---
title: "Against Nudge"
author: "Jayson Margalus and Nathan Matteson"
url: "https://writebook.jaymargalus.com/2/against-nudge"
---

This paper examines the fundamental incompatibility between nudge theory, as developed by Richard Thaler and Cass Sunstein, and design, rooted in aesthetics and metaphor. While nudge theory seeks to influence human behavior through choice architecture, simplifying decision-making using behavioral economics and cognitive science, design operates within the realms of aesthetics, abstraction, and emergent interaction. The paper argues that the reductionist, outcome-driven methods of nudge theory conflict with the pluralistic, reflective processes central to design. Drawing on the philosophical works of José Ortega y Gasset and Graham Harman, this paper critiques the over-simplification of human behavior in nudge theory and its misapplication to design practices. Through case studies and philosophical analysis, the paper reveals that integrating nudge into design diminishes creativity and undermines the reflective, emergent qualities that define true design work. This critique calls for a reconsideration of how behavioral interventions are employed within the field of design.

Design is its own field of understanding that stands alongside others like religion, philosophy, art, and science. While the field of design might be informed by others, it is distinct. In practice, design focuses on the organization or arrangement of objects -- which might include phones, atoms, the earth, mineral deposits within the earth, people, emotions, etc. -- as well as the interactions among those objects, and finally the aesthetics that emerge from those interactions. In its pursuit of these aesthetics, design seeks to create a representation of some thing-itself. In the creation of these imperfect re-presentations, design, in effect, creates metaphors that emerge from interaction with and among objects.

Aesthetics and metaphor are inextricably tied together. Spanish philosopher José Ortega y Gasset asserts that “**the esthetic object and the metaphorical object are the same** [emphasis added], or rather that metaphor is the elementary esthetic object ... [and] satisfies us precisely because in it we find a coincidence between two things that is more profound and decisive than any mere resemblance.” A well-designed object is beautiful because, in it, we find a coincidence between _the thing_ and some _concealed truth_. 

This conception of design is in contrast with concepts laid out in the book Nudge (2009), where authors Richard Thaler and Cass Sunstein open the text by describing a cafeteria menu full of unhealthy choices: would it be better if healthier choices in the menu were displayed more prominently, thereby encouraging its readers to make better choices? Thaler and Sunstein argue yes. In this context, objects are reduced down to their core components. A menu is a piece of printed paper with food items, prices, a restaurant name, and descriptions of food on it. A person is the sum total of their choices: food preferences, savings accounts, marital status, eye color, and a dark sense of humor.

Yet metaphor, and hence aesthetics, explain something about _the thing itself_ that reducing a menu down to a list of options, or a person down to their characteristics, does not.

Choice architecture (the mechanism through which nudges occur) is a method developed by Thaler and Sunstein, economics faculty at University of Chicago and law faculty at Harvard University, respectively. The concept relies on the cognitive biases and heuristic frameworks that lead to “irrational decisions.” One example of this is demonstrated with the marshmallow experiment conducted out of Stanford in 1972. In this test, children are offered one marshmallow, and told they can have it immediately or wait 15 minutes to receive a second. Most children opted for the former, though the latter is clearly the better option. In this irrational decision-making framework, a choice architect might step in to, say, present the two options in a way to make the latter seem more appealing to the child (perhaps by showing them a picture of a child crying when they don’t receive the second marshmallow). This intervention is called a nudge.

Choice architecture proposes novel nudge interventions to counter those decision-making processes. Unlike other forms of intervention, however, the authors propose to provoke these mediations through a method they call _libertarian paternalism_. That is, interventions that have an individual’s best interests at heart, but do so in a way that respects the individual.

Choice architecture has also been adopted across many disciplines -- including design -- where the concept has received much attention in fields ranging from the development of mobile devices to Human-Computer Interaction (HCI). The theory has seen international adoption by institutions like the World Bank and nation-states, as seen in Great Britain’s Nudge Division. With such widespread adoption it seems almost inevitable that our entire designed world will soon be full of nudge-like interventions. 

This paper makes the case that choice architecture and design have inherently incongruent aims and, therefore, can never wholly work together. We will show that choice architecture is a _scientific theory_ first and that design is, orthogonally, a practice in _aesthetics_. We will then discuss the various incompatibilities between science and aesthetics, and close with further areas of research on this topic.

