Scientists and Saints Are Laughing at the Same Joke

Here's a joke I can't stop telling, mostly because nobody who's in it ever notices they're in it. A scientist and a saint walk into a bar. The scientist looks at the saint and thinks: how does a grown adult defend a belief like that, long past the evidence, purely to keep his standing with his little tribe? The saint looks at the scientist and thinks: how does a grown adult defend a belief like that, long past the evidence, purely to keep his standing with his little tribe? They are describing each other word for word. They are both completely right. That's the joke.

I'm being more literal than you'd like. So let me put the two of them side by side, each clutching a distinction one hair wide, each prepared to die on it.

In the year 325, a few hundred bishops met at Nicaea to settle whether the Son was of the same substance as the Father or merely a similar one. In Greek that's homoousios against homoiousios. The entire quarrel turns on a single letter, the iota.¹ Men were exiled over that letter. "Not one iota of difference" is a phrase in the language precisely because, right here, the iota was the whole of the difference.

Now the scientist's version of the same bit. A result is "significant" if its p-value lands under 0.05, and unpublishable noise if it comes in at 0.051.² The line is arbitrary. Ronald Fisher, who set it, more or less picked a round number and called it handy.³ And careers are made and broken on which side of it a finding falls. The bishop had his iota. The scientist has his second decimal place. Tell me that isn't the same man, at the same bar, ordering the same drink.

Every joke has a mark: the one person at the table who's sure the story is about somebody else. The saint knows the mark is the scientist. The scientist knows it's the saint. They are both right.

Why the joke is actually funny

A good joke earns its laugh by being true, and this one is true in a way it took psychologists a while to measure. Dan Kahan's team at Yale gave people a fiddly maths problem. Dressed up as data about a skin cream, the better you were at maths, the more often you got it right. Dressed up as data about gun control, strong maths skills made people better at reaching the wrong answer, as long as the wrong answer flattered the politics they already held.⁴

Read what that means. Being clever did not make anyone more objective. It made them a sharper lawyer for the side they'd already joined. That's the engine of the whole joke. The cleverness each camp is proudest of is the exact instrument that keeps it fooled. The line you'd put on your business card is the line that makes you the punchline.

The part where it's on you

Here's where I stop letting you laugh for free. You've been reading this with a particular small smile, the one that says you can see both fools clearly, from the outside, from a safe distance. That is the mark's smile. You think you're the bartender. You're at the bar.

We even know the shape of it. Emily Pronin and Lee Ross called it the bias blind spot: people reliably detect bias in everyone but themselves, and the effect does not spare the educated, the irreligious, or the proudly rational.⁵ The feeling of getting the joke, of being the one sober person watching two drunks argue theology and statistics, is not evidence that you're sober. Statistically it's the surest sign you're as deep in as they are and have only stopped noticing.

The objection I can't wave away

A scientist has been getting angrier at me for several paragraphs, and the objection is a good one, so I should hand it over properly instead of dodging it. Science is not just the tribal instinct in a lab coat. It is the tribal instinct with an exoskeleton bolted on, one built specifically to catch the instinct in the act. Replication. Peer review. Predictions registered in advance, where you can be made to look wrong in public. Results that have to survive people who would be glad to see them fail. The bishop has none of this. There is no experiment that retires a creed, and no council convenes to un-canonise a saint because the miracle failed to reproduce.

So the tidy symmetry I've been selling you is too tidy, and I knew it while I was selling it. Up close, the p < 0.05 brawl and the iota brawl really are the same animal, the same status game over the same hair-fine line. Step back, though, and only one of the two communities has spent a century building machinery to overrule its own priests. The replication crisis people keep citing is not science failing. It is science doing the one thing religion has no organ for: turning the instruments on itself, in public, and not liking the result.⁶

How much that actually saves us, I don't know. Some weeks it looks like the only thing standing between us and the bishops. Other weeks it looks like a slow and expensive route to the tribe's preferred answer. I go back and forth on it. I'm not going to pretend I've settled it for you in an essay.

The smaller claim, which I'll stand behind

Here is the part that doesn't let me off either. You do not run replication on your own beliefs. There is no peer review for the things you were sure of this morning. Whatever error-correction science built at the institutional level, you are mostly not running it on yourself, and neither am I. The instinct is stock equipment. The exoskeleton is sold separately, and most of us never bother to install it.

So the joke I actually believe is duller than the one I opened with, and worse. It isn't that everyone is a fool. It's that the fix exists, in plain view, and we mostly can't be bothered to reach for it. The man who wrote this very much included.

Next time: we build a camera fast enough to freeze an electron, and it somehow walks the cleverest people in the room straight back to God. They just won't call him that.

❧