget improves transfer
knowledge it would be useful to investigate what other kinds of behavior change it can drive especially
compared to other interventions Creating interventions like Nudget also highlights the need to develop a
theoretical taxonomy of persuasive design techniques which we will discuss in the next chapter
24
4 Taxonomy of Persuasive Design Techniques
Having empirically studied user awareness about persuasive design techniques we turn to a more
theoretical question about these techniques how do they interrelate to form a system of persuasion We use
the datasets of Facebooks and LinkedIns persuasive design techniques gathered in our Nudget user study
along with theoretical insights from the literature to make progress on this question Specifically we map
persuasive design techniques onto a twodimensional behavioral grid to situate them along different
parameters pertaining to human behavior We then identify additional dimensions of interest to flesh out a
fuller design space within which to situate persuasive design techniques
41 Motivation and Goals
James Williams notes that an area ripe for impact is differentiating between different types of
technologydriven persuasion by how well they are aligned to our goals and how much they constrain our
actions 42 He points out the need for a better vocabulary in the attention economy asserting our freedom
of attention means developing its conceptual and linguistic foundations 63 There exists a rich behavioral
and social psychology literature on persuasion discussed in Chapter 2 but we lack a mapping of modern
digital persuasive design techniques onto this theoretical space 64 Thus in this chapter we take a
descriptive theoretical approach with the following goals in contributing to a taxonomy of persuasive design
techniques We want to more systematically categorize the different types of persuasive pulls we feel when
we interact with these massively popular mass consumer products to
1 Show how persuasive design techniques interrelate and function as a system to affect people
2 Provide a more nuanced vocabulary for persuasive design techniques drawing from the relevant
academic disciplines
3 Lay the foundation for pumping intuition about which kinds of persuasion we feel are problematic
or ethical
In this chapter we focus on the first two goals we address the third in Chapter 5
42 Mapping Persuasive Design Techniques onto the Fogg Behavioral Grid
Although interest in taxonomies of influence has existed since Marwell and Schmitts taxonomy of
16 techniques in 1967 no one has mapped the specific techniques used by mass consumer products in the
attention economy onto this persuasive space BJ Foggs 2009 paper The Behavior Grid 35 Ways Behavior
Can Change later updated in 2012 identifies a specific twodimensional theoretical space that categorizes
different digital and nondigital user behaviors based on how often they occur and how novel or frequent
they are 64 Fogg notes that we lack a mapping of both persuasive theories and persuasive design
techniques onto his behavior grid and that this is an important task for future work We focus here on
filling this gap by mapping persuasive design techniques onto the grid We focus on the design techniques
themselves and not persuasive strategies discussed in Section 25 since as he notes a single theory eg
operant conditioning might extend across multiple rows and columns of his grid reducing the usefulness of
the analysis For example such strategies might include allurement emotional appeal or guilt from
25
Levine Whelesss 1990 review of 53 tactics 48 A major ambiguity in dealing with persuasive strategies
would be accounting for all the different ways that a persuasive strategy can be used
Our grid adapts from Foggs 2012 updated 15box grid 65 as well as his earlier 2009 version of a
35box grid 64 Fogg identifies seven rows that capture the schedule on which a behavior occurs a
onetime behavior a onetime behavior that leads to an ongoing obligation such as adopting a puppy rather
than just playing with a puppy once behaviors that last for a period of time behaviors that occur on a
predictable schedule behaviors that occur on an irregular cue rather than a fixed schedule behaviors
performed at will and behaviors that are always performed He identifies five columns that capture the type
of behavior change performed a new behavior an existing behavior and increase in frequency or duration
of the behavior a decrease in frequency or duration of the behavior and a behavior that stops being
performed
We removed the behavior is at will row which Fogg defines as containing behaviors that the user
can perform at any moment primarily because these behaviors depend on the screen that the user is
currently on Due to the abundance of screens and user journeys on both Facebook and LinkedIn clicking on
a single button can take a user to a completely different page the only thing that remains constant is the
horizontal toolbar at the top of the web page that allows options such as Home and Notifications as well
as the search bar which allows users to independently type and navigate to their desired page Many of these
at will user behaviors are captured in the row that includes behaviors on a predictable schedule More
importantly the one and twoclick user journeys allow many different behaviors to occur at will
Furthermore the use of will conflates the line between autonomous action and action that is prompted by a
persuasive technique for which there are no clear metaphysical distinctions due to the ultrapersuasive
worlds into which such platforms place users Foggs nondigital examples of behavior for a period of time
include exercising for 30 minutes and flossing for a longer duration In the context of Facebook and
LinkedIn we took a period of time to refer to behaviors that take longer than a few seconds eg scrolling
for five seconds as opposed to clicking a button for less than a second
What constitutes a persuasive design technique that we might map onto this grid There is no
standard definition for what a persuasive design technique is and opinions vary on how many there are 66
As opposed to persuasive strategies of which there appears to be a finite set based on past literature the
number of design techniques is large perhaps uncountable 66 Every successful interface is designed to
achieve a purpose and guides the user to some degree to achieve a task For this reason from one
perspective almost anything could be considered persuasive as long as the design incorporates an argument
or usage intention 44 Following the standard definition from captology a persuasive design technique
occurs when a designer designs with intent to change an attitude or behavior 19 Since it is difficult to
survey the designers of these platforms for their intent about each specific design element without revealing
potentially sensitive information about the inner workings of their platforms our dataset of persuasive
design techniques consists of 1 techniques we identified on Facebook while designing Nudget and 2
techniques that subjects in our Nudget user study described in Chapter 3 perceived to be persuasive on
LinkedIn These are listed fully in Appendix A and Appendix C Fogg later defined persuasive design
techniques to be a specific implementation of a persuasive experience 64 The design techniques identified
by our study participants are along the lines of what Fogg has in mind as seen in the example he provides of
Amazons technique to offer free shipping for purchases over 25 the timing of Amazons offer and how it
is worded is a design technique 64
26
We take a relatively liberal approach to defining what constitutes a persuasive design technique In
our dataset anything from a monetization strategy technique 10 to the color of a button to the text used
under an icon could be considered the visual conduit of persuasion under a broad notion of what persuasion
is metaphysically We call the visual element through which the persuasion occurs the vehicle In our analysis
each persuasive design technique must have a vehicle as well as a method of persuasion how it persuades The
same visual element can have multiple methods of persuasion which we count as two different persuasive
design techniques eg techniques 6 and 7
421 Case Study Facebook
In Figure 9 we present the behavior grid populated with Facebooks persuasive design techniques
where the dark gray row labels indicate rows present in Foggs 2012 condensed version of the behavior grid
We sort the persuasive design techniques identified while designing Nudget into boxes on the grid We
supplement these with a few additional techniques observed by tracing the vehicles of the Facebook
behaviors that Fogg lists in his grid 64 An example of one of these additional techniques is illustrated in
Figure 10
Facebooks design techniques appear to encourage primarily onetime behaviors and behaviors that
occur on a cue Since Facebook has such a massive culture of heavy daily active use many of the persuasive
design techniques employed by the website upon opening it prompt behaviors for users that automatically
occur both on a predictable schedule in order to form habits with the product as well as behaviors that occur
on an unpredictable cue which relies on the behaviorist theory of variable reinforcement rewards 67
Creating this mapping also teaches us more about the behavior grid and its suitability for persuasive
analysis Many of the design techniques could belong in different boxes in the grid for example showing a
number overlayed over the notifications icon at the top toolbar serves to remind users on a predictable
schedule every time they open the website or app which for 2 billion users is a habitual part of their
monthly routine 8 But during continued use of the website or app the same notification style reappears
on the top toolbar on a variable rewards schedule which serves to increase the frequency and also
presumably the duration of users notificationchecking behavior Clicking on a notification in turn takes
them to other kinds of posts and events on Facebook where they then get transported to the boxes in the
behavior grid relating to liking commenting and other engagement behaviors Thus the entire user
experience can be modelled as a journey across the behavior grid where the initial behaviors determine the
initial conditions of the trajectory and the rest becomes a random walk that is a function of the elements and
design techniques that the user is then exposed to as well as any overriding behaviors the user performs
outside of the grid to direct their own behavior agentially onsite in spite of the pulls of persuasive design
Another example of a persuasive design technique that could lead to behaviors spanning multiple
rows is the design choice of autoenabling comments below each post and displaying all the comments sorted
by an algorithm that places the most active conversations above those that are less active This technique can
have multiple behavioral effects prompting a behavior of scrolling through comments which could be
thought to either always occurs bottom row since every post enables comments or occur for a span of time
greater than a few instantaneous seconds Similarly notifications that prompt you to wish your Facebook
friends a happy birthday trigger a behavior that occurs on cue the prompt is the notification however this
is also a behavior that creates an ongoing obligation due to the human tendency to reciprocate birthday
wishes The mysterious newsfeed feed algorithm seems to rank posts by usefulness which could be seen as
decreasing the behavior of browsing useless posts however it also keeps people scrolling via variable
27
reinforcement by providing mildly interesting posts on an unpredictable schedule which increases time
spent on site
A B C D E
perform newunfamiliar
behavior
perform existingfamiliar
behavior
increase behavior intensity
frequency or duration
decrease behavior intensity
frequency or duration
stop a behavior
1 one time behavior
collection of personal and
network information during
onboarding ad placement and
display uniform with friends
posts display of friends who
have played each gameYou are
now connected on Messenger
chat notification with new
friends
turn on chat calltoaction
prepopulated write a
commenttext boxes
adjacent to profile picture
below posts
displaying posts that friends
are tagged in
2
one time behavior that
leads to ongoing
obligationcost
profilecompleteness cues
selfrepresentation leads to
envy spiraling oneclick
confirm friend menu adjacent
to newsfeed
comments as verbal
approval counts of
likesemoji reactions as
social approval friends
interestin events as social
proof
buried menu to
change
security
settings
3
behavior for a period of
time
twoclick option to play games
with friends
unauthorized autoplay always
enabling comments ability to
display multiples chat boxes
towards bottom of the screen
minimizing video to the right
side of newsfeed to allow you
to keep watching
4
behavior on a
predictable schedule
slot machine feed variable
reinforcement rewards
readtimestamps on
Messenger
glued reminders at the top
menu bar glued search bar
5 behavior is on cue
poking reciprocity
intentional interruption via
notification tagging as social
approval oneclick event
join button
quantified FOMO red color
overlay alerts notifications
prompting birthday wishes
6
behavior is always
performed
mystery algorithm
micropersonalization
autopopulation login
information and profile
picture
endless commentsbreaking
news side feed bottomless
bowl newsfeed
incompleteness of options in
left menu
Figure 9 Persuasive design techniques from Facebook mapped onto the Fogg Behavioral Grid color legend above
The variability in each individual users familiarity with the product might also affect the box in the
behavioral grid into which we map a persuasive design technique For example the row behavior is always
performed differs based on how deeply the user has matched their mental model of the affordance the
28
platform provides with the exclusivity of the means to achieve it For instance teens using Snapchat often
start to measure the strength of their friendships through the apps streak feature they subscribe to the
subculture of the platform believing that the number of consecutive messages exchanged with a contact
parametrizes depth of connection 88 This extreme adoption of the platforms metrics can turn behaviors
that for some less invested users might only be performed on cue eg tagging your friend in a photo into
mandatory behaviors due to the expectations of the social group with which they interact through the
platform eg liking your friends profile picture as a kneejerk reaction to let them know youve seen it
The last two columns of Foggs matrix raise two
interesting questions what does Facebook allow us to do
less of and what behaviors does Facebook stop Fogg notes
that Facebook offered relatively few design techniques to
persuade people to decrease or to stop a behavior as shown
by the lack of items in Columns D and E 64 As Fogg
suggests in box D2 when users become flooded with
notifications that are repeatedly subpar in quality the
persuasive design technique of providing notifications for a
range of Facebook activities and updates may actually have
the effect of training users to ignore certain notifications
Similarly by including over 25 possible buttons to click in
the left column navigation menu pictured in Figure 11 the
platform trains users to not direct their attention to the less
used buttons which frees up more attentional tokens to
focus on the most useful presumably the ones at the top
The menus that social media platforms include provide
different means for connecting with people or engaging in
activities together but by reducing users frame of possibility to the options on the menu they decrease other
29
behaviors that users might take to connect with others such as using other platforms to send messages make
calls or hold facetoface meetings
422 Case Study LinkedIn
We now map the 171 persuasive design techniques identified by Nudget subjects on LinkedIn onto
our adaptation of the Fogg behavioral grid Since they are generated via a close reading of three specific
screenshots integral to a typical user flow on LinkedIn they achieve a level of granularity that our more
general observations of Facebook might not As expected by the social nature of LinkedIn and its
competitiveness in the attention economy LinkedIn uses similar persuasive design techniques to Facebook
However one important difference in LinkedIns design is the subscription business model called LinkedIn
Premium in which users can pay for additional insights into their profile as well as expanded messaging
capabilities among other exclusive features 68 This motivates LinkedIn to use strong persuasive pulls to
convince users to actually put down money and buy their subscription in addition to using their product for
free and paying via their time and attention We studied the persuasive design techniques across three
different screens on LinkedIn the Home screen the My Network tab and the Notifications tab The
data we use here are only the persuasive design techniques that our Nudget study subjects noted and reported
on their transfer assessments numbered and listed in Appendix C so they represent the elements most
salient to users both before and after our system intervention We divide the screens into regions along the
natural grid lines of the web page for clarity in discussion and we map the persuasive design techniques from
each of the three screens onto the Fogg behavioral grid screen 1 Figures 1214 screen 2 Figures 1517
screen 3 Figures 1820
Figure 12 LinkedIn screen 1 appears upon login and is comprised of a profile summary social feed news links and
messaging box
30
A B C D E
Screen 1
perform newunfamiliar
behavior
perform existingfamiliar
behavior
increase behavior
intensity frequency or
duration
decrease behavior
intensity frequency or
duration
stop a behavior
1 one time behavior
5 17 21 22 24 25 41
42 44 58 65 87 89 92
13 14 15 16 18 29 39
40 45 53 69 73
34 52 57 11
2
one time behavior that leads to
ongoing obligationcost
6 7 8 9 59 60 62 63
64
3 19 20 30 31 36 55
76
32 70
3 behavior for a period of time 33 12 35 43 68
4
behavior on a predictable
schedule
66 67 72 77 78 79 83 23 80
5 behavior is on cue 90 91
1 2 4 48 51 74 75 81
82
6 behavior is always performed 28 71 85 88
37 47 49 50 54 56 84
86
10 27 46 61 26 38
Figure 13 Persuasive design techniques from LinkedIn Screen 1 mapped onto the behavioral grid color legend above
Persuasive Vehicle Method of Persuasion
1
Notification badges on the horizontal toolbar
for notifications messages and network
Makes you want to click and see new notifications arouses
curiosity
2
Red color of notification badges on the
horizontal toolbar
Stands outcatches your attentionindicates urgency in order to
redirect your clicks to other peoples or companies pages
3
Number on the notification badges on the
horizontal toolbar
Makes it feel like a todo list and makes you want to get the