### Three guiding principles of ethical choice architecture

Like many theories, choice architecture has its own guiding principles. According to Thaler and Sunstein, choice architecture “helps people make the choices that they would have made if they had paid full attention and possessed complete information, unlimited cognitive ability, and complete self-control.” Nudges, which are the product of choice architecture, are “any aspect of the choice architecture that alters people’s behavior in a predictable way without forbidding any options or significantly changing their economic incentives.”

The three guiding principles of a good nudge are:
- All nudging should be transparent and never misleading.
- It should be as easy as possible to opt out of the nudge, preferably with as little as one mouse click.
- There should be good reason to believe that the behavior being encouraged will improve the welfare of those being nudged.

The subjective nature of these principles _(what is transparency, after all?)_ means that nudge exists in a space that requires ethical judgement. In fact, the authors admit this when they note that, when signing their own books, they implore others to “nudge for good!” Nudge, it seems, can be used for good or ill.

Choice architects -- typically a domain expert using the tools of choice architecture -- use these principles as well as a set of heuristics to influence human behavior. These heuristics are based on early research by Daniel Kahneman and Amos Tversky on cognitive biases. Kahneman and Tversky claim that individuals make choices based on three kinds of biases: anchoring, availability, and representativeness -- terms we’ll revisit later.

Examined broadly, these three heuristics assert that the way we present information to an individual influences how they think. Since information can be presented in a manner that encourages people to make poor decisions, we should use our understanding of the mind to present it in a way that helps people make the best decisions for themselves.

### The logical conclusion of choice architecture

Another key assertion by Thaler and Sunstein is that “there is no such thing as a ‘neutral’ design.” Combined with their guiding principles, as well as their employ of heuristics to leverage cognitive biases, this leads to a line of logic that looks something like:

1. All objects -- people, boats, boxes of cereal, and grasshoppers alike -- influence other objects
2. Using data we can predict, with a reasonable degree of certainty, the outcomes and effects of those influences
3. Therefore, we can (and should) design systems that influence for the most positive effect

Continuing along this path, as choice architects are experts in their domain, they know what the best choices are. Because a choice architect can measure how people behave, and because they know what the best choices are, they’re able to connect an individual to a choice best suited for them.

Finally, because there are experts who understand the best way to navigate those influences, we should leave it up to those experts to _design_ these systems for us.

### Choice Architecture and Science

Choice architecture expresses itself as a science first. Its roots are in cognitive and behavioral science, describing how people understand and act based on stimuli. Put another way, there are people, and there are the things that people interact with outside of themselves. This can be described as a subject-object ontology (Descartes’ _I think, therefore I am_ is an example of such an ontology), wherein the outcome of an interaction between the two is the focal point.

As explained previously, much of Nudge’s inspiration comes from the work of Kahneman and Tversky in heuristics, which explain the generalizations we make in order to save energy when making decisions. The study of heuristics, in turn, is influenced by Kahneman’s own work in cognitive systems. Kahneman refers to these systems simply as System 1 and System 2, but for sake of ease, the authors of Nudge name them the _automatic and reflective systems_.

These systems, Kahneman says, explain the two kinds of thinking that humans engage in. Where the automatic system (also referred to in the book as the _lizard inside_) is more reflexive and uncontrolled, the reflective system is controlled and effortful. The automatic and reflective systems govern how we think, behave, and engage with the world around us. They are _the world inside_ that governs how we experience _the world without_.

But reducing individual behavior down to a set of heuristics and cognitive systems -- like biases, genomics, or cognitive science -- does not explain all of the human qualities that emerge from the individual. Similarly, an IKEA dresser, when assembled incorrectly, is not a dresser even though it contains all of the components that a well-constructed dresser does. People are more than their biases and other reducible features -- there is something here that eludes science.

Graham Harman puts it another way when he says that while science “does allow us to predict [some things, that] is not even the point, since even if we could predict the features of all larger entities from their ultimate physical constituents, the ability to predict would not change the fact that the larger entity actually possesses emergent qualities not found in its components.” Here we begin to understand the issue with choice architecture: people contain multitudes, and multitudes elude scientific explanation.

### Design and Aesthetics

The study of aesthetics and metaphor is deeply intertwined. Metaphor alludes to the truth without being truthful -- an abstract painting is the representation of the truth of something, but is _not the thing itself_, as Harman puts it. Aesthetics give us insight into the real without direct access to it, as if reality hid behind a veil where we could see shadows and shapes, but not the actual details. In design practice, this means that aesthetics emerge from our interactions with other objects, giving us insight into what those objects represent. These aesthetics emerge from _feedback_.

