I spent part of the weekend listening to two friends talk about religion. Neither convinced the other, which is normal, but the exchange got me thinking about what makes a belief dogmatic in the first place.

All dogma is wrong. Yes, I’m aware of the irony in making that flat, unqualified claim. Bear with me.

Everything is a probability

Let’s begin with the premise that no belief you hold sits at exactly 100% certainty (or exactly 0%). The universe itself is probabilistic, and our grasp of it is built from limited observation and limited processing. Both sides of our beliefs are fuzzy: the thing we’re trying to understand, and our understanding of it. Even your best-supported beliefs carry some residual chance of being wrong.

But nobody can live that way. You can’t walk around carrying 0.83 for every proposition you’ve ever encountered. So we round. A belief at 99.99% becomes “true,” one at 0.01% becomes “false,” and we get on with our day. This isn’t a flaw. Rounding is a heuristic, and heuristics are how finite brains get anything done. I round gravity up to 1 every morning and sleep just fine.

Where you round is the whole game

What matters is the threshold—how much evidence it takes before you round up, or down. A healthy threshold responds to two things, the evidence and the stakes. I’ll round a restaurant recommendation up at 70%. I want a great deal more confidence than that before I trust a bridge. Either way, the threshold moves when the information changes, and when the cost of being wrong changes.

Dogma is a threshold that doesn’t move.

Dogma comes in two polarities

The two versions are mirror images:

• Round up, regardless. “There’s nothing you can show me that would make me stop believing this.”
• Round down, regardless. “You’ll have to prove it with 100% certainty before I’ll believe it.”

As an example, pick any wellness fad. The devotee who can’t be talked out of it and the skeptic who won’t accept anything short of perfect evidence are doing exactly the same thing. One has fixed the threshold at “already true” and the other at “never provable.” Neither threshold has anything to do with the strength of the evidence, because neither is a function of the evidence at all. (The second one is sneakier, though, since skepticism often sounds more rigorous.)

Two ways to go broke

Dogma fails twice. The first failure is certainty. A 100% conviction is a check written against a balance that comes up short. The claim itself may even be true, but the confidence was never earned, because nothing in the universe ever transacts at the mythical price of 100% certainty.

The second (and far bigger) failure is process. A probability can only be refined with tools: inspecting assumptions, gathering evidence, testing predictions, entertaining the alternative long enough to see what it predicts. Orthodoxy exists to protect certain claims, and it protects them by removing those tools. Some questions are off limits. Some sources are untrustworthy by definition. Some evidence is classified as irrelevant before anyone has looked at it. The dogmatist doesn’t just hold a number they can’t defend, they’ve lost the means to ever revise it. That’s what bankruptcy looks like: dogma writes checks that reality can’t cash.

The dogmatic skeptic goes broke the same way. If nothing short of certainty counts, then no evidence ever counts, and the probability never moves. Same failure, but with the polarity reversed.

(The same failure mode exists for any orthodoxy, by the way. Party platforms, economic schools, and lifestyle tribes all have their off-limits questions.)

Cromwell’s rule

Statisticians already have a name for the cure. If you assign a prior probability of exactly 0 or exactly 1 to something, no quantity of evidence can ever shift your estimate. That’s just the math. Dennis Lindley dubbed the safeguard Cromwell’s rule, after Oliver Cromwell’s 1650 plea to the Church of Scotland:

“I beseech you, in the bowels of Christ, think it possible you may be mistaken.”

Cromwell was no skeptic. He was a devout man asking other devout men to leave the door open a crack. I find that a useful reminder that the rule has nothing to do with which side of any question you’re on.

A field test

Ask: “What would change your mind?”

A person with a live threshold can usually answer, even if the answer is demanding (“a replicated study with a decent sample,” “a documented case I can’t explain away”). A person who answers “nothing,” or whose bar sits at a height no evidence can reach, has told you where they stand. This works on friends, on strangers, and (annoyingly) on yourself. Try it there first.

The improbability is the point

Back on the topic of faith: Miracles are only noteworthy if they shouldn’t happen. That’s what the word means. A miracle is a low-probability event, and its entire significance comes from the odds against it. So an argument that a given miracle is nearly certain (read: scientific apologetics) has talked itself out of the one thing that made it a miracle. And the flat declaration that divine intervention is impossible has rounded the odds to zero, which means it can never be shown otherwise and has quietly discarded the tools for finding out.

Honest faith, it seems to me, is the kind that states its odds. “This is very unlikely, and I’m choosing to bet on it anyway” is a coherent (even beautiful) position, and it leaves everything open to inspection. The same candor applies in the other direction: “I’m very confident there’s nothing out there, but I can’t put the number at zero.” Neither requires abandoning a conviction. Both just require a humble epistemology.

Method outweighs conclusion

If I’m forced to choose, I’d say that promoting bad epistemology does more harm than promoting a wrong conclusion. A flawed conclusion reached by a live process will eventually get corrected, because the process is still running. But a correct conclusion reached by a frozen threshold teaches freezing, and whoever learns it will apply it to every other belief they hold (including the ones that turn out to be wrong).

TAKEAWAY: Round your beliefs if you must (everyone does). But be honest with yourself about what you’re doing, and keep the tools of revision within reach. Ask what would change your mind, and be suspicious of any answer—yours included—that amounts to “nothing.” And for bonus points, experiment with assigning actual numerical probabilities to your convictions and notice how those numbers shift over time.