numbers to 0 arouses our base desire for having order instead
of chaos
4 Intermittent variable notifications
The delivery schedule of notifications is varied and intermittent
which keeps it changing and thus interesting
refer to Appendix C for full list
89 Promoted ad Encourages you to click on content that will make the site money
90 Use of huge emoji in promoted ad
Distracts you from other useful information to encourage you to
click on the promoted content
91 Blue learn more text below promoted ad The color blue makes it enticing to click on the promotion
92 In 1 week get job offer
Enticing and likely false promises in advertisements encourage
clicking
Figure 14 A sample of the list of persuasive design techniques from LinkedIn screen 1
31
As with Facebook it is interesting to see the rare persuasive design techniques that fall into the last
two columns In region 1A the work feature makes users think it will help them find work to do design
technique 11 but it turns out to be a collection of other LinkedIn features to use Thus it decreases the
behavior of looking for jobs which is what an unknowing user might think they are setting out to do and
instead redirects it to other features such as learning advertising and slideshare In region 1B LinkedIn
does not allow you to custom curate your feed technique 38 Even though you can follow people just as
you can on Twitter and Facebook you do not get to decide whose posts you want to see when so you are
forced to let the platform decide which posts you see first This decreases the behavior of controlling which
people or companies posts you get to see and does not let you actively prioritize what really interests you
Figure 15 LinkedIn screen 2 appears upon clicking My Network on the top horizontal toolbar and is comprised of a
network summary box pending invitations box suggested connections feed and Premium advertisement region
A B C D E
Screen 2
perform newunfamiliar
behavior
perform existingfamiliar
behavior
increase behavior
intensity frequency or
duration
decrease behavior
intensity frequency or
duration
stop a behavior
1 one time behavior
116 117 123 129 131
132
105 109 110 114 111 122 97
2
one time behavior that leads to
ongoing obligationcost
94 95 96 98 100 101
136 137 139 142
93 103 104 112 113
127 128 140
121 124 125
3 behavior for a period of time 99 106
4
behavior on a predictable
schedule
133
5 behavior is on cue 118 119 135 126 130 134 102
6 behavior is always performed 138 107 108 141 115 120
Figure 16 Persuasive design techniques from LinkedIn Screen 2 mapped onto the behavioral grid color legend on P30
32
Persuasive Vehicle Method of Persuasion
93
Showing your number of notifications in the
tab on Google Chrome
Attracts attention towards the LinkedIn tab by making you
wonder about the significance of the number
94
Toast prompting user to download Windows
app
Opens the door to moore notifications even while not on
LinkedIn allowing the site to pull you away even if youre doing
something else and spend more time on LinkedIn
95
Text in toast prompting user to download
Windows app
Use of words like seamless makes it seem like users will have a
better experience if they get the app
96
Placement of toast prompting user to
download Windows app
Layout breaks the grid of LinkedIn since the toast crosses over
into the next region which gets us to notice it
refer to Appendix C for full list
139 Repeating the Try for free call to action
The recurring nature of this call to action repeated left to right
across the screen makes it seem more urgent
140 Echoed stylization of try for free button
The try for free button uses the blue and white stylization of
the accept and connect buttons which have already been
associated with positive actions
141 Default status of messaging tab as closed
Lets you see articles and ads and allow their persuasion to attract
you first so you click on them
142
Highlighting the golden color associated
with the premium membership
Brings to mind the status that the site has associated the color
with
Figure 17 A sample of the list of persuasive design techniques from LinkedIn screen 2
In screen 2 the majority of techniques serve to persuade users to engage in onetime behaviors that
lead to ongoing obligations Since screen 2 is all about growing users network and amplifying the network
effects that boost both the individual use of and the overall number of users of the product there are many
techniques in Row 2 to give users a reason to come back and use the site more often For example in region
2A the use of persuasive language shows the ways LinkedIn appeals to key human needs on Maslows
hierarchy Never lose touch technique 102 and variations of the word connect and contact technique
108 These phrasings guide users to equate LinkedIn with human connection whereas true human
connection is only a small part of what the site persuades users towards By using the design pattern of
highlighting the accept button in blue and the ignore button in gray the latter using a word that alludes
to a negative realworld behavior the site dissuades the behavior of clicking ignore technique 121 By
making the count of notifications seem like a count of the number of action items left it decreases the
behavior of ignoring notifications and not acting upon them technique 125
Since the websites sources of income come from ad views and paid subscriptions the design
techniques that are plentiful in screen 2 are also the ones that have users perform new onetime behaviors
clicking on various creatively positioned and formatted ads that use your own profile photo technique 90
juxtaposing an ad with your profile picture and placing single lines of text permanently at the top of the
feed By deemphasizing and making certain links less symmetric eg Help Center and About LinkedIn
makes clicking on them less appealing and less likely technique 133 The same visual element can persuade
33
in multiple different ways For example placing your profile picture next to the company logo serves initially
as a hook to draw your eye in since we are naturally narcissistically attracted by photos of ourselves
technique 134 Once your attention has been directed to the right side of the screen you process the two
symbols photo and logo on a deeper symbolic level imagining the conceptual blending of the self and the
premium membership Once you get used to seeing who has viewed your profile the inherent curiosity and
selfconsciousness makes this a valuable piece of information that draws you repeatedly back to the platform
technique 136
Figure 18 LinkedIn screen 3 appears upon clicking Notifications on the top horizontal toolbar and is comprised of a
Chrome extension advertisement and a notifications list
34
A B C D E
Screen 3
perform newunfamiliar
behavior
perform existingfamiliar
behavior
increase behavior
intensity frequency or
duration
decrease behavior
intensity frequency or
duration
stop a behavior
1 one time behavior 150 151 153 158 160 169 170
2
one time behavior that leads to
ongoing obligationcost
143 145 147
146 154 155 156 157
159 163 164 165
149
3 behavior for a period of time 152 167
4
behavior on a predictable
schedule
5 behavior is on cue 144 148 168 171
6 behavior is always performed 162 161
Figure 19 Persuasive design techniques from LinkedIn Screen 3 mapped onto the behavioral grid color legend on P 30
Persuasive Vehicle Method of Persuasion
143 Extension download message
Opens to the door to more faster notifications even if not
onsite which creates more opportunities to pull you back to the
platform if youre online doing something else
144
Including the number of notifications 4
written and 2 overlayed on the app icon
Gets you to look at the ad
145
Glimpse of what the Chrome extension looks
like
Helps you imagine what it would look like which makes it easier
to make a decision to get it
146 Home button
Provides an easy exit to more of the feed which is architected
solely by LinkedIn
refer to Appendix C for full list
168 X liked Ys post notification Normalizes liking behavior and encourages you to do the same
169 White color of previous notifications
The softer shade of notifications below the fold makes our eyes
more comfortable looking at the notifications encouraging us to
spend more time in this section at the end
170 White color of previous notifications
Contrast with the blue background color of newer notifications
which makes you want to take action to make these more
uniform with the others
171 Including the timestamp of a notification
Conveys the urgency of reacting by counting the number of
hours elapsed
Figure 20 A sample of the list of persuasive design techniques from LinkedIn screen 3
In screen 3 the majority of techniques lead to behaviors that create an ongoing obligation due to the
nature of messaging a simple prepopulated oneclick reply action by the sender will actually initiate an
official chat conversation with the receiver The receiver is not made aware in the moment that the message
was generated by a LinkedIn button and has no way to distinguish between an organic independent
35
message and one generated by the platform The initial message then opens the door to a chat response
which sparks a longerterm communicative obligation that invests users in the platform Technique 149
uses the phrasing even when youre away to paint the default user state as not being away and to reassure
users that in the nonordinary case when they are away from the platform notifications can bring them back
A B C D E
perform newunfamiliar
behavior
perform existingfamiliar
behavior
increase behavior
intensity frequency or
duration
decrease behavior
intensity frequency or
duration
stop a behavior
1 one time behavior
5 17 21 22 24 25 41
42 44 58 65 87 89 92
116 117 123 129 131
132 150 151
13 14 15 16 18 29 39
40 45 53 69 73 105
109 110 114 153 158
160
34 52 57 111 169 170 11 122 97
2
one time behavior that leads to
ongoing obligationcost
6 7 8 9 59 60 62 63
64 94 95 96 98 100
101 136 137 139 142
143 145 147
3 19 20 30 31 36 55
76 93 103 104 112
113 127 128 140 146
154 155 156 157 159
163 164 165
32 70 121 124 125 149
3 behavior for a period of time 99 33 106 152 12 35 43 68 167
4
behavior on a predictable
schedule
66 67 72 77 78 79 83 23 80 133
5 behavior is on cue 90 91 118 119 135
1 2 4 48 51 74 75 81
82 126 130 134 144
148
168 171 102
6 behavior is always performed 28 71 85 88 138
37 47 49 50 54 56 84
86 107 108 141 162
10 27 46 61 115 120
161
26 38
Figure 21 The full dataset of LinkedIns persuasive design techniques mapped onto the adapted version of the Fogg
Behavioral Grid Most techniques prompt onetime behaviors onetime behaviors that lead to ongoing costs behaviors
on cue Both new and existing behaviors are prevalent whereas changing behavioral habits on the product especially
decreasing or stopping behaviors are deemphasized
These mappings of Facebook and LinkedIns design techniques help elucidate how the seemingly
small individual techniques are interconnected to form a web of persuasive guidance One time behaviors
that are higher up vertically in the grid lay the foundation for more habitual behaviors lower in the grid For
example once users give in to the triggering message of the red notification count clear their notifications
and experience the satisfaction of removing the jarring red overlay and restoring visual symmetry over the
icon it lays the foundation for them to want to do so every time they login Moving from columns A to C
illustrates the typical progression of user behavior for tasks important to the platforms bottom line such as
engagement actions and ad views For example comments likes and emoji reactions are an important
measure of user engagement the creators of these platforms could plausibly aim for new users to like or
comment on their first post continue to do so after they see how easy and satisfying it can be and then
increase their frequency of commenting and liking A persuasive design technique like personalizing and
microtargeting ads causes behaviors that are onetime actions such as ad clicks over time as the platform
learns which topics to present the user with at the optimal times they affect the users mental models of and
stance towards the platform as well Each individual users behavior grid also influences that of other users
due to the interactive nature of social media the more comments under a post the more likely that other
36
users will spend more time with a post or see a comment that then compels them to comment The finite
limit on users time and attention in the attention economy connects the Fogg behavioral grid for LinkedIn
to that of Facebook and every other platform that seeks user attention Increasing the time spent on one
platform column C inherently decreases the time available to spend on another platform column D One
limitation of the grid however is that when placing persuasive design techniques into these buckets due to
the difference in what is novel to some users and not others it creates multiple mappings of the same
techniques which is an issue even Fogg runs into in his original paper where he includes certain behaviors
such as create a group and create a page under multiple boxes A2 and B2
43 Design Space Dimensions
Fogg identifies two dimensions of interest when studying persuasive design the type of behavior
change novelty and increasedecrease and the schedule on which the behavior change occurs ranging from
onetime to permanent We identify the following additional dimensions listed in Figure 22 synthesized
primarily from the persuasion theory psychology and behavioral economics literature These ultimately
provide a richer multidimensional theoretical space that can account more granularly for more of the
psychological concepts of interest in persuasion theory
The first category of dimensions deals with the temporal process of persuasion The Fogg behavioral
model 64 can be used to categorize design techniques based on which part of the workflow of
accomplishing a desired behavior they intervene in motivation ability or prompt Motivators
pleasurepain hopefear and acceptancerejection create the desire to engage in a behavior or have an
attitude ability factors time money physical and mental effort social deviance and routineness determine
the ease of performing the behavior and prompts sparks facilitators and signals offer opportune moments
to perform them In this theory with the right level of motivation followed by sufficient ability followed by
an opportune prompt the persuasive design technique will be successful The second dimension consists of
the 8Es of exploratory learning when using a technology experiment expectation what you predict will
happen evidence gathering evaluation comparing experiment with expectation explanation analysis of
gathered evidence editing revision of theorymodel exploration and exercise 119 Third we list the
cognitive components of the problemsolving process when a user interacts with technology forming an
interpretation of the technology defining goals and subgoals identifying and using operators stringing
operators into methods and selecting between methods using selection rules 119 We can classify
persuasive design techniques based on which cognitive component of this process they target The third
dimension incorporates uses of technology products that are not mindless and are more goaldriven eg
someone logging into Facebook to send a specific message or search for a particular group rather than engage
in the more mindless openended scrolling behavior The second dimension accounts for techniques that
evoke both structured and unstructured behaviors on the platform Since both these types of behaviors are
documented in the HCI literature 11 a strength of our list of dimensions is that they account for both
The fourth category of dimensions deals with the style of delivery of the persuasive techniques Fogg
identifies a functional triad of three different functions computers can play when interacting with people
they can function as tools increasing the users capabilities as mediums providing an experience or as
social actors creating a relationship 19 When a persuasive design technique is delivered by a computer
or technology product acting as a tool it can be recognized by the way it performs tasks like the following
37
reduction simplifying tunneling guiding completion through a sequence of tasks tailoring using
personally customized information as part of the design suggestion timed intervention at an opportune
moment where both motivation and ability are present selfmonitoring making behavior visible through
tracking surveillance making behavior observable by others and conditioning reinforcing target
behaviors When persuasive design techniques are enacted through a computer that functions as a medium
we can recognize them as offering exploration and insight via simulations and using procedural rhetoric 19
Finally when persuasive design techniques are enacted through computers that function as social actors an
approach studied in the field of computermediated communication we can recognize them via their
reliance on physical cues psychological cues persuasive language social dynamics and adoption of social
rules
Label Category
Description of
Dimension
Possible Values
Number
of
Possible
Values
1 Temporal
process
Function of technique in
Fogg behavioral model
motivation ability prompt 64 3
2 Temporal
process
Exploratory learning
process targeted
exploratory learning process targeted experiment
expectation evidence evaluation explanation
editing exploration exercise 119
8
3 Temporal
process
problemsolving substep
targeted
interpretation goals operators methods selection
rules 119
5
4 Delivery
style
interpersonal role of the
computer
tool medium social actor 19 3
5 Delivery
style
personalitymediated
dimensions
authoritativenonauthoritative instruction style
cooperativecompetitive social feedback extrinsic
intrinsic motivation type positivenegative
reinforcement 69
4
6 Users
cognition
Maslows hierarchy of
needs targeted
selfactualization esteem belonging and love safety
physiological 70
5
7 Users
cognition
degree of affective
cognition
receiving phenomena responding to phenomena
valuing organizing values into different priorities
internalizing values 71
5
8 Users
cognition
type of memory
targeted
procedural semantic episodic perceptual transfer
memory 120
5
9 Users
cognition
psychological biases
engaged
stems from not enough meaning need to act fast too
much information memory constraints choice
architecture 34
5
Figure 22 Our list of nine additional dimensions along which to classify persuasive design techniques based on their
occurrence during the temporal process of persuasion their delivery style and the aspects of the users cognition
engaged We list the various dimensions along with the possible values along each dimension
38
The fifth dimension incorporates findings from Halko and Kientz in the personalitybased
moderators of successful persuasive design They identify four parameters that can take on opposite values to