As opposed to a subject-object ontology (as described earlier), design’s approach is pluralistic. A pluralistic approach to design looks at the way objects interact with each other at all scales. When an individual’s behavior is nudged, a pluralist might focus less on the outcomes that emerge from their interactions. Instead, they might also ask: how does this affect the individual internally, and also, how does this affect the other systems that individual interacts with (See Figure 1)?

 ![nudge1.png](https://writebook.jaymargalus.com/u/nudge1-TZczJw.png)
_(Figure 1)_

Aesthetics emerge from these complex, emergent interactions that defy our predictive frameworks. Design expresses this in many different ways throughout its varied fields. Game design uses the Mechanics, Dynamics, and Aesthetics (MDA) framework to describe complex interactions and the aesthetics that emerge from them.

Figure 2 is a representation of the MDA framework, which can be summarized as:

1. The player interacts with the game using a mechanic (run, jump, swim)
2. The system responds to the player, creating a dynamic
3. The player then feels something based on this feedback. This is where the aesthetic emerges.

 ![nudge2.png](https://writebook.jaymargalus.com/u/nudge2-08W0xc.png)
_(Figure 2)_

Human-Computer Interaction teaches a similar concept represented by the Gulfs of Evaluation and Execution. In this approach (pictured in Figure 3) the user interacts with the world, and both the execution and evaluation points are mediated by a gulf that separates them.

 ![nudge3.png](https://writebook.jaymargalus.com/u/nudge3-xzkVeO.png)
_(Figure 3)_

In his seminal design book _The Design of Everyday Things_, Don Norman describes yet another similar approach, and further elaborates on the concept by introducing _mental models_. In many ways, mental models emerge from the same dynamics that aesthetics emerge from according to the MDA framework.

 ![nudge4.png](https://writebook.jaymargalus.com/u/nudge4-xZuMM6.png)
_(Figure 4)_

Aesthetics cannot be completely explained by science because they are withdrawn beyond the gulf. One can only describe outward manifestations of aesthetics (a smile, grimace, appearances of excitement in an individual), because aesthetics emerge from inside the objects (our thoughts and feelings in reaction to stimuli). The inward nature of aesthetics means they are _not transparent outwardly_, but are _intentionally_ reflective inwardly. That is, when you interact with a smartphone, the description of that interaction cannot simply be reduced down to a series of button presses, screen swipes, and success criterion -- rather, the interaction is also your feelings of delight, surprise, and wonder that cannot be discerned except from within.

### Design and Science

Science has long informed design, as have art, philosophy, and religion. Importantly, while design incorporates concepts from science and other fields, it is its own distinct practice from these other fields. Where art abstracts through metaphor, design specifies through reduction, and where science reduces, design abstracts through metaphor. Design exists in the liminal space between abstraction and reduction.

We see science’s influence upon design in, for instance, Human–Computer Interaction (HCI), which is predicated upon concepts such as cognition, perception, and understanding. We see it in graphic design when we discuss how people perceive and understand color and the physiology of the reading process which determines the conventions of typesetting. And we see it in user experience and user interface design when we discuss how people read a website or understand information presented to them in an app. How we display information in design is often based on our understanding of how we know the human brain works, and how the brain will perceive that information.

Nudge, then, attempts to find a home in the design practice, as its roots in behavioral science share the same underlying principles of this reductionist, scientific approach. That is, understanding human behavior as a science informs design methods of many practitioners, and science as a practice attempts to explain the world by reducing it to its constituent components.

Yet design’s foundation is grounded in aesthetics. As opposed to science, which is reductive, the study of aesthetics abstracts upward. The study of aesthetics (in both art and philosophy) tells us that understanding cannot simply be reduced to the constituents of things, but is also constructed through metaphor. Aesthetics is a subjective field of exploration that, contemporarily, has received less attention than scientific (explanatory) methods of design -- ones focused on outcomes over process.

### Design and Choice Architecture Frameworks

Nudge appears in design in many ways. In academic literature, there are many proposed methods for implementing it.

In _Nudge Your Customers Toward Better Choices_, a piece published in the Harvard Business Review (HBR), Goldstein et al. argue that there are two kinds of nudges in product design: mass defaults and personalized defaults. Put differently, there are nudges in product design that can apply to all of your customers, and defaults that can be designed into products that cater to groups or individuals through data collection and recommendations. Nudges are not necessarily behavioral changes aimed at broad groups, then, but can be applied to individuals. This would seem to require more than common heuristics that the authors of Nudge have anticipated, since individual preferences (whether it’s your favorite food, color, or dislike for the TV show Hogan’s Heroes) would also need to be considered. A choice architect in this instance would not be an expert on group preferences, but somehow, needs to be an expert on your preferences.

