Exhibit 04
Not all defaults are benign. A taxonomy of manipulation: the deliberate design choices that use default psychology against the people using the product.
The Shift
Defaults are powerful because they exploit real psychology: status quo bias, loss aversion, cognitive load. Dark patterns are what happens when those forces are deliberately inverted.
In 2010, UX researcher Harry Brignull coined the term "dark patterns" to describe interfaces that appear to serve users but are engineered to benefit the company at the user's expense. The mechanisms are the same ones this museum has explored: friction added where it helps the company, removed where it steers toward a purchase. Frames chosen to make the preferred option feel inevitable. Defaults set to maximize data collection, not user welfare.
What distinguishes dark patterns from ordinary bad design is intent. Bad design is usually an accident of constraints, competing priorities, or deadline pressure. Dark patterns are the product of deliberate A/B testing: designers knew users were confused, ran experiments, and chose the confusing version because it performed better for business metrics.
In documented cases, internal documents show explicit discussion of how to prevent users from finding the opt-out. Amazon Prime's cancel flow, at its most complex, required navigating six separate screens designed to discourage completion. The FTC sued over it in 2023. That is not design failure. That is design success for the wrong goal.
The scale is significant. A 2022 FTC report found dark patterns in the vast majority of popular subscription services. A 2019 crawl of over 11,000 shopping websites identified 1,818 distinct dark pattern instances across 183 sites. The EU's Digital Markets Act, enacted in 2022, explicitly bans several categories. These are not edge cases. They are industry practice.
Exhibit A
The tablet spins around. Three big buttons: 20, 25, 30 percent. The cashier is right there. Somewhere below, in small gray type, is a way out.
Before this screen reached every counter in America, it was in the back seat of a New York City taxi, where a quirk between two payment vendors created one of the largest natural experiments in the history of defaults. Pay for your coffee on the terminal, then read what the data says just happened to you.
In 2009, New York taxis processed credit card payments on screens built by two competing vendors. One suggested tips of 15, 20, or 25 percent. The other suggested 20, 25, or 30 percent on fares above $15. Same city, same cabs, same riders. The only difference was the buttons.
Economists Kareem Haggag and Giovanni Paci analyzed 13 million of those rides. The higher suggestions raised the average tip by 27 to 30 cents over a baseline of $2.22, roughly 12 percent more money from nothing but a label change. Multiplied across New York's ride volume, the buttons were worth millions of dollars a year.
There was a catch. Screens with the higher suggestions were over 50 percent more likely to receive no tip at all. Push the anchor too high and some riders push back with zero. Psychologists call that reactance, and the authors flagged it as the built-in cost of an overreaching default.
Their 2014 paper described the setup as a case in which default effects were exploited by a for-profit industry. Then tablet point-of-sale systems carried the same three-button screen to every coffee shop, food cart, and bakery in the country, with the percentages creeping upward and the escape hatch shrinking. The experiment never ended. It changed venues.
What you just felt has names. Six of them are below.
Six Techniques
Each card contains a live example. Interact with it, then see what was done to you.
The opt-out is written to make declining feel foolish or ungrateful. The choice is technically free. The framing is not.
Get 15% off your first order
Join 200,000+ savvy shoppers. Unsubscribe anytime.
The phrase "I prefer paying full price" is not a neutral description of your preference. It is a label designed to make the refusal feel irrational. You are not choosing to pay more. You are choosing not to share your email. The framing punishes that choice with language that implies a character flaw. Studies show confirmshaming reduces opt-out rates by up to 25%.
Consent forms written in deliberately ambiguous or double-negative language. You think you opted out. You opted in.
Communication preferences
You are opted into everything.
Box 1: "Uncheck if you prefer NOT to receive" means leaving it checked = you want emails. Box 2: "Deselect if you do NOT wish to opt out" means leaving it checked = you accept ads. Both were checked by default. The double-negative syntax is so hard to parse that most people give up and leave everything as-is, which is the entire point.
Signing up takes one click. Cancelling takes a phone call, a retention interview, and a waiting period. The asymmetry is not accidental.
Click "Start free trial" to see how easy it is to get in.
One click in. Six steps out.
Amazon Prime's cancel flow at one point required navigating through six separate screens, each offering discounts or asking you to reconsider. The FTC sued Amazon over this in 2023, calling it an illegal dark pattern that enrolled customers without consent and made cancellation unreasonably difficult. Amazon settled for $25 million.
Your attention is guided toward the option that benefits the company. The option that benefits you is technically present. Just visually buried.
The "Recommended" badge, gold border, and center position are not neutral. They are a deliberate visual hierarchy designed to make Pro feel like the default choice. Eye-tracking studies show most users do not consciously register the free tier in pricing layouts like this. The free plan is visible. It is not designed to be seen.
Countdown timers, "Only 2 left!", and "47 people viewing now" manufacture pressure. Studies show these figures are frequently fabricated.
All three signals are fabricated.
The countdown would have reset to 10:00 when it reached zero. The stock count has shown "3 left" for weeks. The viewer count is generated by a random number function. The FTC has taken action against multiple retailers for fabricating these signals. Urgency and scarcity cues increase conversion by 9 to 14% on average, which is why they persist even after regulatory attention.
The price shown on the product page is not the price you pay. Fees appear only at the final checkout screen, after your commitment is established.
Order summary
27% more than advertised.
The $49.99 you clicked on became $63.74 at checkout. This technique, called "drip pricing," exploits commitment and consistency bias: by the time the fees appear, you have already spent time selecting your seat, entering your details, and investing in the purchase. Abandoning it now feels like a loss. The FTC estimated drip pricing costs US consumers $8.8 billion annually.
The Scale
Dark patterns are not a fringe phenomenon. They are industry standard.
The Research
Five bodies of work that defined, catalogued, and measured dark patterns in the wild.
The talk and website that named the phenomenon. Brignull, a UX researcher, began cataloguing instances of deceptive interface design and built the first public taxonomy. The site darkpatterns.org became a community archive of documented examples, and the term entered regulatory vocabulary within a decade.
The study that predicted the tip screen. Analyzing 13 million New York City taxi rides, the authors showed that raising suggested tip buttons from 15/20/25 to 20/25/30 percent increased average tips by roughly 12 percent, while making zero tips over 50 percent more likely. Published in AEJ: Applied Economics, it documented default effects "exploited by a for-profit industry" years before that screen reached every coffee counter in America.
The first major academic taxonomy, presented at CHI 2018. The authors identified five high-level dark pattern strategies: nagging, obstruction, sneaking, interface interference, and forced action. This framework gave regulators and researchers a shared vocabulary and remains the most widely cited classification system.
The largest empirical study to date. Researchers automatically crawled 11,076 shopping websites and identified 1,818 instances of dark patterns across 183 sites. Larger, more-visited sites were significantly more likely to use dark patterns, suggesting deliberate adoption rather than accidental design. The paper won a Best Paper Award at CSCW 2019.
The FTC's formal report on dark patterns in subscription services and e-commerce. The agency documented widespread use of hidden fees, difficult cancellation flows, and manipulative consent interfaces among major commercial platforms. The report preceded enforcement actions against Amazon, Vonage, and others, and recommended legislative action to codify restrictions.