Nudge
Influencing without Infringement
Author: Richard Thaler and Cass Sunstein
Recommended by my startup co-founder who I recently met for the first time 5 years after we gave up Mohnish Shah while chatting in a Target parking lot at 11pm.
Choice Architecture
Small changes in context can greatly influence the decisions of people, the people who have the power to make these changes are called choice architects.
Choice architects have the responsibility for organizing the context in which people make decisions, and yet, most choice architects don’t realize they are one.
Parallels can be drawn between regular architects and choice architects:
There is no such thing as a neutral design choice
Every choice will be interpreted by a multitude of subjectivities - each person will interpret the context you provided in their own way
Arbitrary decisions will inevitably influence downstream interactions
Given that even the smallest details can have a major impact on people’s behavior, you should assume that everything matters. The power of these small details comes from the focusing of attention in a particular direction - this insight can be both paralyzing and empowering.
Good architects realize that they can’t build a perfect building; it doesn’t exist; but they can make choices that have beneficial effects.
This concept strikes me as oddly tantric. The idea of directed attention without infringements or constraints captures the essence of spacious passion as outlined in Vividness. The idea that there is no eternal truth or inherent flaw in the universe, but that pragmatic improvements can be made is also one that many find both paralyzing and empowering - the world is our playground and there are no right answers, where do we even begin?
Perhaps we can start by nudging people in directions we think will serve them well.
Libertarian Paternalism
This is a key mindset that pervades the book, and perhaps not a combination many would find particularly endearing on first impression.
To make things worse, they’re somewhat contradictory. However, when properly understood, they show a lot of attractive qualities - even more so when combined.
The first hurdle we must overcome is the baggage that the words now carry beyond their original definitions due to dogmatists.
Libertarianism - as long as people are not harming others, they should be free to do what they like.
Libertarianism posits that we should strive to design policies that maintain or increase freedom of choice. It simply means liberty-preserving, not burdening those who want to exercise their freedom.
This may be caveated by two points:
When people are inflicting harm on others, freedom of choice is not the best idea
When people are inflicting harm on their future selves, nudges might not be enough
Paternalism - it is legitimate to influence people’s behavior to make their lives longer, healthier, and better.
Paternalism posits self-conscious efforts should be made by institutions to steer people’s choices in beneficial directions, with “good” being defined by the chooser. We should make it easier for people to make the choices they would have made given complete attention, complete self-control, and complete information.
Those who reject paternalism are typically implicitly committed to the idea of Homo economicus, the “economic man” - the notion that every person thinks and chooses well, and fits the typical economist’s depiction of a rational, self-serving human being. Counterexamples of the idea of the “economic man” are obvious and abundant: diet, smoking, drugs, drinking, etc.
The reason for the nonexistence of Homo economicus is not that people make incorrect decisions, this is bound to happen because the future is unknown - they are however, expected to make unbiased forecasts, predictions cannot repeatedly err in predictably, and people tend to be predictable.
Libertarian Paternalism Weakens Paternalism in Exchange for Liberty
Libertarian Paternalism is a soft, non-intrusive type of paternalism - choices are not blocked or forced. Choice architects simply try to nudge people in directions that will make their lives better.
By accepting paternalism and rejecting the idea of Homo economicus, we accept that humans err predictably. This alone doesn’t allow us to do much, but when paired with libertarianism, allows choice architects to act on the predictability of humans to nudge them towards better choices.
Avoiding Reactance - one of the major benefits of the soft approach to paternalism. When people feel ordered around, they might experience negative feelings that lead them to do the opposite of what was asked.
What is a Nudge?
A Nudge is an aspect of the choice architecture that predictably alters people’s behavior without forbidding any options or significantly changing their economic incentives.
Nudges are not:
Taxes
Fines
Subsidies
Bans
Mandates
Sources of Systematic Heuristics & Biases
Anchoring - people tend to be biased towards numbers close to ones they already know.
When asked to guess the population of a mid-sized city, residents from a larger city will overestimate the population, and residents from a smaller city will underestimate it.
Availability - people tend to be biased towards things where examples come to mind more readily.
People will perceive crimes reported on the news to have a higher likelihood of occurring.
Representativeness - people tend to be biased towards things that fit the stereotypes in their heads.
People will perceive a gender studies major to be more likely to be a “feminist banker” than a “banker”.
Unrealistic Optimism / Overconfidence - People tend to be overconfident about their own abilities.
Most people think they are above average.