Another approach to choice architecture in design is outlined in _23 Ways to Nudge: A Review of Technology-Mediated Nudging in Human-Computer Interaction_. In it, Caraban et al. create an elaborate framework that outlines 23 distinct mechanisms of nudging, grouped within 6 categories, and leveraging 15 different cognitive biases. The six categories range from confrontation (“nudges that pause an unwanted action by installing doubt”) to deception (or, nudges that “use deception ,.. in order to affect how alternatives are perceived ... or experienced, with the goal of promoting particular outcomes”). Significantly, the piece takes a very prescriptive (see: scientific) approach to including choice architecture in design. As we have noted earlier, it’s this prescriptive nature that makes it difficult to fully reconcile choice architecture with design.

The authors elaborate on this work with a later piece introducing a “Nudge Deck” -- a design tool meant to help designers who are implementing nudges in their own work -- a kind of nudge of its own. The deck of cards -- in much the same way as science prescribes a course or set of actions that must be followed -- instructs designers to use the deck in order to incorporate nudge responsibly into design. So, nudges can be targeted at groups or individuals. We can do so in a way that is prescriptive, and that will result in best outcomes for that group or individual. If designers follow a series of steps provided, they can nudge in an ethical manner.

In parallel with literature on _how to think_ about choice architecture in design, studies have been conducted on its efficacy in design.

In _Nudge & Influence Through Mobile Devices_, Eslambolchilar et al. propose a workshop that argues mobile devices are well-situated toward facilitating nudges because they are able to collect data on their owners, in situ, which in turn could be used to influence habituated behaviors. Central to the authors’ assertion is the concept of peer influence (“comparing individual performance with relevant social group performance [and] social network sites running on the device facilitate communication of personalized descriptive social norms that relate to the participant’s self-defined community.”). We see, again, that nudges can be targeted based on data collected from an individual. Problematically, the authors in this case also conflate nudge and choice architecture with social architecture -- particularly, that of mobile and online social networks. But social networks are built on what and who we know, and therefore limit our choices by creating echo chambers -- nudges, in this case, would only come from a limited selection of options that our own networks provided us with.

Choice architecture and social media seem to be a topic of interest in design. In _Nudge for Deliberativeness: How Interface Features Influence Online Discourse_, Menon et al. conducted a between-subjects experiment to determine how nudging could mitigate invective in online discourse. The nudges used to influence behavior of the participants were: _Partitioning Text Field, Reply Choice Prompts, and Word Count Anchor_. Perhaps unsurprisingly, the nudges seemed to actually increase the number of arguments that people made. The only nudge that seemed to have a positive effect was the _Reply Choice Prompt_ asking people if they wanted the original poster to reply to their argument. The authors admit that the social pressure might, however, reduce creativity and reduce peoples’ willingness to speak their minds.

There is, then, a prescriptive way to apply choice architecture to individuals or groups. That prescription can be administered in many ways, including through social media, which can be used as the mechanism to enforce change (through peer pressure, etc). It is at this point that we must ask: where is the design in choice architecture and nudge? This methodology seems more like a social scientific approach to human behavior than it does an aesthetic one.

### Design and Nudge are Incongruent

Caraban et al. (referenced earlier) assert that there are 23 distinct manifestations of nudge in design. In 23 Ways to Nudge, a list of nudge types are plotted along axes representing transparency to user (non-transparent to transparent) and response elicited from user (reflective to automatic, explained more below). See Figure 5.

First, the authors argue that dark pattern, or ethically questionable, nudges fall within the _automatic and non-transparent_ quadrant. These nudges might aim to deceive or trick the user by taking advantage of their brain’s automatic response systems while hiding a nudge’s true intentions. Social media companies are notorious for these kinds of nudges, or dark patterns, when their algorithms intentionally incentivize emotionally manipulative content in order to increase attention and engagement. In this case, the nudge (content that will incite anger) is not transparent to the user, while the response (anger) is automatic.

Conversely, positive or benign nudges fall within the _reflective and transparent_ quadrant. These kinds of nudges remind, suggest, or demonstrate possibilities. When your music player suggests songs based on your preferences, this is a positive nudge that demonstrates or suggests other possibilities, leading you down a path toward additional discovery of songs. The nudge is reflective, in that it offers choices without coercion, while being transparent in its actions (“You might also like…”).