be more successful at persuading people with different personality types persuading using an authoritative
vs nonauthoritative instruction style through an agent persuading via the notion of cooperative social
feedback and teamwork vs encouraging competition persuading using extrinsic motivators like trophies or
badges vs intrinsic motivators like feeling fulfilled and persuading via negative reinforcement removing
aversive stimuli vs positive reinforcement adding positive stimuli 69 The sixth dimension refers to
Maslows hierarchy of needs a widely accepted ordering of the basic psychological social and physical needs
of humans from the highestorder to the lowestorder selfactualization esteem belonging and love safety
and physiological needs 70 We can categorize persuasive design techniques based on which needs they
primarily fulfill Next the seventh dimension in the user cognition category is taken from Bloom et als 1956
taxonomy of the affective domain which has to do with the stages of acquiring processing and internalizing
values Along this dimension persuasive design techniques can fall into the buckets of having to do with
receiving phenomena responding to phenomena valuing organizing values into different priorities or
internalizing values 71 The eighth dimension sorts persuasive design techniques by the type of memory
they target procedural which stores skills and procedures semantic declarative memory that stores facts
episodic which stores events perceptual which stores sensory data or transfer memory the interference
of old and new memories 120
Finally there exist several psychological limits and behavioral economics biases and the ninth
dimension categorizes persuasive design techniques by which biases they involve For example whether an
action is described in terms of risk or gain can involve the illusion of scarcity sunk cost fallacy and relative
memories to intentionally persuade 32 These biases have been grouped under different categories and can
arise when there is not enough meaning eg the halo effect where impressions in one area influence those
in another when fast action is required when there is too much information when memory processing
constraints are triggered 34 We list some of these specific biases in Chapter 2 In this dimension we also
include persuasive design techniques that are a direct result of choice architecture manipulation a design
method that asks what optimal users would truly want and then tries to minimize the costs imposed on
those who might satisfice nonoptimally plan and conduct behaviors in order to nudge them towards
the optimal behavior 72
Other dimensions that would be harder to parameterize accurately might be the utility level of the
persuasive direction for the users end goal is it helping you do something you wanted to do or is it
persuading you to do something you didnt want to do Using additional data sources that would be
accessible to the makers of these sites such as clickthroughrates or deviance in time spent due to controlled
introduction of a new feature we could categorize design techniques based on quantitative measures of their
performance in the product However identifying these nine dimensions makes progress on the second goal
of the taxonomy listed at the beginning of this chapter of providing a more expansive and nuanced
vocabulary with which to describe these techniques Although these dimensions are not perfectly orthogonal
eg the motivation parameter in dimension 1 might include needs from Maslows hierarchy in dimension
6 we could make a simplifying assumption that each persuasive design technique must take on a value along
each of the dimensions
This allows us to ask questions like which persuasive design technique serves to provide motivation
dim 1 as the user is engaged in the exploration part of the learning process dim 2 that is building the
users interpretation of the platform dim 3 while the computer is acting as a social actor dim 4 with an
39
authoritative feedback style providing extrinsic motivation dim 5 Taking the product of the number of
possible values across all dimensions gives rise to a space of 900000 possible persuasive design techniques
Conceptualizing different techniques in this space gives us the opportunity to discover new techniques and
identify the proper context in which to use them based on a deeper understanding of the theoretical
underpinnings from which they arise It also allows us to reverseengineer a design choice we find persuasive
and explain temporally and psychologically how and why it functions Our approach thus far does not
directly address the third goal of the theoretical approach moralizing persuasion and pumping intuition
about how to determine which persuasive techniques are ethical in which contexts How might we
determine what patterns maximize agency and minimize unauthorized persuasion This is a question we
address in the next chapter using a philosophical approach
40
5 Implications Ethics and the Philosophy of Technology
If we make explicit how certain technologies shape our lives we can create the distance we need to be able to
relate to these forces This generates the space to experiment with the use of technology keeping a sharp eye on
the quality of the practices resulting from them and based on the realization that every practice in which a
technology is used shapes our own subjectivity as well PeterPaul Verbeek 40
Our discussion so far has helped create the distance to which philosopher of technology Verbeek
refers such that we can step back and analyze persuasive design techniques and the way they interlock to
form digital systems of persuasion We can now turn to the third goal of the theoretical approach to
persuasive design techniques mentioned in Chapter 4 gaining a stronger intuition about which techniques
are considered ethical in different contexts Philosophical concepts like intentions and values will play a large
role in this discussion and they will perhaps contribute more to changing the way we frame questions rather
than providing direct answers However by raising these questions in light of the past two chapters we can
make philosophical progress In this chapter we begin by taking a descriptive approach and surveying past
work on agency and ethics We then transition from this philosophical review to a prescriptive argument in
which we analyze a digital product through the lens of philosophy of technology By doing so we
demonstrate a philosophical approach to measuring the merits of a technology
First why does the question of what constitutes ethical persuasion arise Many contemporary
accounts of ethical issues in digital technology focus on information security and privacy However there is
an underlying and sometimes overshadowed factor driving the design choices that lead to information crises
the fact that information abundance in the attention economy has created the need to profit by maximizing
the capture of attention using digital products Williams notes that the core challenge of the Internet is that
it optimizes more for our impulses than our intentions which creates important and sometimes unsalutary
threats to attentional freedom 73 Indeed as our relationship with technology has evolved we have slowly
allowed persuasive design to chip away at our agential control and more readily allowed platforms to steer us
perhaps similar to the way we will eventually cede control to selfdriving steering and took our hands off the
wheel When is the user steering and when is the platform To determine the ethicality of persuasion we
must first begin by understanding the nature of human agency and the balance between it and technological
structure in guiding behaviors and changing attitudes
51 Structure and Agency
Philosophers and social scientists have long studied the question of whether human behavior is
determined by social and contextual structure external to individuals or whether humans act freely of their
own agency The literature on this debate between structure and agency defines structure as the
recurrent patterned arrangements which influence or limit the choices and opportunities available and
agency as the capacity of individuals to act independently and to make their own free choices 74 On one
hand theorists emphasize the power that agents have in constructing the world However persuasion
theorist Kelton Rhoads claims that our lack of awareness about persuasive pulls creates an illusion of
freedom Id contend we only enjoy the limited amount of freedom that remains between the powerful
influences that largely determine our everyday lives Because we are not consciously aware of those
41
influences we perceive freedom 66 Products in the attention economy tread the fine line between
designing for users autonomous choices and persuading them to do things they did not autonomously decide
or want to do before coming onto the platform Overall it is undoubtable that they do persuade us but we
can also acknowledge that our level of control over our behavior plays a role in mediating the effects of these
products in our lives
This is why taking a middle ground stance is reasonable Modern theorists view structure and
agency as coexisting and coevolving structure influences human behaviour and humans are capable of
changing the social structures they inhabit 7577 Along this vein Verbeek proposes a hybrid model of
agency in which we must view the human and the persuasive technology as a hybrid system because the
defining questions of ethical persuasion dont exist without one or the other when people develop
connections with technology these connections form the places where freedom must be located freedom
is therefore also a hybrid affair distributed over people and artefacts 40 Freedom then is not the lack of
outside influence but the capacity to develop a relationship with these influences Developing this idea we
posit that freedom of attention is the capacity to have full agential control over actions in a digital setting
such that reflection on our relationship with these technologies yields a satisfactory balance between the
influence of the self and of the platform Verbeek further explains that every persuasive technology is
grounded in normative claims or assumptions and that in analyzing persuasive technology a design may
never be seen purely as instrumental but always as mediatory 40 Whether the normativity is rooted in a
financial bottom line from the attention economy or in a nonmonetary desire for behavior change it is
important to consider the technology as mediating between the human persuader and the human user
52 Ethical Persuasion
To answer which forms of persuasive design are ethical it is often easier to begin by identifying
forms that are not There exists a spectrum of different ways to persuade and in fact a spectrum of modes
of influence discussed in Section 25 including education compliance coercion manipulation seduction
and deception Correspondingly there should exist a spectrum of ethicality rather than welldefined
brightlines Consider the difference between persuading a dieting friend to eat a donut either by mentioning
donuts to them or by placing a box of donuts by their bedside every morning 12 Intuitively the latter
seems more unethical than the former yet they seem to fall along a contextdependent continuum To take a
modern technological example something seems similarly unethical about data analytics firm Cambridge
Analyticas micropersonalized targeted political ads that were ripe with disinformation and were delivered
on Facebook using personal information often gathered without consent 78 The recent indictment of
foreign adversaries that used platforms like Facebook and Twitter exactly as intended in order to
microtarget spread propaganda and incite protests similarly seems wrong 79 Perhaps one reason why
these persuasive operations feel wrong is that people are used as a means to an end and if one considers the
persuaders ends unethical then the operation becomes even more unethical In fact both Facebooks
former president and former head of user growth have admitted tremendous guilt over the products
divisive role in the public sphere and its addictive effects on children 80
In addition to harnessing intuition about the opposite of ethical persuasion we can look to past
work which has commented on the ethicality of specific types of persuasive design techniques For example
consider the question of how we should evaluate the strategic design of choice architectures eg persuasive
42
design techniques like optins and optouts in various contexts rather than forced constraints on choice
altogether According to Thaler and Sunstein such nudges are the most reasonable middle ground
between rigid structure and freeform agency especially since they do not coerce by prohibiting ability to
take the alternate course of action However the line between persuasion and coercion is not clear Miller
would argue that much persuasive discourse is indirectly coercive that is the persuasive effectiveness of
messages often depends heavily on the credibility of threats and promises proffered by the communicator
26 Sunstein and Thaler contend that the sheer complexity of modern life and the astounding pace of
technological and global change undermine arguments for rigid mandates or for dogmatic laissezfaire and
that emerging developments should strengthen at once the principled commitment to freedom of choice
and the case for the gentle nudge 72
According to Wood the persuasion present in advertisements bypasses rationality undermines
agency and manipulates us without blatantly deceiving us which is even more appalling than pure
deception 38 He says that even advertisements with beneficial ends such as giving up smoking do so by
unethical means since they corrupt the root of rational communication and preclude the possibility of any
free human community 38 Fogg takes a more moderate view on the use of emotion in persuasion
intuiting that it is only unethical when its intent is to exploit users or when it preys on peoples naturally
strong reactions to negative emotions or threatening information expressed by others 19 He suggests that
techniques like operant conditioning or extreme surveillance should raise red flags 19 Lockton suggests
considering intended commercial benefit intended social benefit and helpfulness to the user as orthogonal
dimensions of what he calls the DwI Design with Intent space in HCI 44 For example the design choice
of making it difficult to put a TV on standby is intended to be environmentally beneficial but could
inconvenience individual users However he admits that measures along these individual dimensions will
remain controversial which will make it even harder to scope out a space that will be ethically useful 44
The risk inherent in making general intuitive claims about different categories of design techniques
stems from the difficulty in accounting for the intentions of the designer the intentions of the users the
specifics of the means used and the actual ends achieved However we synthesize a list of useful ethical
heuristics and questions that users designers and ethical analysts can ask when developing a normative
judgement about a persuasive technology First we should consider the ultimate outcome of the persuasion
as a factor in determining ethicality Baron has advocated considering the utility or worthiness of the end
arguing that manipulation is less objectionable if the worthiness of any end taken by itself is sufficient to
justify manipulation as a means to it 38 Berdichevsky and Neuenschwander further argue for the
importance of considering the end since if something is unethical for you to do of your own volition it is
equally unethical to do when someone persuades you to do it 16
Second we should consider the means when evaluating normativity of influence and should
specifically consider if a form of influence can undermine rationality and agency by its very existence Harris
poses the following thought experiment what if YouTube knew the perfect video to play for you at a given
time that would be completely entertaining and helpful and it autoplayed that perfect video 81
Something about the act of autoplaying the video and taking over your control of the dimension of time still
makes it feel like structure is dominating agency and skewing the balance of the hybrid model of agency
Even if a technique steers a user back to a course that a reflective and rationally selfgoverning person would
have followed without needing to be steered what matters most according to Wood is the way the
manipulation undermines and demeans the person manipulated by violating and disrespecting his rational
capacities to choose for himself how to live 38
43
Adar et al tie together the consequentialist and deontological considerations by proposing a model
of ethical analysis of design based loosely on criminology which deals with intent as well as means and ends
This threepart model includes motive why it happens means how to do it and opportunity when it
works 32 Measuring motive can be tricky and some have cautioned against it saying that choice is such
a messy thing to dive deep into because then you realize that nobody knows what it means to choose 82
But taking into account intentionality when determining ethicality seems unavoidable If we merely rely on a
probabilistic operational picture of which laws of human behavior make certain outcomes more likely or
which means feel the most problematic we miss out on the underlying motivations and intentions that
ground the entire existence of the persuasion in the first place A useful tool in incorporating judgement of
intentions in analysis of persuasion is Foggs stakeholder costbenefit value analysis which asks the
questions of which stakeholders have the most to lose and gain 19 Mismatches in stakes could indicate
mismatches in motivation which would tilt the ethical equation
Figure 23 This flowchart adapted from Berdichevsky and Neuenschwander 16 illustrates one quick method of gaining
intuition for designer culpability or praiseworthiness based on the intentionality ethicality and predictability of the
outcome of persuasion However instead of serving to provide a definitive framework such a chart reveals several of
the key unanswered questions relating to ethical persuasion how should we gauge both user and designer intention
How do we explore the ethicality of the means and ends of the persuasion without collapsing it into a binary decision
And how do we attribute culpability given that persuasion is a symbolic transaction between two agents
One of the most important and interesting categories of ethical heuristics for persuasive design is
centered around the ideas of disclosure and transparency These ideas appear under many names throughout
the literature but they all boil down to a counterfactual transparency test of sorts whether or not the
persuasive technique still works if the persuasive strategy and vehicle are revealed to the user If revealing the
persuasion technique causes the opposite effect ie if people are so vigilant of not falling prey to the target