Gains and Losses - people are loss-averse, they react to losses twice as strongly as gains.
People value things more once they own them. People don’t want to lose money.
Status Quo - people have a tendency to stick with the current situation - this is pretty exploitable.
People will tend not to cancel subscriptions that were initially free.
Framing - people will react differently to the same thing said in different ways.
A 90% chance of survival sounds better than a 10% chance of death.
Thinking, Fast and Slow.
Nobel Laureate Daniel Kahneman, who authored the book which this sub-section is named after, introduced the idea of the brain consisting of two components or systems.
System 1 - Automatic system that is fast and intuitive
System 2 - Reflective system that is slow and reflective
The latter typically leads to better or more accurate thoughts at the expense of taking longer to rethink the original answer given by System 1. If everyone always used System 2, we could more reliably rely on humans to make rational choices. However, we cannot rely upon everyone to do this - policies have to be designed for system 1 thinking.
Resisting Temptation
When facing problems around self-control, we can borrow the framework outlined above that posits the individual containing two semiautonomous selves - Systems 1 and 2.
System 1 adopts the personality of a myopic “Doer”, who has a strong will and impulsive thoughts.
System 2 adopts the personality of a farsighted “Planner”, who tries to promote your long-term welfare while resisting the temptations presented by System 1.
While strategies of various scales can be employed by the Planner to reign in the Doer, sometimes third-party intervention is required. In some cases, this is implemented from a pure paternalistic approach (eg. government intervention) but often we prefer something less intrusive.
A libertarian paternalistic approach often involves the provision of voluntary self-control services that the Planner can enroll in to remove the potential for the Doer to opt for unwanted behaviors from the action space.
Herd Mentality
Understanding how social influences work is important to choice architects for two reasons:
Most people learn from others, this is generally a good thing and is how individuals and societies develop. Misconceptions also come from others, and social influences can cause people to have false or biased beliefs.
This is one of the most effective ways to nudge (for good or for evil) as we are easily nudged by other humans; we like to conform.
Social influences fall into two main categories:
Information - if many people do or think something, their actions and their thoughts convey information about what is best for you to do or think.
Peer pressure - if you care about what other people think about you (particularly those you relate to or hold in high regard), you might just go along with the crowd to avoid potential (often imagined) social consequences.
Collective Conservatism is the tendency of groups to stick to established patterns and perpetuate arbitrary traditions even as new needs arise, and a common outcome of the herd mentality. Of course, the more problems a practice causes the more likely the group is to shift; but when uncertainty is still part of the question, people may prefer sticking to what they know.
Informational Cascades are a phenomenon that occurs when an initial opinion nudges subsequent decision-makers who do not feel strongly about the given topic to conform to prior consensus.
Pluralistic Ignorance occurs when we follow a practice on the belief that other people like it. This can cause widely despised social practices to persist for the simple reason that people don’t know that other people also dislike a practice.
This can often be resolved by nudging people towards speaking and acting in accordance with their actual views, or simply informing people about what others are thinking and doing.
The Tools of the Choice Architect
Before answering the “how” of nudging, we should establish the “when”.
The golden rule of libertarian paternalism is to offer nudges that are most likely to help and least likely to inflict harm.
Opportunities for effective nudging typically arise when people face decisions that demand attention, introduce high complexity, lack immediate feedback, or are obscure and not easily parsed into simple terms.
A few key considerations when designing choice environments:
Reminders and Prompting
People have finite self-control, attention, and memory; this is why we make lists and set reminders - in case we forget. Reminders have become ubiquitous and can be a terrific form of nudging. For instance:
Checklists - Pretty straightforward. Works especially well when all relevant individuals are authorized to issue reminders.
Implementation Intentions - Elicit questions that prompt the user to state specifics about their intention to do something.
Advance Benefits, Defer Costs
Most people need encouragement for longer-term “investment-type” activities that are beneficial in the long term but provide little or no immediate gratification gratification.
Nudges can be designed to mitigate issues around self-control by providing immediate benefits and deferring costs for certain actions.
Easing Difficulty
People are less likely to need help with easier problems. We can even cope with harder problems if solutions are readily available to them. However, when solutions are scarce or not easily accessible, we can assist by lowering the difficulty through helpful nudges.
Frequency
Hard problems become easier through practice, and finding solutions becomes automatic. However, most important decisions do not come with many (if any) opportunities to practice. Difficult and rare choices are good candidates for nudges.
