The mercy of the default

Somewhere in my editor there is a settings page with a few hundred switches on it. I have opened it maybe three times in as many years. I know, the way you know a road you have never driven down, that some of those switches would make my days easier. A keybinding to save a motion I make forty times an hour. A layout that matches how I actually work. People write careful posts about exactly these, and I have read a few, felt the small thrill of I should set that up, then closed the tab and gone back to work, because the work was already happening and the default was already fine.

That is the whole life of a default, and it has almost nothing to do with being tricked.

The default didn't win by hiding

We talk about settings as if the only reason anyone keeps the default is ignorance: you didn't know the option existed, or someone buried it three menus deep. Sometimes that is true. Mostly I know precisely where the switch is. I keep the default because flipping it is work, and the work is the expensive part.

Configuring is unpaid labor. An hour reading other people's opinions about tab width. An evening lost to a config file I will only tweak again in six months. Attention spent now against a payoff that is small and far off. And attention is not the resource it was. Gloria Mark has spent two decades timing how long people hold focus on a single screen, and the line only falls: about two and a half minutes per screen in 2004, seventy-five seconds by 2012, and lately around forty-seven seconds, a number five separate teams have now landed on. I don't feel that as a statistic. I feel it as an unwillingness to spend any of a shrinking budget on a settings page when real work is one keystroke away. The default doesn't win because I missed it. It wins because it is cheaper, and it keeps winning, quietly, by that same small margin every day.

The part I was slower to admit

There is a second thing underneath the first, and it took me longer to say out loud. Some days I am not merely unwilling to choose. I am relieved not to.

By the time I sit down to write code I have already decided what to eat, what to wear, what to answer and what to leave, which of three small fires to put out first. Deciding is the texture of an adult day. So when a tool quietly settles something for me and turns out to be right, I don't feel robbed. I feel lighter. One less thing to hold open.

The popular name for this is decision fatigue, and I want to handle it carefully, because the tidy laboratory version has come apart. The idea that willpower is a fuel that burns down over a day, what psychologists called ego depletion, did not show up when twenty-three labs ran the careful experiment together and found nothing. I am not going to tell you your self-control has a gauge on it.

I don't need the gauge to explain why the toggle never moves. Samuelson and Zeckhauser named the real thing back in 1988, and unlike ego depletion it has held up: status quo bias, the plain human pull to leave things as they sit. A default is not only the lazy path. It reads as a recommendation, the maker murmuring most people are fine here, and to someone already spent that murmur is a kindness you take gratefully. We don't keep the default because we missed the switch. We keep it because deciding is exactly the thing we were trying to do less of.

There is no flat floor

Put those two together and the usual story about on-by-default AI starts to wobble.

The usual story is theft. A company ships an assistant switched on. Microsoft did it with Recall, until a backlash shoved it back to opt-in. Google turned Gemini on across Gmail and Chat and Meet. LinkedIn flipped your profile into training data and left you to find the off switch. The story tells itself: they knew you wouldn't look, they leaned on your inertia, shame on them. And the fix is always the same one word. Opt-in. Let people choose.

I used to believe that fix completely. I believe it less now, for two reasons.

The first is that there is no neutral setting to ship. You cannot offer a feature with no default. Off is a default too, a tilt toward you probably don't want this, and it is every bit as sticky as on. Thaler and Sunstein made the point years ago and it still holds: whoever lays out the choice has already shaped it, and "don't shape it" is not on the menu. The floor is never flat. The only real question is which way it leans, and toward whom.

The second reason is the uncomfortable one. If part of why defaults stick is that I am relieved to have the decision lifted off me, then the person on the other side of my toggle is not an innocent I am robbing. They are tired, and they are handing me the choice on purpose. That makes the duty heavier, not lighter. Inertia I could exploit with a clean conscience and a straight story about consent. Trust I cannot. When an organ-donation default moves a country from twelve percent donors to over ninety, nobody on either side feels tricked. They each did the easy thing, which happened to be the thing their system quietly recommended. The power was never in hiding the switch. It was in being the one who gets to decide what counts as easy.

The floor is never flat. The only question is which way it leans, and toward whom.

What I can't tie off

So I have stopped reaching for "make it opt-in" as though it settled the matter. It moves the lean, it does not remove it, and it pretends the user wanted to be asked, when half the time the kindest thing you can do for a tired person is decide well on their behalf and be right.

That is the part I can't close. A good default is a real mercy. It is the senior engineer setting sane configuration so ninety-five other people never have to learn what they don't care about. It is the feature that helps you without making you hunt for it first. And the same relief that makes that a gift is the exact thing a worse company reaches for when it wants something from you and is counting on you being too spent to say no. From the tired side of the screen, at forty-seven seconds a glance, I honestly cannot always tell the two apart. The longer I spend building things other people will leave on default, the less certain I am that anyone downstream of my toggles can tell either.

Sources

• Do Defaults Save Lives?, Eric J. Johnson and Daniel G. Goldstein, Science, 2003 (organ-donation consent under opt-in versus opt-out defaults). SSRN
• Status Quo Bias in Decision Making, William Samuelson and Richard Zeckhauser, Journal of Risk and Uncertainty, 1988; supported again by a preregistered replication in 2021.
• A Multilab Preregistered Replication of the Ego-Depletion Effect, Hagger et al., Perspectives on Psychological Science, 2016 (null result, d = 0.04). Frontiers commentary
• Gloria Mark on shrinking attention spans, Speaking of Psychology, American Psychological Association, 2023. apa.org
• Nudge: Improving Decisions About Health, Wealth, and Happiness, Richard Thaler and Cass Sunstein, 2008 (choice architecture; there is no neutral design).
• "Microsoft reverses course, makes Recall feature opt-in," The Record, 2024. therecord.media