behavior that they actively perform the opposite behavior then perhaps there is a case to be made for why it
is not ethical John Rawls called this the publicity principle in 1971 In its simplest form the publicity
principle bans government from selecting a policy that it would not be able or willing to defend publicly to
its own citizens If it adopts policies that it could not defend in public it fails to manifest that respect
Instead it treats citizens as tools for its own manipulation 72 This is supported by Adars HCI definition
of benevolence in deception which says that if the enduser would prefer an experience based on the
deceptive interface over the experience based on the honest one we consider the deception benevolent
32 This notion of truth is important yet elusive since designers themselves could be unaware of the truth
44
or could be targets of a form of deception themselves eg noisy user data or miscalculated user intentions
32 Knowing this it is important to be aware of the slippery slope of increased social acceptance of
persuasion that initially seems unethical 83 Designers should maintain the essence of their initial ethical
intuitions about human agency even once their technology has become ubiquitous
One challenge is that there is no apparent way to systematically crowdsource user choice about each
persuasive design technique in a timely manner without greatly interrupting the user experience
Furthermore it is difficult to predict whether users will consent to or be able to continue being persuaded by
a technique once they have reverseengineered its mechanisms 16 Nevertheless Berdichevsky and
Neuenschwander claim as an ethical principle what they call the disclosure principle that creators of
persuasive technologies ought to disclose their motivations methods and intended outcomes except when
such disclosure would significantly undermine an otherwise ethical goal 16 The Nudget system described
in Chapter 3 provides an extreme example of what such transparency might look like if inbuilt on social
networking sites Efficient disclosure however should aim for direct concise and standardized
communication rather than Nudgets educational goals of generating interest and longerterm knowledge
retention
It is interesting to imagine what user response would be if each major design technique had to
display the reasoning behind its design onscreen or report time spent on site due to that technique These
ideas are beginning to be put into practice For instance viewing total time spent on site is currently
functionality implemented by HabitLab a Chrome extension that provides behavioral interventions to
reduce distraction by products in the attention economy 62 Google announced its forthcoming
implementation at a platform level of time spent statistics in its latest Android P operating system and so
did Apple in the Digital Health initiative incorporated in its newest operating system iOS 12 84121
However adopting mass transparency will still be difficult since meansadaptive persuasive systems are
likely weakened by disclosing how they adapt to individuals whereas endadaptive systems are more likely to
benefit from disclosure 17 Ploug et al explain that when Amazon for example makes personalized book
sale recommendations it is functioning as an endadaptive system that benefits from presenting reasons why
it recommended certain products whereas when it tweaks language to include superlatives like
international bestseller it is meansadaptive since the persuasive technique targets the means of persuasion
17 Drawing back the curtain on the latter kind of technique could reduce the success of the target behavior
85 So this kind of disclosure of means or intent would then affect the achieved end of the behavior which
would feed back into our ethical equation of the motive means and opportunity
Another important ethical heuristic is the golden principle based on Rawlss veil of ignorance In A
Theory of Justice Rawls argues that we would only agree to obey the ethical rules that benefitted us
regardless of who we were in society had we no idea who we were 16 So the golden principle applied to
persuasive design states that the creators of a persuasive technology should never seek to persuade anyone of
something they themselves would not consent to be persuaded of 16 Furthermore the accuracy
principle states that designers ought not to lie especially because humans have a harder time distinguishing
computers lying than humans due to the lack of access to behavioral and bodily cues If human lying is
considered unethical then by extension so is computer lying
Bell and Whaley identify hiding and showing as two kinds of deception 32 which suggests that
we might begin an ethical analysis by asking ourselves what is hidden and what is shown and with what
intention It is also worthwhile to consider the extent to which the persuasion undermines your autonomy
ability to perform the nondesired behavior If the persuasion completely prevents the ability to perform
45
the nondesired behavior then perhaps it is more constraining coercive and unethical than one where the
nondesired behavior can readily be performed Furthermore what is the level of reciprocal equality That is
can the user exert the same forms of influence over the technology or the particular technique in question
1932 If yes then that points to a balance in Verbeeks hybrid agency structure Lieto and Vernero pose the
question of whether the persuasive argument is reducible to a logical fallacy 86 If yes then that could point
to an undermining of rationality Fogg also points out the question if a human were using this strategy to
persuade me would it be ethical 19 This can help focus on the ethicality of the means since presumably
the motivation and outcome would be the same Finally we pose the question of how the persuasive
technology evolves a users original intention Is the change in intention eg from send a quick message on
LinkedIn to catch up on all my notifications considered a corruption of intention or an elevation of
intention How does this change in mental framing of intention align with the users ultimate goals
Answering these questions and checking different persuasive designs against these heuristics will
provide support and reasoning to back our intuition about each persuasive design technique in its own
context These questions and heuristics can also help answer questions of responsibility and blame as
Berdichevsky and Neuenschwander illustrate via the flowchart in Figure 23 Verbeek complicates their
picture by saying that due to the hybrid nature of agency between humans and persuasive technology
ambient intelligence and persuasive technologies therefore also interfere in the moral responsibility of
people for actions arising in interaction with them 40 What does this mean practically in terms of our
institutions of liability as well as the design practices that give rise to such situations Some say that
designers of persuasive technologies should be held responsible only for reasonably predictable outcomes
16 But what should be the threshold of reasonability given the platforms ability to conduct largescale
behavioral experiments that can track the impact of changing every pixel on the screen And what should the
responsibility of these platforms as a whole be in bringing about more ethically palatable forms of persuasion
and discarding problematic ones Before considering the possible designlevel and institutionlevel changes
that could help we consider a few counterarguments
53 CounterArguments
Some argue that the effects of persuasion and the call for more ethical persuasive design in the
attention economy is overblown One counterargument revolves around the use of advertisement blockers
since ads are the linchpin of the attention economy critics of the ethical awakening in persuasive design
argue that using ad blockers significantly reduces possible agencyundermining effects of these platforms
However as Williams explains not viewing ads still exposes you to the design that exploits your
nonrational psychological biases in ways that work against you a product or service does not magically
redesign itself around your goals just because you block it from reaching its own 31
Proponents of the agency side in the agency vs structure debate claim that instead of focusing on
the problem of how to make persuasion more ethical we should focus on empowering people to have more
selfcontrol While selfcontrol is an important part of responsibly using technology that is meant to addict
us we should remember the power of design techniques borrowed from settings like casinos and bottomless
buffets 43 that constantly pull us into the digital world of the attention economy With the massive
micropersonalization efforts and artificial intelligence underlying technology we are subjected to
algorithms that have been trained to observe and adapt to our behavior personality and most intimate
46
preferences For example we should remember that the placement and wording of many advertisements
occurs through sophisticated techniques such as ad format selection via contextual bandits 87 While some
argue that products in the attention economy are just the latest iteration of mass influence technology such as
print advertising radio or television it is their ubiquitousness and daily active use pattern ubiquitousness
that make their influence different and often more insidious Portraying the problem as one in which we just
need to be more mindful of our interaction with apps can be likened to saying we need to be more mindful of
our behavior while interacting with the artificial intelligence algorithms that beat us at chess equally
sophisticated algorithms beat us at the attention game all the time 88 When we remember that hundreds
of engineers and designers predict and plan for our every move on these platforms it seems more justified to
shift the focus of the discussion towards ethical persuasion Williams maintains that asking users to adapt
to distraction become more media literate and choose options that arent on a platforms menu when they
dont align with the users values or goals comprises a pessimistic and unsustainable view of technology
89 Understanding that the problems with persuasive technology are not simply of information
management and user overuse but fundamentally of our attention is an important realization
Finally some say that it is difficult to reject the unethical side effects of persuasive design techniques
without allowing for the benefits they introduce into our lives The impact of these platforms has been far
from completely negative Products like Twitter and Facebook have brought the world together helped
organize resistance against totalitarian governments uplifted the voices of oppressed populations and
facilitated movements countering police brutality 90 But at what cost to the individuals cognitive
resources and freedom of thought and attention As long as there remains a mismatch in the bottom line
goals of the platforms and the deepest goals in our lives the way the technologies influence us will be far
from optimal and could often be unjustified By trying to hold our technology products to a higher standard
of ethical persuasion we open the door to more harmonious designs that continuously empower us instead
of distracting and frustrating us But what kinds of changes are needed in the way we make and use
technology to move towards this vision We now review and discuss improvements that could be made at
different levels in the attention economy
54 Improvements
My view of helpful steps to move persuasive technology in the attention economy away from
problematic modes of influence on users include 1 more contextawareness of users when considering the
behavior patterns for which we design in the attention economy 2 devicelevel changes to better inform
and empower users 3 better metrics for designers to design for what truly matters and 4 possible
institutional changes for widerreaching impact
541 Awareness of Users Ambient and Cognitive Context
As Mark Weiser famously remarked in The Computer for the 21st Century There is more
information available at our fingertips during a walk in the woods than in any computer system yet people
find a walk among trees relaxing and computers frustrating Machines that fit the human environment
instead of forcing humans to enter theirs will make using a computer as refreshing as taking a walk in the
woods Past work has demonstrated the successful design of a contextaware cell phone application SenSay
that uses multiple sensors to determine the optimal time and way to send notifications 91 Tsikerdekis et
al have shown the successful use of nonverbal detection techniques to prevent contextspecific
47
identitybased persuasion to safeguard online communities from deception 92 Hiniker et al have
described how designers can detect when a product is being used intentionally or mindlessly and can change
its interface to adapt to the type of use by foregrounding the primary use case and backgrounding others
unbundling single common patterns of use and designing for disengagement once the primary purpose is
achieved eg when Gmail displays a congratulatory message over an empty inbox 11 When products
become more intentional about realizing the users purpose and style of engagement then they can design for
minimization of overreaching manipulation and elimination of all distraction besides what is necessary to
fulfill the primary intention at hand
542 Devicelevel Changes
Positive change in the movement towards ethical persuasion should also come at the device level
The very design of home screens and default notification structures drives many persuasive techniques that
can be disrespectful of our attention by placing us in a catchup mindset 13 We also need better
standards for labeling in a critical review of captology Atkinson advocates labelling persuasive elements
such as advertisements with consistent labels across apps which can bolster users rights to accept or reject
the content of persuasion 20 Harris suggests that the Apple App Store should reward apps that lead to a
greater proportion of time well spent determined by users through questionnaires instead of solely
rewarding the ones that are trending or have high download counts 12 In other words the menu for what
users can choose from should be sorted by whats good for you like placing healthier options at arms reach
on a shelf Another design change he advocates is the practice of having your phone or computer routinely
giving you detailed breakdowns about your time spent on various sites asking you whether this is how you
want to be spending your time and guiding you through changing your habits if it is not 12 Overall
architects of the platforms that host apps in the attention economy should design more intentionally to
prevent the possible unsavory side effects that their persuasive design causes and they can equip users with
higher levels of information about their use of the platforms healthier defaults and ways to continue
improving their digital habits Changes that come from the makers of smartphones and operating systems
will have a wide reach in selfregulating the attention economy and holding it to more consistent ethical
standards across the board
543 Better Metrics
Metrics are important because they play a key role in the feedback loops that drive design we
improve and iterate our designs based on what we can measure They also play a role in users mental models
of their purpose as they enter and spend time on these platforms often we adopt as our personal metrics
those the site deems important eg Facebook likes Tinder swipes and Snapchat streaks As designers we
parameterize a users needs and values through measured behavior using such metrics but we lose
information about their deeper driving desires Users do not come to these products with the intention of
solely spending time or money or sending a high number of messages or comments yet that is often what is
measured and optimized for This demonstrates the importance of creating more meaningful metrics that
parameterize what users truly care about as well as encapsulating the root of the needs and goals that bring
them to these platforms How might we move towards a more phenomenological understanding of questions
like Which YouTube videos helped you learn an instrument What comedy made you laugh What
encouraging videos helped you face your fears 93 We address a possible approach to these questions in
the next two subsections It might initially seem that answering such questions would not create metrics that
48
link usefully to measuring revenue for the sites however metrics that align better with user values are not
always contrary to the longterm business profits of companies in the attention economy they actually pose a
market opportunity Holmen explains that people increasingly want to spend time well not spend more of it
If its our shallowest self which is reflected to us every time we open Facebook Instagram and YouTube
the best business opportunity around might be to begin to cater for our aspirational selves 94 Metrics that
help companies encourage the formation of healthy digital habits could actually inspire more loyalty and
customer satisfaction Analogous historical evidence for this can be seen in the sales successes of car
companies that offered seat belts almost two decades before they were federally mandated in response to
consumers desire for a safer user experience 122 Due to the nature of the attention economy such a
change made by one product could spur ripples of change across other products Finally better metrics
enable more introspective designers who can more easily determine the ethicality of particular forms of
persuasion with clearer answers about their users deeper values and the experiential and cognitive effects of
design choices
544 Institutional Changes
Legal regulation of persuasion is rarely an optimal route for the fastmoving technology industry
Nonetheless in light of the recent national impacts of technologies like Facebook Twitter and YouTube in
our fundamental democratic institutions eg foreign interference in the 2016 presidential election
789596 people have begun to call for regulation at an institutional level to reduce the social fallout of
these technologies by moderating the techniques they employ Proponents of such change compare
regulation of the attention economy to regulation of the addictive tobacco industry regulation of extraction
economies like coal or the rise of the userdemanddriven organic food movement 1297 In this way
design ethicists call for the attention economy to turn to green solutions that are regenerative and
replenishing since attention has become the modern mined resource Some have even called for the need for
a Silent Spring awakening in the attention economy that leads to a Digital Environmental Protection
Agency 98 transitioning control from the private sectors market regulation to the public sectors policies
