Germany and Austria share a border, a language, a religion, a rough level of wealth, and a continent. They do not share an organ donation rate. In Germany, which asks people to opt in, roughly 12% of the population is registered as a donor. In Austria, where donation is the assumed position and objecting requires an active step, the figure sits at 99.98%. Nothing about the two populations differs in any way that would explain a gap of that size. What differs is a single line of pre-printed text on a form that most people never think about, and the gap it produces has been the workhorse example for a whole field of policy design ever since.
🧠 The Form Nobody Reads
The name for it is the default effect: the option you receive by doing nothing is the one you are most likely to keep. It belongs to nudge theory, and its definition is almost too plain. A default is the course of action that applies if the chooser never specifies anything.
The citation that made the organ donation comparison famous is a two-page 2003 paper in Science by Eric Johnson and Daniel Goldstein, titled Do Defaults Save Lives? Their answer was yes, and the mechanism they proposed was not that Germans secretly oppose donation while Austrians secretly support it. It was that most people hold no settled position on what happens to their organs after they die. They are not opposed. They are unattached. The form arrives before the preference does, so whichever box is pre-ticked becomes the answer by default.
That argument is still being tested in real legislatures. Wales switched to deemed consent in 2013, and England followed in 2020. The question every one of those debates had to answer was the same one Johnson and Goldstein raised: if the checkbox can move the number that far, who should get to decide what the checkbox says?
🤔 It Was Never Laziness
The lazy explanation is that people cannot be bothered to fill in a form. The research literature lists five explanations, and laziness is not one of them.
Cognitive effort comes first. When the options are close together or the judgement is genuinely hard, the mind falls back on a rule that reads, more or less, if there is a default, do nothing about it. In one set of studies, participants who were busy with a demanding second task were more likely to copy the snack choice they had seen a previous participant make. Load someone’s attention and they will follow whatever was already there.
Switching costs are the second, and they are dull in the honest way that real constraints are dull. Leaving a default can mean searching for information, paying a fee, mailing a form, hiring a lawyer. When the cost of leaving exceeds the benefit of arriving, staying put is the rational move.
Loss aversion is the third, and it is the prettiest. Take three pension plans, A, B and C, which trade present consumption against future savings. Make A the default and B and C now look like losses in the savings dimension. Make C the default and A and B look like losses in the consumption dimension. Whichever direction gets coded as a loss starts to dominate the decision, so the chooser keeps the default in both worlds. Same menu, two reference points, one winner: the one that was already ticked.
Then comes the part that makes defaults genuinely strange. A default is read as an implicit recommendation. Someone set this value, the reasoning goes, and if that someone seems competent and on your side, following them is sensible. Under that reading the default stops being a neutral starting position and starts changing what people want. The fourth mechanism is recommendation, and the fifth is meaning. Organ donation supplies the sharpest version: under an opt-out system, declining to donate reads as morally worse than it does under opt-in, and families may even read a registered donor as less sincere about it, which makes them more likely to refuse when the moment comes.
That last point keeps the headline number honest. Around a quarter of families refuse to donate a loved one’s organs, and refusal is the second gate. Changing the law opens the first gate. It does not walk you through the second.
🔗 The Defaults You Never Chose
Strip away the organ donation framing and the lesson lands wherever a product meets a person. Defaults are the cheapest design decision available and often the most powerful one: the order of a list, the state of a switch, the fields already filled in, the length of a conversation that starts without the user asking for anything.
This matters most for users who are least equipped to fight a bad default. Fewer options help, which is the old finding about choice overload. A sensible default beats even that, because it removes the decision rather than shrinking it. Anyone building anything for an older relative is choosing between a screen that asks and a screen that simply arrives in a workable state.
The same mechanism runs the other way in half the products on your phone. Free trials that renew silently. Marketing consent that is pre-ticked. Privacy settings that begin public. Automatic subscription renewals, which the default-effect literature discusses openly as a firm’s instrument. Researchers even have a category for the sleazy version, where the default is presented as though it were the only option and the alternatives sit three menus deep.
For anyone designing an AI companion, the question is blunt. What does the user receive if they never touch a single setting? That answer is the product. Everything else is a preference the user has to spend attention to claim.
🎲 Count Your Defaults
The literature sorts defaults into two families, and both make handy checklists.
Mass defaults apply to everyone. A benign default is the company’s best guess at what most people want, like a phone’s stock ringtone. A random default is assigned arbitrarily, which is what happens when nobody has a preference worth guessing and a researcher needs two comparable groups. A hidden default is the one where the alternatives have been quietly buried, and that is where design turns into something closer to a trap.
Personalised defaults are tailored: smart ones use static facts about you, persistent ones extrapolate from your past behaviour, adaptive ones update as you go.
Here is the exercise. Open your phone’s settings and find five options you have never once changed. For each one, guess who set it and what they wanted. The answer is rarely what is best for me. It is usually what was most convenient for the person setting it.
Nobody chose your defaults. That was the entire point of them.
The interesting question was never what you prefer. It is what happens when you say nothing.