Feedback
In an unstructured environment without effective learning opportunities, practice can only help so much. Further, we may only get feedback on the choices we opt for, not the ones we forego. When feedback is ineffective or alternatives have not been explored, a nudge might be beneficial.
Guiding Through Uncertainty
When people are unable to predict how a choice will affect them, or do not understand the available alternatives, a nudge towards narrowing their options or even suggesting one may be welcomed.
Implementing Choice Architecture
If you want to encourage people to do something, make it easy - here are a few ways to do so:
Defaults
Defaults tap into people’s inclination to opt for the path of least effort or resistance - until the default becomes obviously bad enough for the user to pick otherwise. Defaults are ubiquitous, powerful, and to some extent, unavoidable; something has to happen if the user does nothing.
Required Choice - users must actively make a decision, this is a good strategy if the decision is important and easily overlooked, or requires the person’s explicit consent.
Prompted Choice - users are presented an option where a selection is not mandated, but acts as a strong recommendation by the choice architect.
Expect Error
A well-designed system expects its users to err and is as forgiving as possible. The idea is to identify sources of error and build “nudges” that prompt users to make their originally intended action into the system.
Give Feedback
Well-designed systems tell people when they are doing well and they are making mistakes. An even better system will preemptively tell you when things are about to go wrong.
If an overwhelming amount of feedback is given and the automatic system within the brain takes over, the feedback is rendered useless.
Mappings
An accurate mapping between a choice and an outcome improves the ability of people to predict the consequences of their choices with considerable accuracy.
A well-designed system should help people improve their ability to form good mappings and select the option(s) that will leave them better off.
Choice Complexity
People use different strategies to make choices depending on the action space and complexity of each option.
When faced with a small number of well-understood alternatives, we tend to closely consider all options and make trade-offs when necessary.
As the set of choices gets large, we may use alternative strategies that can lead us to sub-optimal outcomes. We may have to use simplifying strategies like “filtering” to eliminate options by their attributes, without considering trade-offs that may lead to more welfare.
A good choice architecture should provide structure in such situations, where choices are made more structured and simplified to the user; nudges can be immensely helpful to choice-making through well-designed sorting or personalized recommendations.
An important consideration is to remember that surprise and serendipity can be pleasant too - it may also be good to nudge people in directions they might not have originally considered choosing.
Incentives
Aligning incentives is standard economics, but we can enrich them by remembering the agents are humans who may err or act irrationally. One of the biggest factors to account for is salience - are all major and relevant incentives noticed by the decision maker?
Salience can be manipulated and good choice architects should direct people’s attention to incentives.
Taking Breaks
An often neglected yet important tool available to choice architects is scheduling intermissions. A properly timed intermission should be added without ruining the experience while allowing the users to stretch their legs (either literally or figuratively).
Curation
A tool available to choice architects to massively improve user experience is curation - the ability to provide surprises, serendipity, and delight to a customer; the ability to provide only the best of the best.
Curation can massively reduce the amount of navigation that needs to be done to find something you want given a sea of choices.
Fun
The first mantra of nudging is to make it easy to make the desired decision - choosing fun is easy. Fun piques our curiosity builds excitement and anticipation, and increases the willingness and even perceived value of performing a given activity.
Good and Evil
The corollary to the basic principle of good choice architecture is “if you want to discourage a behavior, make it harder”. This means employing punishments, obstacles, barriers, and disincentives.
Nudges and behavioral science have a dark side too, they can be used for both good and evil. People can be defaulted into decisions that will hurt them, harmful actions can be incentivized and made easier. Friction can be added to manipulate people and make it harder to obtain an outcome to make them better off.
Save More Tomorrow
Saving for retirement is a relatively new challenge for our species, arising from a mix of rising life expectancy and geographical dispersion of families. It’s also one of the situations where directional nudges are highly warranted - we are pretty confident that people will be better off by improving their retirement savings.
How much to save is something we won’t dive into, but in general, the cost of saving too little is much greater than that of saving too much. Yet in actuality, people’s retirement accounts reflect the opposite.
Many people know they should be saving more, but few change their behavior. We know people are favorably disposed towards being nudged on this issue, this is an opportunity to be helpful.
There are five important psychological principles relevant in this context to be aware of:
Most people want to and plan to save, but never actually do it.
Self-control restrictions are easier to adopt if they take place sometime in the future, not now.
Loss aversion - people hate seeing their paychecks go down
Money illusion - losses are felt in nominal dollars, not adjusted for inflation
Inertia plays a powerful role - people typically stick to their initial decisions.