Technosociologist Zeynep Tufekci explains why the time is ripe for public discourse about political action
regarding these platforms drawing on an analogy with the automobile industry Facebook is only 13 years
old Twitter 11 and even Google is but 19 At this moment in the evolution of the auto industry there were
still no seat belts airbags emission controls or mandatory crumple zones The rules and incentive structures
underlying how attention and surveillance work on the internet need to change But there are few
solutions to the problems of digital discourse that dont involve huge tradeoffsand these are deeply
political decisions 99
Undoubtedly more innovative business models will play a major role in restructuring the
motivations that compel companies to exploit psychological weaknesses in the name of profit Roger
McNamee an early investor in Facebook and a personal mentor to its creator Mark Zuckerberg recently laid
out the argument for a subscriptionbased adfree model of Facebook claiming that the lack of ads and
sponsored clickbait could improve the user experience enough to charge a subset of users a small fee 100
However switching to a monthly payment or payperuse business model would pose new challenges such
as peoples valuation of the product being highly variable due to the endowment effect where users in one
survey said they would pay an average of about 7 per month to use Facebook but would need 75 per
month to stop using it 101 There have also been creative ideas such as monetizing attention in escrowed
bonds at the outset of persuasion for a more decentralized marketbased solution regulated by consumers
49
themselves based on their judgement of the outcome of the persuasion 102 However as we have discussed
earlier removing ads in lieu of other revenue streams will not solve all the issues that arise with
psychologically manipulative design McNamee has also called for greater user control over the social graph
including the power to export it more easily to other social networks which could give smaller platforms
access to network effects and make them more competitive with the monopolistic companies that currently
dominate the attention economy 100 Regulating the large monopolies that set the standard for design in
the attention economy could level the playing field and reset design standards for all apps
These institutional changes will require time and consideration to become enactable However there
are more immediate steps to be taken in the shorter term One such step is the wider adoption of the process
of valuedriven design an approach based on the principle of foregrounding deeper user values in design
We now briefly discuss this approach along with some of its challenges
55 ValueDriven Design
On every platform a person who wants to be attentive to their friends can find themselves in a state of
frazzled distraction As users we end up acting and socializing in ways we dont believe in and later regret
We act against our values by procrastinating from work by avoiding our feelings by pandering to other
peoples opinions by participating in a hateful mob reacting to the news and so on The coded structure of
push notifications makes it harder to prioritize a value of personal focus the coded structure of likes makes it
harder to prioritize not relying on others opinions and similar structures interfere with other values like
being honest or kind to people being thoughtful Joe Edelman 103
How might we keep persuasion that leads to a target behavior unaligned with the users deeper
values as low as reasonably achievable One approach is valuedriven design which maintains that we must
gauge and then design to support each users true values not just goals and that we must distinguish true
values from ideological commitments broader socialized values norms standards chosen due to social
consequences and goals or fears 103 Even direct values themselves include allegiances situational values
and performative values a range that stems partly from the fact that people have tiers of preferences about
what they want and what they want to want which social psychologists have encapsulated in theories of
multiples selves 104105 In the longer term this valuecentered design might involve rethinking
traditional separations between the user and the designer in the design process and wider adoption of design
methods like ValueSensitiveDesign VSD and Participatory Design PD 106107125
But an added challenge comes from the fact that persuasive technology influences human intentions
and values itself Not only does it condition human behavior but it also helps shape the interpretations on
the basis of which human beings make intentional decisions 40 In an extreme case of such influence our
capacity for moral reflection can be swapped for a voluntary exposure to influence from technology 40 So
then how do we truly gauge values and design for the values that will actually lead to organically defined
wellbeing What mental states or behavior patterns denote true human flourishing and how might we
strive towards those These are important questions for designers of persuasive technologies of the future
who are creating interfaces for users tired of being pulled and cajoled to take microactions that arent aligned
with their deepest intentions Getting people to understand their own values and goals will be a challenge
given the disparity between the experienced and remembered self 108 But there are things that we know
50
we cherish and want in life when we are in touch with our highest deepest most reflective selves and
finding the vocabulary to describe and create those experiences will prove beneficial
From an analytical perspective we can move towards an experiential understanding of what we lose
and gain when we interact with a specific technology by drawing from foundational work on
phenomenology in the philosophy of technology We now analyze a specific technological product using a
phenomenological lens and we demonstrate the emergence of subtle and important insights about the
experiential impacts of technology
56 Case Study Phenomenological Analysis of Google Maps
Google Maps is one of the most prolific navigation technologies in the world and over the thirteen
years since its launch it has come to play an integral role in the daily lives of users worldwide 109 Recent
studies have shown that using Google Maps is linked to the weakening of particular cognitive functions
related to navigation Such findings shed light on the broader personal impact of such a seemingly
innocuous benevolent technology This aspect of Google Maps motivates questions about the user
experience that can be approached in unique ways using foundational concepts and opinions from the
philosophy of technology Although Google Maps is not directly a competitor to the mainstream products in
the attention economy its cognitive underpinnings make it insightful for this kind of analysis
Phenomenology provides a mode of inquiry into the conditions of what makes things appear as such and
lets us probe the coconstitutive nature of humans and technology and the complete experience in all its
sensory and cognitive fullness of interacting with a technology 110 We now deconstruct the technology
through the lens of two philosophers of technology and their phenomenological perspective towards
technologys impact on our minds and lives We draw from Heideggers 1977 essay The Question
Concerning Technology and Albert Borgmanns 1984 essay Focal Things and Practices By analyzing this
technology from the lens of philosophy we demonstrate a method of analysis that can be applied to products
in the attention economy in order to encapsulate some of the more abstract and experiential positives and
negatives with which they are associated
One of Google Mapss core features is providing voice and textbased navigational guidance Using
location data satellite imaging crowdsourced reviews and live traffic data it enables users to get accurate
realtime directions for the best route to a selected
destination There exist many similar GPSbased
routing services like Apple Maps Waze and
Hopstop and several of the critiques we discuss here
could apply to them However we choose to focus
on Google Maps since it is one of the oldest and
most farreaching solutions hosting maps of over 28
million miles since 2005 of locations ranging from
North Korea to the Great Barrier Reef 109111 It
also outranks its competitors in numbers of users by
millions and now has over a billion monthly users
112 We focus our analysis on the mobile use case
an illustration of which is captured in Figure 24
51
Now we discuss the recent work in neuroscience and psychology that sheds light on how
navigational aids like Google Maps adversely impact cognition More specifically when using GPS
technology to navigate we put less work into generating our own internal picture of the world 113
These internal representations are called mental maps and consist of relative positional relations between
worldly landmarks According to one researcher They are very individualThe things which matter to you
might be completely different to those that matter to your wife or your children 113 These mental maps
allow people spontaneity and flexibility while traveling without them you can never deviate from the route
you know look for shortcuts or improvise if the situation calls for it 113 Furthermore additional
findings suggest that people who use a set of directions rather than traditional paper maps draw their route
and surroundings less accurately after the trip and are more susceptible to unknowingly taking redundant or
retraced paths 113 Neuroscientists have even demonstrated that English taxi drivers a seasoned group of
manual navigators have more gray matter in the posterior hippocampus of their brains 123 Although not
an inherent cognitive benefit in itself this represents a lack of risk of psychiatric disorders associated with
lower levels of gray matter
These cognitive findings draw attention to the link between Google Maps and changing subjective
experience As philosophers of technology we can then ask the question of what reality is like for someone
using Google Maps What qualities characterize the experiences of a Google Maps user and what representations of
the world are available to them Due to the phenomenological nature of the questions it seems natural to call
upon Heidegger to provide a formal approach His concept of enframing represents one of the main tenets
of his philosophy of technology that the essence of technology has to do with its way of revealing our
notions about our relationship to technology itself 114 For example Heidegger explains how upon
construction of a dam our conception of a river changes from that of a majestic flow of water to that of a
standingreserve of energy waiting to be tapped 114 By disentangling our own attitudes we can use our
understanding of our enframements of technology as windows into understanding the phenomenological
role of the technology Since society has conceived and developed Google Maps to supplant navigational
cognitive pathways and neural mapping capabilities we have signed away the capacity to experience the
fullness of selfconstructed direction sense This loss is not merely cursory but indicates the loss of a deeper
part of our reality As physicist John Huth puts it Its almost like depriving yourself of music or a
conversation with another person Theres a richness that youre missing out on 113 In this way
understanding what aspects of our brain we conceal via our collective enframing of Google Maps reveals
what we miss out on due to repeated usage of the tech
However it is undisputed that things are also gained phenomenologically Google Maps enables us
to feel the comfort of setting out fearlessly to new destinations with the guarantee of accurate rerouting in
case of misstep But is this enough to extinguish the concern of missing a deeper richness As Google
executive Dan Sieberg puts it the technology allows us to arrive anywhere Edinburgh Cologne Tokyo
and within moments know our way to our hotel have a list of the bestrated restaurants and know the best
route to take on the metro 111 It arguably reduces the friction of having to make manual searches phone
calls and physical map consultations to iron out these details ourselves It is not hard to imagine that being
forced to do this for every potential destination would be an unbearable burden on our time But are there
times when undertaking such actions would not only contribute to our own cognitive good but also return a
more encompassing splendor to our quality of experience We believe Heidegger would say yes Heidegger
might point out that regardless of the alleviation of certain mental efforts using the technology makes it so
that reality becomes more transactional The user interface constantly visually highlights the path to take
52
and a voice regularly blurts out the next mental step a user should make or the next street name or turning
point to selectively seek This transforms a smooth continuous journey into a choppy discrete set of
miniature destinations one for each instruction in the route Users attention is focused so intently on the
directions from point A to B that it becomes hard to meaningfully observe let alone remember the current
surroundings
Another aspect of the human travel experience that is concealed when using Google Maps is the
spontaneity and serendipity of whimsicalness taking the long way just because you feel like it or even
thinking to turn down an unexplored road In addition there is a risk that the foresight afforded by Google
Maps could take away the magic of seeing a place for real for the first time 111 The integration of
thirdparty reviews of points of interest along the way could even make it so that everyone will head for the
same wellreviewed destinations 111 creating a sameness that would detract from the thrill of stumbling
upon exoticlittle known locations that become backdrops for unique memories only you have Overall the
same affordance of Google Maps the ability to always have a perfect virtual travel guide in your pocket
both adds and takes away from aspects of what is required to make travel completely and subjectively
fulfilling phenomenologically
Apart from the question of what is lost and gained when using the app is the question of the nature
of the embodied sense of presence felt by users of Google Maps Here it will be useful to draw upon
Borgmanns characterization of focal practices and their associated sense of wholesome presence which we
will now recount According to Borgmann focal practices are central to living the good life They consist of
a few simple things and are liberating and invigoratingdeeply touching and fleetingwhere effort and
joy are one split between means and ends with a unity of achievement and enjoyment of competence and
consummation 111115 He concretizes this by using the example of an outdoor runner running is simply
to move through time and space stepbystep But there is splendor in that simplicity he says describing
the wholeness of the experience of delighting in both the process of undertaking an activity as well as
completing it
Applied to Google Maps this concept spurs the following question If the purpose of travel is to
undertake a journey qua journey what are the implications of orienting the user in a such a way as to hide
key aspects of how the journey came to be Indeed the technology is designed such that the backend work
required to chart a path observe traffic adjust the path correspondingly and recommend suggested stops or
leisure activities is abstracted to the point where the user is never exposed to the nuanced decisionmaking
behind the scenes This disables the full endtoend understanding that characterizes a focal practice
Heideggers idea that everything you reveal conceals something else is also apt here Why reduce the possible
enjoyment and enrichment a user would experience in understanding the inner workings of the means of
travel by foregrounding the ends
Another relevant aspect of the technology under a focal practice lens is the egocentric reorientation
of the map to put users in the center Studies have shown that following egocentric cues leads to lower recall
of landmarks passed on the route as compared to traditional maps 113 In fact in one recent experiment
researchers studied the varied effectiveness in acquiring landmarkreferenced allocentric knowledge and
viewreferenced egocentric knowledge 116 They found that learning navigation using cartographic
maps led to faster improvements in judgements of relative direction while learning using routes led to more
accurate results in scene and orientationdependent pointing This sheds light on the fact that using a
routebased guide like Google Maps foregrounds aspects of the individuals perceived role in navigation and
gives users a quantifiably stronger sense of how the world points out away from them This contrasts with
53
the allocentric presentation inherent in paper maps which forces you to plan and frame your route within a
meaningful context towns forts universities parks and natural features named for local heroes and history
or local flora and fauna 117 Although Google Maps does display these features when used as just a map it
strips away all extraneous details as soon as users enter a destination 117 In this way using the technology
lessens the potential to truly and focally make the practice of travel ones own
One could argue that Google Mapss recent integration of more threedimensional street views and
intelligent recommendations about midway stops represents attempts to make the technology more
allocentric Nevertheless just like Borgmanns concept of the outdoor runner there is something more
organically satisfying gained from disengaging with the technology to undertake the crux of the practice of
travelling One journalist who decided to navigate around Boston without Google Maps poignantly
recounted such an experience That evening as I drove down the Pike my window down and my phone
buried deep in my pants pocket the city snapped into shape around me Suddenly I was not just a guy who
had learned a set of moves I was a guy who knew his way 113
Viewing Google Maps through the lens of Heidegger and Borgmann allows us to articulate more
precisely what aspects of the human experience are at stake when we use the technology Thinking about
these issues as agents for technological change brings up important design questions to consider as we shape
the technology in coming years For example how might we design Google Maps to support mental
mapping instead of supplant it Might we include quizzes or routebased games to keep cognitive skills
sharp Could we create different modes of usage corresponding to peoples different intentions behind travel
allowing more transactional interactions for urgent targetoriented journeys versus more serendipitous
freeflowing travel experiences for leisure trips This could take a form ranging from a simple question
