Pick one. A: 2,500 with a 33% chance, 2,400 with a 66% chance, nothing with a 1% chance. B: 2,400, guaranteed.

Most people pick B — 82% of them, in the room where this was actually asked. A sure thing over a gamble that's only slightly better on average. Reasonable enough. Nobody likes risking a solid 2,400.

Now a second pick, same room, same people. C: 2,500 with a 33% chance, nothing otherwise. D: 2,400 with a 34% chance, nothing otherwise. This time 83% picked C (Kahneman & Tversky, 1979).

Here's the catch: choosing B in the first pair and C in the second doesn't add up. The second pair is just the first with an identical 66% chance of winning 2,400 stripped out of both sides — nothing about how the two gambles compare to each other should have changed. And it wasn't just two different crowds landing on two different averages. Tracked person by person, 61% of individual respondents made exactly that contradictory pair of choices (Kahneman & Tversky, 1979).

None of that would surprise anyone who's read Dan Ariely. Predictably Irrational is basically a tour of this same territory — anchoring, decoy pricing, the free-shipping trap — and by now it barely counts as news that humans aren't the tidy rational actors classical economics assumed (Ariely, 2008). What's more interesting is a quieter finding sitting right next to it. People don't just make the inconsistent choice. Ask them beforehand how they'd behave, or ask them afterward whether the pattern applies to them, and the answer is rarely a flat denial. It's softer than that — something closer to sure, that happens, probably not to me, or not that much.

Flip the sign on the money and the pattern gets stranger. Kahneman and Tversky reran their gamble pairs as losses instead of gains, and the preferences didn't just weaken — they mirrored. Most people preferred an 80% chance of losing 4,000 over a certain loss of 3,000, the exact opposite of how they'd treated the same odds when the numbers were positive (Kahneman & Tversky, 1979). Certainty is comforting when there's something to protect and unbearable when there's a loss to avoid — the same mental shortcut producing risk-aversion in one direction and risk-seeking in the other, depending only on which side of zero the number sits.

They found something similarly strange in insurance. Offer people a policy at half the normal premium that only pays out half the time — a straightforward discount for reduced coverage — and most turn it down, even when the arithmetic favors it (Kahneman & Tversky, 1979). Burglar alarms, new tires, quitting smoking: all the same shape, a partial reduction in risk for a partial cost, and all of them chronically undervalued relative to the all-or-nothing version. Nobody plans to be the person who skips the alarm system for a bad reason. They just are.

The self-discount shows up even when nobody's gambling at all. Ask people to rate their own susceptibility to a list of well-known cognitive biases against everyone else's, and the asymmetry holds up — not just in one lab, but across cultures, and starting in childhood (Pronin & Hazel, 2023). A pre-registered replication out of Brazil in 2024 found the same gap in a population nothing like the original Stanford undergraduates the effect was first documented in back in 2002 (Seda et al., 2024; Pronin, Lin, & Ross, 2002). It travels. It isn't a quirk of one country's students being asked one specific question.

Michael Lewis spent an entire book on the two men who documented all of this (Lewis, 2016). What stuck with me, going back to their actual data, is that the subjects were Israeli and American university students and faculty — people who looked a lot like the researchers running the study. The bias blind spot isn't something that happens to other people. It's the default setting, running in the room where it was discovered, too.

That tracks with something Jordan Peterson would call a status problem more than a math problem. Believing yourself to be a reasoning creature — someone who arrives at decisions rather than gets pushed into them — isn't just a nice self-image. It's load-bearing. Admit that a coin flip's framing moved you more than the actual odds did, and you're not just wrong about a gamble. You're wrong about being the kind of person who doesn't get moved that way. The explanation that shows up after the choice isn't lying, exactly. It's protecting the hierarchy where the "rider" outranks the "elephant." It's the same architecture this project ran into all the way back in Post 1 — the intuition moves first, and the part of you that explains why shows up afterward, doing PR for a decision it didn't make (Haidt, 2001).

What I keep sitting with isn't whether the blind spot is real — the data's fairly settled on that. It's the size of the gap between the two moments: the one where the framing quietly does its work, and the one, sometimes seconds later, where you'd swear under oath you weighed it fairly. Worth wondering, next time a choice feels obviously reasoned through — was it, or did the story just arrive first?