Finally, we have the puzzling _automatic and transparent_ quadrant. An automatic and transparent nudge might be something done without the user’s explicit approval, but is transparent in its intent and typically gives the user the ability to change it. Default settings on a smartphone and ambient feedback are some such examples. A phone, by default, might be set to provide automatic updates for security purposes. Yet the user of the phone also has the ability to turn these settings off through a transparent settings panel if they no longer wish to partake in automatic updates.

 ![nudge5.png](https://writebook.jaymargalus.com/u/nudge5-Hpxjup.png)
_(Figure 5)_

This leaves us with one quadrant empty. A keen observer may have noticed that none of the nudge types fall within the _reflective and non-transparent_ section of the chart. But what is a reflective and non-transparent act? One might assume that it is something that elicits thought from the user, but does so in a way that eludes perception.

It seems curious to develop a four-quadrant model with one quadrant entirely empty. What fits there?

Going back to an earlier assertion, we know that aesthetics are not transparent and are _inwardly affecting_. One of the most recognized pieces of music to come from the Agnus Dei section of Roman Catholic Latin mass is the round _Dona Nobis Pacem_ which translates to “give us peace.” The song is often sung during Christian Holy season as a call to unification. But it also has a secular history, such as when it was sung in the streets during the reunification of Germany by many who were likely unaware of the song’s literal text, but who understood its intent.

Like most art, _Dona Nobis Pacem_ eludes literal understanding -- its contents are not entirely transparent. But understanding the literal piece is also unimportant, as demonstrated by its adoption from those who didn’t even understand its words, but understood its intent. The song, then, is able to inwardly affect those who hear it while eliciting a reflective response. Looking back at Figure 5, aesthetics seem to fit nicely, then, into a quadrant that is non-transparent and reflective. _Dona Nobis Pacem_, Van Gogh’s _Starry Night_, and yes, even the objects of design channel aesthetics to evoke these reflective and non-transparent effects. If the authors of _23 Ways to Nudge_ are correct, then, nudge does not manifest itself in this quadrant because it is incongruent with aesthetics.

If nudge is incongruent with aesthetics, and if aesthetics is a central feature of design, then nudge is also de-facto incongruent with design.

Including nudge in design contradicts the aim of design -- almost without being aware of these very contradictions. In design, aesthetics serve more than the user's needs, but also their desires. “Design ideally is about service on behalf of the other—not merely about changing someone's behavior for their own good or convincing them to buy products and services.”

Yet choice architecture is precisely an exercise focused on changing behavior without addressing the vital role that aesthetics and service to the user have in design. Design addresses our desires by treating objects as particular representations of the truth, while disregarding the notion that we can understand the truth itself. We are godlike in creation, but not in our ability to understand what those creations will bring forth.

In this way, design stands in opposition to science and, by extension, choice architecture which attempts to exert _process_ and _prescription_ upon the very nature of being. Choice architecture would claim that we are godlike _both_ in our creations _and_ in our understanding of what those creations will bring forth. _We can change people and make them better.
_
Yet while we can predict _particular aspects_ of what our creations will manifest, it is impossible to predict the sum-total of changes we introduce into a system. All we are left with, then, is the pursuit of aesthetics. Creating objects that interact with other objects, not to affect how they behave, but toward the end of understanding the world through the act of creation and engagement with it.

Design is not the pursuit of perfection through reduction or abstraction. It is not perfection of the individual, or the things the individual interacts with, or the things that interact with other things. As Nelson and Stolternman note

> … it is important to appreciate the danger of creating a design motivated by a quest for the absolute ideal design solution. This often leads to the creation of something that cannot be supported, maintained, afforded, or controlled by the proposed beneficiaries of the design. Attempts to create perfectly glorious designs can bring ruin, or the threat of ruin, because they are not formed by the intention of designing the adequate, but by the unrealized quest for the comprehensive and utopian.

There is no utopia, and no such ideal exists within our grasp. No processes or micromanaging will get us closer to perfection.

Design is the rejection of outcomes-oriented processes like nudge, and the embrace of the warm, noumenal possibilities contained within aesthetics. Without aesthetics, there is no metaphor and without metaphor, there is no sublime. By introducing choice architecture into design, we not only apply a methodology incongruous with the practice, but we remove an opportunity for the individual to access the very thing that informs design practice -- creativity through the messiness of the universe.