requiring user input about how leisurely their travel is before giving them a route to a redesign of the
routegeneration algorithms to increase room for seemingly suboptimal suggestions to parameterize
serendipity Furthermore how might we reduce the egocentric worldview the current technology induces
One writer notes that You can redirect most smartphone navigation apps to align with the magnetic
compass instead of your direction of travel 117 Might we make this an easier more quickly accessible
choice on the screen Might we remove the screen altogether and have the guiding voice be more mysterious
and whimsical
Even more ideas arise when responding to the insights gained from viewing Google Maps through
the focal practice lens How might we preserve the focal nature of travelling without ridding ourselves of the
ease of the tech Borgmann takes a step beyond Heidegger to say that we do not have to seek out
pretechnological enclaves to encounter focal things so might it be possible to use the global reach of Google
Maps to not just support the idea of travel as a focal practice but fully transform travel into a focal practice
Considering Borgmanns ultimate view of focal practices as providing a profounder commerce with reality
that deepens charity and compassion could we include prompts or incentives for travelers to engage
constructively with others along their journeys rather than being lone arrows on the screen Could we
achieve this by not providing statistical insights gained from others travels to allow for more
personalization What if we made the maps annotatable to allow users to scrawl personalized nicknames for
landmarks As a frontend engineer or user experience designer working on Google Maps or even a
designer using the Google Maps API for my own app such questions would be very relevant to my work
Framing them using the rhetoric of philosophical analysis provides a unique window into such design
choices
54
In the end looking at Google Maps through the lens of Heidegger allows us to critically examine the
nature of the reality available to users showing both the experiential gains and losses Examining Google
Maps through Borgmanns concept of focal practices helps elucidate how abstracting information and
portraying egocentric symbols contributes to and detracts from wholesomeness of travel as a focal
experience As philosophers and technologists such analysis allows us to propose new directions for even the
most established technologies In a world where innovation often outpaces critical discourse bringing the
voice of philosophers into modern discussions is especially worthwhile as we strive to gain a more complete
understanding of the positives and negatives of technologies on the human experience
55
6 Future Work and Conclusion
In the past three chapters we have taken a system design approach a persuasive theory approach
and an ethics and philosophy of technology approach to studying persuasive design techniques We began by
creating and evaluating a system that makes persuasive design techniques visible on Facebook feeds finding
that our system could improve transfer knowledge of such techniques significantly more than traditional
methods of educating users We then mapped specific persuasive design techniques from Facebook and
LinkedIn onto the behavior grid and theoretically analyzed how they interconnect to function as influential
systems offering a standard behavioral vocabulary with which to classify them Finally we synthesized
several normative heuristics to provide techniques for designers to intuit what constitutes ethical persuasion
providing a sample phenomenological analysis of Google Maps to advocate for experiential and valuedriven
perspectives in the development of future products in the attention economy
There is important future work to be done in all three of these domains On the empirical side we
measured the effects of a system like Nudget on user knowledge but how might such a system affect user
behaviors engagement styles or time spent on the platform We also hope to more granularly measure the
impact that design techniques in different categories of persuasive theory have on user engagement
attitudes and cognitive resources on social media Furthermore our participant pool was highly trained
which leaves scope for future work to answer the question of how a more representative sample would
respond to a system like Nudget and how results would change over the course of a longer longitudinal
study Using the insights from the user feedback on systems like Nudget can help create better design tools
for the architects of mass consumer social media sites by giving them more effective ways to understand
how their persuasive design techniques affect peoples cognition
From a persuasion theory standpoint we can continue to probe deeper into the Fogg Behavioral
Grid by more closely examining techniques in individual grid boxes of interest Using the additional
dimensions we identify in Chapter 4 it would prove incredibly fruitful to use a methodology like
morphological analysis to expand the theoretical space that persuasive design techniques occupy This would
help us discover new methods of persuasion as well as boundary conditions and extreme cases of persuasion
which would further develop our ethical intuition about persuasion 118 We could use such a taxonomy to
discover new persuasive techniques and to better understand how to benefit from the same cognitive
weaknesses and abnormalities that sometimes lead us to act against our goals by turning them into creative
ways to benefit from our flaws and perform behaviors we deeply want
We are at a critical point in our relationship with popular technology products in the attention
economy In the coming years understanding the ways in which they persuade us will be a crucial problem
to get right because the stakes and influence of these technology is higher than ever before These products
shape our thoughts and actions both when we are on and off their platforms Via personalization and filter
bubbles they change the spread of information and subsequent social and cognitive stratification of society
Maintaining a demand for ethical persuasion and freedom of attention demonstrates the need for more
constructive design institutionlevel conversations and shifts towards ethical persuasion We should
ultimately continue to find test and implement ways for designers of these platforms to design for our
deeper values and use technology to nourish rather than distract us to encourage us to fulfill our goals rather
than scroll mindlessly
56
There is a slow but sure social awakening occurring in the technology industry and in the attention
economy concerning the ethical responsibility of tech platforms to examine how they persuade and there
needs to be further work at the design level platform level and institutional level to make sure we maintain
the right level of control over our cognitive resources Understanding the way technology guides and shapes
our behaviors and attitudes is crucial to understanding how we function in the world today By empowering
both everyday users the creators of these technologies persuasion scholars and policymakers with better
ways to understand discuss and mindfully design the persuasive effects of our platforms we can make
progress towards a world in which people feel more empowered with an understanding and control of their
own attention autonomy and digital lives
57
References
1 Simon HA Designing Organizations for an InformationRich World Internet 1 Sep 1969 cited 28
May 2018 Available httpzeuszeitde200739simonpdf
2 Kelly K Better than Free Internet 5 Feb 2008 cited 28 May 2018 Available
httpswwwedgeorg3rd_culturekelly08kelly08_indexhtml
3 Wu F Huberman BA Novelty and collective attention Proc Natl Acad Sci U S A 2007104
1759917601
4 Fogg BJ Mass Interpersonal Persuasion An Early View of a New Phenomenon In Persuasive
Technology Lab Internet cited 29 May 2018 Available
httpcaptologystanfordeduwpcontentuploads201403MIP_Fogg_Stanfordpdf
5 Social Media Fact Sheet In Pew Research Center Internet Science Tech Internet 5 Feb 2018
cited 29 May 2018 Available httpwwwpewinternetorgfactsheetsocialmedia
6 Smith A Anderson M Social Media Use in 2018 In Pew Research Center Internet Science Tech
Internet 1 Mar 2018 cited 29 May 2018 Available
httpwwwpewinternetorg20180301socialmediausein2018
7 Shi E Niu Y Jakobsson M Chow R Implicit Authentication through Learning User Behavior Lecture
Notes in Computer Science 2011 pp 99113
8 Matthieu The 50 numbers to know about social networks in 2018 In Spitche Internet 8 Jan 2018
cited 29 May 2018 Available
httpswwwspitchecomdigitalmarketingen50numbersknowsocialmedia2018
9 Hackett C McClendon D Posts Bio Christians remain worlds largest religious group but they are
declining in Europe In Pew Research Center Internet 5 Apr 2017 cited 29 May 2018 Available
httpwwwpewresearchorgfacttank20170405christiansremainworldslargestreligiousgroupbuttheyaredecliningineurope
10 Feldman B Most of the Time Youre Watching What YouTube Wants You to See In Select All
Internet 11 Jan 2018 cited 29 May 2018 Available
httpnymagcomselectall201801mostofyourtimespentonyoutubeisdictatedbyyoutubeht
ml
11 Lukoff K Yu C Kientz J Hiniker A What Makes Smartphone Use Meaningful or Meaningless
Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies 20182 126
12 Harris S Waking Up Podcast 71 What is Technology Doing to Us Sam Harris In Sam Harris
Internet cited 25 May 2018 Available
httpswakinguplibsyncom71whatistechnologydoingtous
13 Edelman J Is Anything Worth Maximizing On Metrics Society and Selves Internet Available
httpnxhxorgmaximizing
14 Roettgers J For Facebook Change Is Political For Google Its Personal In Variety Internet 8 May
2018 cited 25 May 2018 Available
httpvarietycom2018digitalnewsgooglepersonalwellbeingonlineabuse1202803655
15 Constine J Instagram CEO confirms upcoming time spent Usage Insights In TechCrunch
Internet 15 May 2018 cited 25 May 2018 Available
httpsocialtechcrunchcom20180515instagramusageinsights
58
16 Berdichevsky D Neuenschwander E Toward an ethics of persuasive technology Commun ACM
199942 5158
17 Ploug T Hasle P OinasKukkonen H Persuasive Technology 5th International Conference
PERSUASIVE 2010 Copenhagen Denmark June 710 2010 Proceedings Springer Science Business
Media 2010
18 Kaptein M Eckles D Selecting Effective Means to Any End Futures and Ethics of Persuasion
Profiling Lecture Notes in Computer Science 2010 pp 8293
19 Fogg BJ Persuasive Technology Using Computers to Change what We Think and Do Morgan
Kaufmann 2003
20 IJsselsteijn W de Kort Y Midden C Eggen B van den Hoven E Persuasive Technology First
International Conference on Persuasive Technology for Human WellBeing PERSUASIVE 2006
Eindhoven The Netherlands May 1819 2006 Proceedings Springer Science Business Media 2006
21 Davenport TH Beck JC The Attention Economy Understanding the New Currency of Business
Harvard Business Press 2002
22 Stothart C Mitchum A Yehnert C The attentional cost of receiving a cell phone notification J Exp
Psychol Hum Percept Perform 201541 893897
23 Mark G Gudith D Klocke U The cost of interrupted work Proceeding of the twentysixth annual
CHI conference on Human factors in computing systems CHI 08 2008 doi10114513570541357072
24 Stone L Just Breathe Building the case for Email Apnea In HuffPost Internet HuffPost 8 Feb 2008
cited 28 May 2018 Available
httpswwwhuffingtonpostcomlindastonejustbreathebuildingthe_b_85651html
25 Daniel C McFarlane Kara A Latorella The Scope and Importance of Human Interruption in
HumanComputer Interaction Design HumanComputer Interaction 2009 Available
httpswwwtandfonlinecomdoiabs101207S15327051HCI1701_1
26 Dillard JP Pfau M Chaiken S The Persuasion Handbook Developments in Theory and Practice
SAGE 2002
27 Shrager J Demandance Internet 2015 Available httparxivorgabs150701882
28 Mole C Attention 2009 Available httpsplatostanfordeduentriesattention
29 Hallden L Towards a Brafree Instagram Experience In NewCo Shift Internet NewCo Shift 30
Dec 2017 cited 29 May 2018 Available
httpsshiftnewcocotowardsabrafreeinstagramexperience3e43273b611f
30 Williams J The philosophy and ethics of attention and persuasion Internet 28 Jun 2016 cited 25
May 2018 Available
httpsonlinedialoguenlwpcontentuploads201606JamesWilliamsDPItalkinUtrechtJune20
16compressedpdf
31 Williams J Why Its OK to Block Ads In Practical Ethics Internet 16 Oct 2015 cited 25 May
2018 Available httpblogpracticalethicsoxacuk201510whyitsoktoblockads
32 Adar E Tan DS Teevan J Benevolent deception in human computer interaction Proceedings of the
SIGCHI Conference on Human Factors in Computing Systems CHI 13 2013
doi10114524706542466246
33 Cialdini R PreSuasion A Revolutionary Way to Influence and Persuade Random House 2016
34 Benson B Cognitive bias cheat sheet Better Humans In Better Humans Internet Better Humans
1 Sep 2016 cited 25 May 2018 Available
59
httpsbetterhumanscoachmecognitivebiascheatsheet55a472476b18
35 Petty RE Cacioppo JT The Elaboration Likelihood Model of Persuasion Communication and
Persuasion Springer New York NY 1986 pp 124
36 Eslami M Karahalios K Sandvig C Vaccaro K Rickman A Hamilton K et al First I like it then I
hide it Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems CHI 16
2016 doi10114528580362858494
37 Rhoads K Definitions Persuasion Compliance Propaganda Brain Washing Social Psychology
Internet cited 25 May 2018 Available httpwwwworkingpsychologycomdefinithtml
38 Coons C Weber M Baron M Barnhill A Manipulation Theory and Practice Oxford University
Press 2014
39 Rhoads K Cialdini R Working Psychology Applying Psychology to Influence Persuasion
Teamwork Internet cited 25 May 2018 Available httpwwwworkingpsychologycom
40 Verbeek PP Ambient Intelligence and Persuasive Technology The Blurring Boundaries Between
Human and Technology Nanoethics 20093 231242
41 Coons C Weber M Paternalism Theory and Practice Cambridge University Press 2013
42 Williams J NEXT17 Distraction by Design Why the Attention Economy Is in a Moral Crisis
Internet 2017 Available httpswwwyoutubecomwatchvy32RWsZXCL4
43 Wansink B Painter JE North J Bottomless bowls why visual cues of portion size may influence
intake Obes Res 200513 93100
44 Lockton D Harrison D Stanton N Design with Intent Persuasive Technology in a Wider Context
Lecture Notes in Computer Science pp 274278
45 Rhoads K Marwell Schmitts 16 Internet cited 5 June 2018 Available
httpwwwworkingpsychologycommarwellhtml
46 Marwell G and Schmitt D Dimensions of ComplianceGaining Behavior An Empirical Analysis on
JSTOR In American Sociological Association Internet cited 31 May 2018 Available
httpwwwjstororgstable2786181
47 Dillard JP The Current Status of Research on SequentialRequest Compliance Techniques Pers Soc
Psychol Bull 199117 283288
48 Rhoads K 53 Persuasion Tactics The Levine Wheeless Taxonomy Internet cited 25 May 2018
Available httpwwwworkingpsychologycomlevinehtml
49 Cialdini RB Influence The Psychology of Persuasion Harper Collins 1993
50 King W The persuasive effect of graphics in computermediated communication Computers in
Human Behavior Pergamon 19917 269279
51 Cialdini RB Demaine LJ Sagarin BJ Barrett DW Rhoads K Winter PL Managing social norms for
persuasive impact Social Influence 20061 315
52 Tormala ZL Briol P Petty RE When credibility attacks The reverse impact of source credibility on
persuasion J Exp Soc Psychol 200642 684691
53 Kohavi R Deng A Longbotham R Xu Y Seven rules of thumb for web site experimenters
Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data
mining KDD 14 2014 doi10114526233302623341
54 Evangelista B Check your phone 86 times a day Tech insiders say thats by design In San Francisco
Chronicle Internet San Francisco Chronicle 4 May 2018 cited 24 May 2018 Available
httpswwwsfchroniclecombusinessarticleCheckyourphone86timesadayTechinsiders1288
60
8512php
55 Contempo BETA Internet cited 25 May 2018 Available httpscontempothesocialedgecom
56 Rosen LD Mark Carrier L Cheever NA Facebook and texting made me do it Mediainduced
taskswitching while studying Comput Human Behav 201329 948958
57 Tufekci Z Not This One Am Behav Sci 201357 848870
58 Whittaker S Kalnikaite V Hollis V Guydish A Dont Waste My Time Proceedings of the 2016 CHI
Conference on Human Factors in Computing Systems CHI 16 2016 doi10114528580362858193
59 Moll B Kollberg L Nasri B Lindefors N Kaldo V Living SMART A randomized controlled trial
of a guided online course teaching adults with ADHD or subclinical ADHD to use smartphones to
structure their everyday life Internet Interventions 20152 2431
60 Bator RJ Cialdini RB New Ways to Promote Proenvironmental Behavior The Application of
Persuasion Theory to the Development Of Effective Proenvironmental Public Service
Announcements J Soc Issues 200056 527541
61 Fogg BJ The New Rules of Persuasion Internet cited 25 May 2018 Available
httpcaptologystanfordeduwpcontentuploads201502RSAThenewrulesofpersuasionpdf
62 HabitLab for Chrome Build better habits online Internet cited 25 May 2018 Available
httpshabitlabstanfordedu
63 Williams J Stand out of our Light Freedom and Resistance in the Attention Economy Cambridge
University Press 2018
64 Fogg BJ The Behavior Grid 35 ways behavior can change Proceedings of the 4th International
Conference on Persuasive Technology Persuasive 09 2009 doi10114515419481542001
65 Fogg BJ The Behavior Grid 15 Ways Behavior Can Change Internet cited 24 May 2018
Available httpbehaviorgridorg
66 Rhoads K Ethics of Influence Persuasion Internet cited 24 May 2018 Available
httpwwwworkingpsychologycomethicshtml
67 Skinner BF Operant behavior American Psychological Association 1963 doi101037h0045185
68 LinkedIn Premium Internet cited 2 Jun 2018 Available httpspremiumlinkedincom
69 Halko S Kientz JA Personality and Persuasive Technology An Exploratory Study on
HealthPromoting Mobile Applications Persuasive Technology Springer Berlin Heidelberg 2010 pp
150161
70 Maslows hierarchy of needs Oxford Reference Internet cited 25 May 2018 Available
httpwwwoxfordreferencecomview101093oiauthority20110803100138279
71 Forehand M Blooms Taxonomy Internet 11 Jul 2012 cited 2 Jun 2018 Available
httpswwwd41orgcmslibIL01904672CentricityDomain422BloomsTaxonomypdf
72 Thaler RH Sunstein CR Nudge Improving Decisions About Health Wealth and Happiness
Penguin 2009
73 Williams J Orwell Huxley Banksy In Rough Consensus Internet 24 May 2014 cited 1 Jun 2018
Available httpsblogsoiioxacukroughconsensus201405orwellhuxleybanksy
74 Barker C Cultural Studies Theory and Practice SAGE Publications 2003
75 Bourdieu P Nice R Outline of a Theory of Practice Cambridge Cambridge University Press 1977
76 Bourdieu P The Logic of Practice Stanford University Press 1990
77 Hurrelmann K Social Structure and Personality Development The Individual as a Productive
Processor of Reality CUP Archive 1988
61
78 Rosenberg M Confessore N Cadwalladr C How Trump Consultants Exploited the Facebook Data of
Millions New York Times Internet 17 Mar 2018 cited 25 May 2018 Available
httpswwwnytimescom20180317uspoliticscambridgeanalyticatrumpcampaignhtml
79 Internet Research Agency Indictment In US Department of Justice Internet 16 Feb 2018 cited 1