Borrow More Today
Humans suffer from self-control problems rooted in being present-biased, putting undue weight on things they can have right now. To be able to retire, you have to figure out how to keep your spending less than your income - you need to save. Unfortunately, before this is greater problem: people borrow money to spend more today.
In particular, credit cards exacerbate this problem by providing instant liquidity in the form of borrowed money, allowing you to easily spend more than you have.
In a vicious cycle, the record of responsibility of using a credit card determines one’s credit score, which determines one’s interest rates. Using a credit card responsibly (paying it off in full every month and avoiding interest payments) requires significant self-control, and those who spend irresponsibly are plenty - Americans have racked up over $1 trillion in credit card debt. This is not unique to the US, China has surpassed this figure already.
Some stats to show their misuse in the US:
43% of balances are not fully paid off each month, only 31% are entirely paid off (the remaining have no balance or are inactive)
$1.1 trillion was the total credit card debt as of February 2020
The average credit card debt of a household was $6000 across 3.1 credit cards as of 2019.
Total interest payments were $121 billion with an average interest rate of 14-18%
Fees accounted for ~5.5% of cycle-ending balances in 2018, with just under half of that being late fees.
Objections to Nudging
Slippery Slope
“If we do some X, then there is a serious chance it will lead to Y, then Z. Even if X is a good idea, Z is scary. So, we shouldn’t do X unless we are willing to accept Z.”
The problem is this does not provide any evidence of an actual slope, and the track record of slippery slope forecasts is rather low.
Applied to the context of nudging, the common criticism is “First it’s nudge, then it’s shove, then it’s shoot.” - even though the whole point of nudging is to avoid shoving. Nudge creep is not a concern as nudges maintain freedom of choice by definition.
Freedom and Active Choice
Some people believe we should always provide people with all the information necessary to make an informed choice, and then let people choose for themselves.
Required active choosing is suitable when choices are simple, and doing so with complex decisions can be quite paternalistic - you’re forcing more responsibility onto someone, and you may not be respecting their preference to not make a choice. In certain situations, it can be intrusive, and in many situations, it is simply impractical to apply required active choosing.
Active choosing is a good idea, but in many domains, curation and well-designed defaults are a blessing.
Don’t Nudge, Boost
Some emphasize that in a free society, people have the right to be wrong.
This criticism does not directly conflict with libertarian paternalism, as it generally favors opt-out rights, as long as harm is not being caused to others.
The difference and criticism lie in that education should be emphasized over nudges. We should be boosting the capacity of people to make good choices rather than enlisting a choice architecture.
These are not conflicting ideas, nudging is simply an offering of well-curated options or suggestions that we can reject if we wish, but are often appreciated. Also, it is unrealistic to think that we can increase individual competence to the level where they are able to handle all complex choices in life on their own.
Nudging is Sneaky
Mandates, bans, and taxes are clear and direct - people know why they are there, and nobody is fooled or deceived. Some criticize nudges for being covert and manipulative - they affect people without their knowledge.
For most nudges, this does not apply; intentions are usually clear and transparent. However, if a nudge is not transparent and is devious in its implementation, it does fall into the “evil” category - these are not the kind of nudges we want.
Some nudges indeed work even if those who are affected by them do not notice they are there. In these situations, the design of the choice architecture is not hidden, but the reason may be. Studies have even found that telling people they have been nudged and the reason behind it increases the nudge’s effectiveness and impact.
So, are nudges manipulative? The philosophical consensus depends on whether or not it adequately respects people’s capacity for rational deliberation.
Beyond Nudging: Mandates and Bans
Libertarians are worried that we may start with nudges and move to shoving. Some critics of a more progressive persuasion have the opposite worry, that we will stop at nudging when stronger measures are required.
Firstly, it is clear that not all problems can be fully addressed with the light-touch intervention of nudging. They might help, but nudges are often not enough to control externalities for more serious matters.
Nudges are also unlikely to discourage many policy and decision-makers from taking stronger measures - they are often used in conjunction with stronger economic instruments or even mandates and bans.
The line when a nudge should become a mandate or ban is certainly blurry. The more general dilemma lies in enthusiastic paternalists who are worried about the risk of poor choices being made by individuals.
In truth, there isn’t a clear definition of when something should be. Both extremes are justified in certain contexts.
As long as people are making informed decisions about how to live their own lives, we should favor an attitude of humility and respect - and hence a presumption in favor of freedom of choice.