Jun 2018 Available httpswwwjusticegovfile1035477download
80 Meyer D Facebook Is Ripping Apart Society Former Executive Warns In Fortune Internet cited
2 Jun 2018 Available httpfortunecom20171212chamathpalihapitiyafacebooksociety
81 Harris Sam Waking Up Podcast 71 What is Technology Doing to Us Sam Harris Internet
cited 25 May 2018 Available httpswwwyoutubecomwatchvjlPF9_1VIso
82 Gallagher B Modern Media Is a DoS Attack on Your Free Will 2017 Available
httpnautilusissue52thehivemodernmediaisadosattackonyourfreewill
83 Marcuse H Sozialforschung IF Some Social Implications of Modern Technology Zeitschrift fr
Sozialforschung 19419 414439
84 Fowler G Google wants to cure your smartphone addiction In Washington Post Internet The
Washington Post 8 May 2018 cited 1 Jun 2018 Available
httpswwwwashingtonpostcomnewstheswitchwp20180508googlewantstocureyoursma
rtphoneaddiction
85 Williams P Fitzsimons GJ Block LG When Consumers Do Not Recognize Benign Intention
Questions as Persuasion Attempts J Consum Res Oxford University Press 200431 540550
86 Lieto A Vernero F Influencing the Others Minds An Experimental Evaluation of the Use and
Efficacy of FallaciousReducible Arguments in Web and Mobile Technologies PsychNology Journal
201412 87 105
87 Tang L Rosales R Singh A Agarwal D Automatic ad format selection via contextual bandits
Proceedings of the 22nd ACM international conference on Conference on information knowledge
management CIKM 13 2013 doi10114525055152514700
88 Harris T Making Technology Less Manipulative Entire Talk In eCorner Internet 3 Nov 2017
cited 25 May 2018 Available
httpsecornerstanfordeduvideomakingtechnologylessmanipulativeentiretalk
89 Williams J Extracts from Stand Out of Our Light Freedom and Persuasion in the Attention Economy
The Nine Dots Prize In The Nine Dots Prize Internet 30 May 2017 cited 25 May 2018
Available httpsninedotsprizeorgextractsstandlightfreedompersuasionattentioneconomy
90 De Choudhury M Jhaver S Sugar B Weber I Social Media Participation in an Activist Movement for
Racial Equality Proceedings of the International AAAI Conference on Weblogs and Social Media
International AAAI Conference on Weblogs and Social Media NIH Public Access 20162016 92
91 Siewiorek D Smailagic A Furukawa J Krause A Moraveji N Reiger K et al SenSay a contextaware
mobile phone Seventh IEEE International Symposium on Wearable Computers 2003 Proceedings
doi101109iswc20031241422
92 Tsikerdekis M Zeadally S Multiple Account Identity Deception Detection in Social Media Using
Nonverbal Behavior IEEE Trans Inf Forensics Secur 20149 13111321
93 Edelman Joe Is Anything Worth Maximizing On Metrics Society and Selves Internet Available
httpnxhxorgmaximizing
94 Holmen M A Crisis of Attention Part II Internet cited 2 Jun 2018 Available
httpwwwbibblioorgblogacrisisofattentionpartii
62
95 Internet Research Agency Indictment Case 118cr00032DLF Document 1 Internet 2018 Feb
Available httpswwwjusticegovfile1035477download
96 Lewis P Fiction is outperforming reality how YouTubes algorithm distorts truth In the Guardian
Internet 2 Feb 2018 cited 2 Jun 2018 Available
httpwwwtheguardiancomtechnology2018feb02howyoutubesalgorithmdistortstruth
97 Harris Sam Waking Up Podcast 71 What is Technology Doing to Us Sam Harris In Sam
Harris Internet cited 25 May 2018 Available
httpssamharrisorgpodcastswhatistechnologydoingtous
98 Tristan Harris on Twitter In Twitter Internet cited 25 May 2018 Available
httpstwittercomtristanharrisstatus913130707978936321
99 Tufekci Zeynep Its the DemocracyPoisoning Golden Age of Free Speech In WIRED Internet
WIRED 16 Jan 2018 cited 25 May 2018 Available
httpswwwwiredcomstoryfreespeechissuetechturmoilnewcensorship
100 McNamee R Longman M LeTourneau N Atkins D How to Fix FacebookBefore It Fixes Us In
Washington Monthly Internet 7 Jan 2018 cited 2 Jun 2018 Available
httpswashingtonmonthlycommagazinejanuaryfebruarymarch2018howtofixfacebookbefore
itfixesus
101 Sunstein Cass How Much Is it Worth to Use Facebook A Behavioral Perspective Harvard Law
Review Internet cited 28 May 2018 Available
httpsblogharvardlawrevieworghowmuchisitworthtousefacebookabehavioralperspective
102 Loder T Van Alstyne M Wash R An economic answer to unsolicited communication Proceedings
of the 5th ACM conference on Electronic commerce EC 04 2004 doi101145988772988780
103 Edelman J Can software be good for us Human Systems In Medium Internet Human Systems
17 Jan 2018 cited 25 May 2018 Available
httpsmediumcomwhattobuilddearzuckfd25ecb1aa5a
104 Edelman Joe Human Values Internet Available
httpsmediumcomwhattobuildhumanvaluesaquickprimerb01ef9617925
105 Edelman J Whats Next Human Systems In Medium Internet Human Systems 6 Sep 2017 cited
2 Jun 2018 Available httpsmediumcomwhattobuildwhatsnext4b4d00bd9403
106 Davis J Design methods for ethical persuasive computing Proceedings of the 4th International
Conference on Persuasive Technology Persuasive 09 2009 doi10114515419481541957
107 Staying free in a world of persuasive technologies Oxford Internet Institute Internet 29 Jul 2013
cited 25 May 2018 Available
httpswwwoiioxacukblogstayingfreeinaworldofpersuasivetechnologies
108 Neisser U Five kinds of selfknowledge Philos Psychol 19881 3559
109 Google Today we turn 10 In Google Lat Long Internet cited 25 May 2018 Available
httpsmapsgoogleblogcom201502todayweturn10html
110 Introna L Phenomenological Approaches to Ethics and Information Technology 2005 Available
httpsplatostanfordeduarchivessum2011entriesethicsitphenomenology2
111 Chivers T The story of Google Maps In Telegraphcouk Internet 4 Jun 2013 cited 25 May 2018
Available
httpswwwtelegraphcouktechnologygoogle10090014ThestoryofGoogleMapshtml
112 Perez S Google Maps is turning its over a billion users into editors In TechCrunch Internet 21 Jul
63
2016 cited 25 May 2018 Available
httpsocialtechcrunchcom20160721googlemapsisturningitsoverabillionusersintoeditors
113 NeyfakhReporter L Do our brains pay a price for GPS The Boston Globe In BostonGlobecom
Internet 18 Aug 2013 cited 25 May 2018 Available
httpswwwbostonglobecomideas20130817ourbrainspaypriceforgpsd2Tnvo4hiWjuybid5
UhQVOstoryhtml
114 Heidegger M The Question Concerning Technology and Other Essays Harper Collins 1982
115 Michalos AC Borgmann A Technology and the Character of Contemporary Life A Philosophical
Inquiry Nos 198620 573
116 Zhang H Zherdeva K Ekstrom AD Different routes to a cognitive map dissociable forms of spatial
knowledge derived from route and cartographic map learning Mem Cognit 201442 11061117
117 Dobbs D Are GPS apps messing with our brains In Mother Jones Internet 19 Nov 2016 cited 25
May 2018 Available
httpswwwmotherjonescommedia201611gpsbrainfunctionmemorynavigationmapsapps
118 Ritchey T Swedish Morphological Society Internet cited 25 May 2018 Available
httpwwwswemorphcom
119 Shrager Jeff SymSys 245 Interaction Analysis Glossary Lecture notes Stanford University cited 3
June 2018
120 Shrager Jeff Cognition for readers with short attention spans Lecture notes Stanford University
cited 3 June 2018
121 Perez Sarah Apple to launch its own digital health features in iOS 12 says report In TechCrunch
Internet 1 June 2018 cited 4 June 2018 Available
httpstechcrunchcom20180601appletolaunchitsowndigitalhealthfeaturesinios12saysre
port
122 Eyal Nir Why Do Google and Apple Want You to Use Your Phone Less In Nir Far Internet
cited 4 June 2018 Available httpswwwnirandfarcom201806googleapplelessphoneusehtml
123 Maguire Eleanor A et al Navigationrelated structural change in the hippocampi of taxi drivers
PNAS April 11 2000 97 8 43984403 httpsdoiorg101073pnas070039597
124 Brignull Harry Dark Patterns Internet cited 5 June 2018 Available darkpatternsorg
125 Friedman B Kahn P Borning A Value Sensitive Design Theory and Methods UW CSE Technical
Report 021201 12 February 2001 Internet Available
httpciteseerxistpsueduviewdocdownloaddoi1011118020reprep1typepdf
64
Appendix A Nudget Intervention Text
Nudget Location Nudget Text
Characte
r Count
Near a video in the
feed
Unauthorized Autoplay For Facebook your time is so it autoplays all
videos without consent to grab your attention
123
Near an
advertisement post
in the feed
Clever Ad Placement Facebook shapes ads like your friends posts making it
hard to ignore selling your attention to the highest bidder
139
Below the Home
button
Glued Reminders Facebook affixes these buttons here like a billboard
reminding you to keep clicking for more and more
119
Below the
Notifications
button
Quantified FOMO Facebook shows a number to tell you exactly how many
important things youre missing out on What is actually important
142
Below the
Notifications
button
Footinthedoor Seemingly harmless Facebook has a tiny request check
your notifs But were bad at forecasting time and 5 mins later youll probably
still be here
171
Below a post with
comments
Endless Conversation Facebook always enables comments so endless chatter
arouses your curiosity even if its not worth your time
127
Below a post with
comments
Comments Comments social approval Without them theres no incentive to
post content help Facebook earn even more
124
Below the Event
section of the right
scrollable feed
Social Proof With events Facebook lists your friends to tap into your
tendency to accept social proof as validity These friends might not even go
151
Below the Like
button of a post
Lucrative Likes Likes selfvalidation Facebook designed likes so youd click
to see friends names Its their oneclick ticket to buying more of your time
160
To the left of the
side feed
Silly Side Feed With this scrollable feed Facebook creates an artificial sense of
breaking news depicting a permanent flurry of activity
141
Near the top of the
feed
Bottomless Bowl Allyoucaneat buffets increase food intake Similarly your
Facebook feed never ends so they keep you scrolling
133
Near the top of the
feed
Mystery Algorithm Even thousands of FB employees dont really know how
the feed algorithm shaping your thoughts everyday works
128
Near the top of the
feed
Slot Machine Feed Facebooks algorithm makes useful posts appear
intermittently so we keep refreshing never sure when well receive that
dopamineactivating prize
165
Near the left menu
list
Missing Options The more we rely on Facebook the more we limit our sense
of the menu to what is shown here What about options for human connection
NOT on the menu
166
65
Near the Pokes
feature on the left
menu pictured
above
Reciprocity Facebook created pokes to leverage our vulnerability to social
reciprocitywhereas the initiators poke was evoked by a web of manipulation
in the first place
174
Near the
Messenger icon on
the top menu bar
Intentional Interruption Facebook knows an instantly delivered message is
more likely to elicit a response
108
Above the
messenger chatbox
bottom right of
screen
Receipts Facebook always tells the sender when you saw their message
increasing social pressure to respond
109
Below the rightmost
arrow for the
dropdown on the
top menubar
Dont Leave Compare how easy it is to see your feed vs deactivate your
account in settings Facebook purposefully makes it hard for you to exit
146
Near the top of the
feed
Digital Trails Facebook uses your every click to understand your preferences
in layout and content Megapersonalization keeps you clicking
141
Above a post that
begins was tagged
in a photo
Tagging Social Approval Facebook makes you think a tag is an organic
action whereas the tagger is often prompted by a targeted notif
138
Near a post with a
photo
SelfRepresentation Facebook makes it way too easy to conflate a perfect
selfcurated photo with real life
108
Below the
Notifications
button on the top
menubar
Loud Red Facebook knows that red is the optimal color to get you to click Red
biologically draws you in Its a race to the bottom of the brainstem
150
66
Appendix B Nudget Transfer Assessment
To measure transfer learning in our Nudget study we included the following 3 images of LinkedIn screens
each divided into 3 regions A B C We left 6 blank numbered spaces below each screen and requested
participants to identify and write out all the persuasive design techniques they think LinkedIn used in that
region We did not blur the images so as to be as realistic as possible and instead merely redacted last names
Screen 1
Screen 2
67
S
c
r
e
e
n
3
68
Appendix C LinkedIn Persuasive Design Techniques
Below we list all the persuasive design techniques identified by our participants in the pretransfer
assessments To constitute a unique persuasive design technique both of the following factors must be
different from all other techniques 1 the vehicle ie a UI element or group of elements and 2 a reason
why that element creates persuasion The second factor took many forms for instance participants described
their personal interactions with an element on the page the underlying social strategy at play network
effects visual theory and hypotheses about how users interact with the element in a way that aligns with the
websites bottom line goals to name a few
Screen Persuasive Vehicle Method of Persuasion
1A 1
Notification badges on the horizontal toolbar
for notifications messages and network
Makes you want to click and see new notifications arouses
curiosity
1A 2
Red color of notification badges on the
horizontal toolbar
Stands outcatches your attentionindicates urgency in order to
redirect your clicks to other peoples or companies pages
1A 3
Number on the notification badges on the
horizontal toolbar
Makes it feel like a todo list and makes you want to get the
numbers to 0 arouses our base desire for having order instead of
chaos
1A 4 Intermittent variable notifications
The delivery schedule of notifications is varied and intermittent
which keeps it changing and thus interesting
1A 5 Textual ad at the top Ready for a change
Tries to get you to click on that page by appearing organic and
relevant
1A 6
Try Premium for free at the top right of the
horizontal toolbar
Uses its prime accessible location near the first things you see at
the top to get you to pay more money
1A 7
Try Premium for free at the top right of the
horizontal toolbar
Places this action on equal footing with the other calls to action
making it seem like a normal action that most people would take
1A 8 Gold color of Try Premium for free
Evokes the concept of royalty and makes this feature seem even
more special
1A 9
The name Premium for the monetized
version
Makes the monetized version seem like a special offer
1A 10 Autobilling after your free trial has expired
The free offer will automatically turn into a payment unless you
actively cancel which makes it harder to remember
1A 11
Misnomer of the Work feature on the
horizontal toolbar
Makes you think its about jobs for you but is actually a collection of
LinkedIn products they want you to use
1A 12 Learning feature
Removes any guilt of time spent on that feature due to use of a
positive productive word
1A 13
My Network element on the horizontal
toolbar
Encourages you to click by arousing your curiosity about what
others are up to
1A 14 Messaging element on the horizontal toolbar
Makes it appear like all messages are organic conversation in order
to persuade you to check the sponsored ones as well
1A 15 Ordering of elements on the horizontal toolbar
Placing Notifications next to your photo which naturally attracts
your attention first makes you more likely to click on Notifications
rather than My Network Fitts Law which funnels your
prescribed clicks
69
1A 16 Number of elements on the horizontal toolbar
Breaking notifications into so many categories increases your
number of clicks
1A 17 Simplicity of icons Makes them friendly looking at not intimidating
1A 18
Placing My Network first in reading order
leftright on the horizontal toolbar
Foregrounds the rewarding nature of connecting with someone an
important theme throughout the site Maslows hierarchy of needs
70
1A 19
Including a photo of yourself on the horizontal
toolbar
Generates a feeling of selfworth or selfidentity which frames your
reference of mind and feels like it gives you agency when
interacting with the rest of the site
1A 20
Including a photo of yourself on the horizontal
toolbar
Makes you want to check out your own profile like mounting a
mirror next to your desk
1A 21 Me section
Foregrounds the idea of personal identity and representation
whereas it is primarily a way to get more data about you to drive
personalization
1A 22 Jobs button on the horizontal toolbar
Mismatch between expectations and reality this button draws you
in by making you think the features main purpose is to add value to
your job search whereas the sites priorities are split between you
and the companies that advertise on it
1A 23 Jobs button associated email campaign
The site sends daily emails saying it found jobs for you to pull you
back to the jobs feature
1A 24
Placement of the text ad Ready for a
change
Placing this ad about jobs near the Jobs icon makes it an outlet to
satisfy your use case of searching for job making it more likely to be
clicked
1A 25
Placement of the text ad Ready for a
change
Placing this at the top means you are forced to read it of you look at
things lower on the site even if you might not want to read the ad
1A 26 No X button to remove ads Forces you to read ads reducing your agency
1A 27 Placement of Home button
As the leftmost button provides an easy way to refresh the page to
load new content to the feed which is an unending cycle
1A 28 Overall color scheme
LinkedIn uses colors that are known to be trustworthy to gain
your trust
1A 29 Different color of horizontal toolbar Makes it stand out as the prominent elements
1A 30
Showing your number of notifications in the
tab on Google Chrome
Attracts attention towards the LinkedIn tab by making you wonder
what kind of notifications await you
1A 31
Showing a little red dot on the LinkedIn icon in
the tab on Google Chrome
Attracts attention towards the LinkedIn tab by making you wonder
what kind of notifications await you
1B 32 New posts pill Prompts loading more updates which increases time spent on site
1B 33
Interactionbased expansion reexpanding of
comment box when you click on an article and
return to feed
Suggests that you should write a comment on your thoughts
1B 34 Likes and comments
As a content creator these keep you coming back to the site to see
who has responded to your post
1B 35 Comments
Keeps you reading below a post even when youre done reading the
post itself
1B 36
Interactionbased recommendations after an
article click you see similar articles or after
following someone you see similar people to
follow
Suggests that you should do similar actions
70
1B 37 Personalization of all content in the feed
You are only shown content first that you are likely to find
interesting according to the sites algorithm
1B 38
No choice of what you want to see in the feed
LinkedIn chooses all the content
Allows them to display what they think is best for their success
metrics my clicks
1B 39
Akshay likes this text indicating your friend
likes a post
Increases likelihood of post engagement via social proof If my
friend X liked this maybe Ill like it too
1B 40
Akshay likes this text indicating your friend
likes a post
Gives you a false sense of your friends digital presence on the site
1B 41 Existence of ads as part of the news feed Gets you to click on ads
1B 42
Enticing text at the beginning of the Aha
advertisement
Makes you want to keep reading arouses curiosity
1B 43 Unverified content in news articles
Not factchecking the content in news articles allows for
sensationalized content in feeds which gets more clicks
1B 44 Calling advertisements like Aha promotions Makes it appear like ads are special deals
1B 45
Deemphasis of the word promoted on a
promoted post
Makes ads look like other content in the feed
1B 46 Showing profile pictures along with posts
Makes the posts seem more personal and also indicates that
content is personalized for you
1B 47 Having the second post cut off halfway Encourages us to scroll down to see the rest of it below the fold
1B 48 Including blue links in posts Encourages us to click on people and things mentioned in posts
1B 49 Bottomless bowl feed is an infinite scroll list Keeps you endlessly scrolling to the bottom
1B 50 Adoption of feed structure from Facebook
Borrows mindless scrolling behavior from Facebook by borrowing
the feed element
1B 51
Variable reinforcement temporal delay upon
reload of feed
Builds anticipation for the content thats loading
1B 52
Default sort by top ordering posts on the
feed by popularity not recency
Makes popular articles even more popular snowball effect
1B 53
Box at the top of feed providing ways to write
an article etc
Makes content generation which otherwise requires commitment
and time very accessible and increases the pool of content
available to be shown in feeds
1B 54
Presence of jobs recommended for you in the
news feed
Brings personalized job recommendations outside of the Jobs
page to the feed increased personalization means more relevance
and thus aroused interest
1B 55 Like button
A oneclick indication of content relevance which allows LinkedIn
to improve personalization in the future
1B 56 Left column stays fixed Allows affixed cues to continue to cue you as your scroll
1B 57 Promoting famous people
Uses your friends names as bait to get more followers to already
famous people
1B 58
Tooltip with a summary profile when you
hover over someones name
Gets you to easily click on their profile to learn more
1B 59
3 shared connections in the tooltip that
appears when hovering over someones name
Makes you want to know who those people are by clicking on their
profiles as well
1B 60 Brown in badge for Premium users
Displayed like a status symbol which makes it compelling to want
and respect nudging you towards a Premium membership
1B 61 Displaying your number of connections Makes this a metric that you want to maximize
71
1B 62
Displaying a message encouraging you to
access exclusive tools
Makes these tools such as finding out how many people viewed
your profile a metric of interest that you want to maximize
1B 63 Displaying Try Premium for Free message
Makes it easy for you to maximize insights by clicking and
subscribing to the paid version
1B 64 Placement of Grow your network message
Placing grow your network near your number of connections
makes it seem like growing this number is as easy as the click of a
button
1B 65 Grow your network message
Makes you feel like youre not doing enough and could do more
through LinkedIn
1C 66 New message notification Cues user to click on the message
1C 67 Highlighting unread messages Prompts you to read them first and check them off your list
1C 68
Chatbox element as opposed to chat being a
different tab
Allows chatbox to be ubiquitously present and unable to be fully
rid of This allows message responses without leaving engagement
on a particular page
1C 69 Chatbox element
Makes it appear like all messages are organic conversation in order
to persuade you to check the sponsored ones from recruiters as
well
1C 70 Chatbox element
Drives conversion to the site since you now have a concrete
investment in a communication pipeline through the site
1C 71 Size of chatbox
Doesnt completely cover the articles and ads which lets you be
distracted to click on those even as you write a message
1C 72 Size of chatbox
Makes messages prominently displayed which conveys a sense of
urgency to reply
1C 73 Message preview size
Not showing the full message arouses curiosity about the rest of
the message and encourages users to click on the message
1C 74 Online status of messaging contacts Makes it easier to start a conversation with a contact
1C 75
Automatic popup of chat box when a contact is
online and you visit their profile
Makes it easier to start a conversation with a contact
1C 76
Automatic new message when you make a
connection
Prompts you to start sending messages to each new contact
1C 77
The text What people are talking about now
at the top of the trending articles
Uses social proof to show that you can stay in the loop with others
by focusing your attention on this section
1C 78
The text What people are talking about now
at the top of the trending articles
Making it seem like this content is driven purely by others hides the
fact that LinkedIn actually curates it
1C 79 Trending topics list
Uses social proof what others are interested in to convince you to
spend longer on the site
1C 80
Number of readers listed below each article in
the trending topics
Greater number of people increases your likelihood of clicking
bandwagon effect
1C 81 Blue dot next to each trending topic
Similar to notification badges makes you want to click through all
the headlines to get rid of the blue dot
1C 82 Timestamp on each news item
Entices you to make sure you are caught up with news over a time
period
1C 83 Show more link below the news items
Encourages us to interact with the box by expanding it downward
and keeps us hooked to the new content since we purportedly
chose to see more content using our own agency
1C 84 Tracking of your clicks on the trending topics Leads you to see content that you are more likely to click on
72
1C 85 Default status of messaging tab as closed
Lets you see articles and ads and allow their persuasion to attract
you first so you click on them
1C 86 Right column stays fixed Allows affixed cues to continue to cue you as your scroll
1C 87
Incentivizing text like USERNAME are you
ready for a job
Directly appeals to your need of finding a job Maslows hierarchy
of needs 70
1C 88
Personalized ads matching your photo with a
company logo
Makes you wonder what it would be like to have that company on
your profile which is an artificially constructed representation of
self in the first place
1C 89 Promoted ad Encourages you to click on content that will make the site money
1C 90 Use of huge emoji in promoted ad
Distracts you from other useful information to encourage you to
click on the promoted content
1C 91 Blue learn more text below promoted ad The color blue makes it enticing to click on the promotion
1C 92 In 1 week get job offer
Enticing and likely false promises in advertisements encourage
clicking
2A 93
Showing your number of notifications in the
tab on Google Chrome
Attracts attention towards the LinkedIn tab by making you wonder
about the significance of the number
2A 94
Toast prompting user to download Windows
app
Opens the door to moore notifications even while not on LinkedIn
allowing the site to pull you away even if youre doing something
else and spend more time on LinkedIn
2A 95
Text in toast prompting user to download
Windows app
Use of words like seamless makes it seem like users will have a
better experience if they get the app
2A 96
Placement of toast prompting user to
download Windows app
Layout breaks the grid of LinkedIn since the toast crosses over into
the next region which gets us to notice it
2A 97
Size of X button on toast prompting user to
download Windows app
The relatively smaller size of the X button makes it harder to close
the toast
2A 98
Gray bar info icon on toast prompting user to
download Windows app
Makes it seem like the toast needs your attention and the icon
makes it look especially important
2A 99 Prompt to See all connections
Proposes and normalizes a seemingly ridiculous action of scrolling
through hundreds of people youve already connected with
2A 100 Your contact import is ready text
Making it seem like a difficult helpful task is already done hides the
fact that this is LinkedIns way of getting more information and
furthering its network effects
2A 101 More options below the connect button
Makes you feel another click invested in the process if you click on
it a process whose purpose is to get more data
2A 102 Textual framing never lose touch
Makes user think site is suggesting an action in the users best
interest whereas adding more contacts is beneficial to keeping you
on the site more content for feed more possible people to message
and interact with
2A 103 Number of connections displayed
Foregrounding this key metric reinforces your commitment to the
site at a user showing how much investment youve already made
sunk cost
2A 104 Font size of number of connections displayed
Makes this seem like a measure of how effectively you are using
2A 105 Thumbnail photos of your connections
Makes you want to click on their profiles and want to add even
more people
2A 106
Location of thumbnail photos of connections
adjacent to see all
Tiered effect human faces attract you to the photos which makes
you more likely to click on See all
73
2A 107 Repeated use of the word connect
Makes you want to connect with people need for human
connection Maslows hierarchy of needs
2A 108
Repeated use of the word your your
connections your contacts
Makes the product feel more personal improving your relationship
with it
2A 109 White box on gray background Color contrast focuses attention more efficiently on white box
2A 110
Consistent stylization blue text and same font
for all calls to action See all Get the app
Connect
Makes it easier to spot the calls to action
2B 111 Text ad at the top of the screen
This ad persists throughout screens making it a permanent fixture
that appears more often with a greater chance of being clicked
2B 112
Personalized invitation snippet See more
prompt
Snippet arouses curiosity and See More gives users a way to
satisfy that curiosity
2B 113
Showing a message from people below their
invitation
Gives the invitation a more personal context and incentivizes you
to connect with them by making you feel like you must reply
2B 114
Showing the positions of people who sent you
invitations below their names
Serves as a hook to make you want to learn more about them
2B 115 Showing profile photos
Makes everyone seem more human which draws upon the norms
and pressures of human interaction
2B 116
Showing mutual connections you share with
people who sent you invitations
Serves as a hook to make you want to learn more about them
because of your shared network social proof
2B 117
Showing mutual connections you share with
people who sent you invitations
Helps you develop more of a personal bond with a potential
connection
2B 118
Personalized suggestions of who to connect
with People you may know
Creates a goal that you then embrace makes the norm of behavior
to want to increase the number of connections you have by
exploring the network of people in your network
2B 119
Personalized suggestions of who to connect
with People you may know
Increases the likelihood of you making connections with more
people which accelerates LinkedIns network effects
2B 120
Bottomless bowl list of people you know is an
infinite scroll list
Increases the number of new connections you might makes which is
linked to how much time you spend on the site
2B 121
Accept button blue highlighted vs Ignore
button grayed out
Makes you more likely to notice and click on the accept button
which is the desired action that increases LinkedIns network
effect which leads to more time invested in the site
2B 122 Loaded word ignore
Makes it harder to choose this action which has a negative
connotation
2B 123 Loaded word invitation
Makes it more tempting to open a connection request by calling it
an invitation rather than a reviews
2B 124 Manage all feature
Allows you to accept all invitations instantaneously which
accelerates LinkedIns network effects
2B 125
Showing the number of invitations in
parenthesis
This large number in parentheses looks unappealing and makes us
want to decrease the number
2B 126 Red color of notification badges
Stands outcatches your attentionindicates urgency in order to
redirect your clicks to other peoples or companies pages
2B 127 Show more button
By not showing all the relevant details encourages you to
clickscroll further to truly find out more information given that
youve come this far You can keep going with it forever
2B 128 Blue color of Show more button
Makes it more enticing to click on it blue stands out compared to
the rest of the color palette
74
2B 129
In 1 week get job offers from top companies
coming straight to you text in the ad
Enticing vivid language in advertisements shown on LinkedIn
prompts clicks
2C 130 Red new message badge
Makes you want to click to reopen the chatbox even when its
closed
2C 131 Custom personalized ad in the right column
Directly forces you to envision yourself in a context that LinkedIn
wants
2C 132 Size of ad in the right column
Makes ad most likely item to clicked whereas useful things like
Help Center are tiny links Fitts Law
2C 133 Layout of the About Help Center links
This layout is formless and unattractive which serves to add more
emphasis to what we should be looking at the big square ad
2C 134 Showing your photo next to company logo Your own face attracts your eye hook
2C 135 Showing your photo next to company logo
Makes you think of what it would be like to work there and have
that company on your profile another place on the site you look at
your photo
2C 136 See whos viewed your profile
Makes you curious in a narcissistic way a manufactured desire
that can be mitigated by the solution presented purchasing
LinkedIn Premium Ironically this rarely improves your professional
life
2C 137
Use of your name NAME start your free
trial
Makes the monetization request more personal
2C 138 Promoting LinkedIn ads LinkedIn ads ensure that you will stay on the site
2C 139 Repeating the Try for free call to action
The recurring nature of this call to action repeated left to right
across the screen makes it seem more urgent
2C 140 Echoed stylization of try for free button
The try for free button uses the blue and white stylization of the
accept and connect buttons which have already been
associated with positive actions
2C 141 Default status of messaging tab as closed
Lets you see articles and ads and allow their persuasion to attract
you first so you click on them
2C 142
Highlighting the golden color associated with
the premium membership
Brings to mind the status that the site has associated the color
with
3A 143 Extension download message
Opens to the door to more faster notifications even if not onsite
which creates more opportunities to pull you back to the platform if
youre online doing something else
3A 144
Including the number of notifications 4
written and 2 overlayed on the app icon
Gets you to look at the ad
3A 145
Glimpse of what the Chrome extension looks
like
Helps you imagine what it would look like which makes it easier to
make a decision to get it
3A 146 Home button
Provides an easy exit to more of the feed which is architected
solely by LinkedIn
3A 147 Blue fill of Get the extension button
The use of blue is enticing and filling the buttons with blue makes
them the most visually attractive
3A 148 Depiction of notifications
Makes it seem like they are comprehensive reminders whereas
LinkedIn selectively chooses notifications
3A 149 Even when youre away text
Makes you feel like you should be thinking about LinkedIn even
when youre not using it
3A 150
Horizontal text ad crosses over into this
vertical region
Guides your eye across the screen from left to right
75
3B 151 Text ad at the top of the screen
This ad persists throughout screens making it a permanent fixture
that appears more often with a greater chance of being clicked
3B 152 Selection of notifications shown
Showing notifications related to connections you already care
about increases the likelihood of you checking notifications
3B 153
Suggested reply buttons Say congrats Say
happy birthday
Lowers the barrier time you need to spend to reply
3B 154
Suggested reply buttons Say congrats Say
happy birthday
Makes it seem like a lower effort interaction whereas these buttons
actually send a message to the other person
3B 155
Suggested reply buttons Say congrats Say
happy birthday
Makes you feel bad if you dont follow the Wish person X
instructions
3B 156 Bolded suggested reply buttons Entices you to click on the more prominent option
3B 157
Blue color of suggested reply buttons Say
congrats Say happy birthday
Conditions us to associate the color blue with good
3B 158
Existence of birthday and work milestone
reminders
Makes it more likely for you to outsource your memory of
birthdays or work milestones to this feature
3B 159 Receiving birthday and milestone wishes
Makes these seem like organic messages to the receiver whereas
they were prompted by targeted notifications
3B 160 Text of suggested replies
Makes it seem like youre building or strengthening your
connections
3B 161 Recording the replies you choose to send
Allows LinkedIn to personalize your feed to rank these connections
posts higher both in the feed and in these notifications
3B 162 Like button
Allowing you to like a post from within your notifications makes
them more interactive and gives you a calltoaction for the
notifications about posts
3B 163 Cue to sign up for a premium membership Gives you a way to satisfy your desire for more features
3B 164
Placement of cue to sign up for a premium
membership at the top of the notifications
Draws your attention to your need to know what others think of
you first also what makes the site more money first
3B 165
3 people viewed your profile before the
Premium call to action
Hooks you by feeding into your curiosity about how others
perceive you and taunts you with your lack of further information
3B 166
Text giving a taste of a premium membership
free trial stay anonymous
Whets your appetite for the greater number of features to engage
with
3B 167 Premium membership itself
Makes you feel like you should make the most of your membership
and use the platform more
3B 168 X liked Ys post notification Normalizes liking behavior and encourages you to do the same
3B 169 White color of previous notifications
The softer shade of notifications below the fold makes our eyes
more comfortable looking at the notifications encouraging us to
spend more time in this section at the end
3B 170 White color of previous notifications
Contrast with the blue background color of newer notifications
which makes you want to take action to make these more uniform
with the others
3B 171 Including the timestamp of a notification
Conveys the urgency of reacting by counting the number of hours
elapsed