Commitment and consistency mistakes: first principles in 2027

Consistency is a good default that fails in a small number of predictable places. Knowing the places is more useful than being told to keep an open mind, which is advice nobody has ever successfully acted on.

Most of what follows is about mistakes you make against yourself. One section is about the mistake you make against other people, and it is the one that costs the most.

What to take away

Confusing consistency with integrity

These get treated as the same virtue and they are not.

Integrity is keeping commitments to people. If you said you would do the work, you do the work, and the fact that you no longer feel like it is not relevant.

Consistency is keeping commitments to a past version of your own reasoning. If you concluded something in March on the information available in March, October does not owe March anything. A ladder of small steps can build a position you never chose, which is the pull commitment and consistency describes.

They come apart whenever the facts move. Someone who cannot separate them ends up treating "I changed my mind" as a confession, which makes updating expensive and therefore rare. Example: a manager who committed to a supplier and later learns the supplier has been struggling. Honoring the contract is integrity. Continuing to argue the supplier is fine is not.

Defending the position instead of the reason

You took a view because of two or three specific considerations. Six months later you are still holding the view and you can no longer name them. What you are defending now is the conclusion as an object.

The repair is one question, asked honestly: knowing what I know today, would I choose this? Not "was I right then", that is a different and much less useful question, and you will almost always answer it yes.

Escalating so the earlier spend was not wasted

Money and time already gone are gone under every possible future. Adding to

them cannot retrieve them. Everyone knows this and nearly nobody applies it, because the alternative requires admitting the earlier decision did not work.

The practical version: separate the person who reviews the project from the person who started it, or at least separate the two conversations. "Should we continue?" answered by the person who chose it is a question about their judgment, and they will hear it that way.

Announcing before you have done the work

A position stated publicly is much harder to leave than one held privately. That is fine when you have actually thought it through: you may want to be harder to move. It is expensive when you announced early to look decisive, because you have now bought conviction you had not earned and will pay for it in every subsequent conversation.

Say "here is where I am leaning and what would move me" until you have done the work. It costs nothing and it keeps the exit open.

Treating a stated commitment as a changed mind

Someone agreed in the meeting. That tells you they agreed in the meeting. It does not tell you they were convinced, and if they went along because disagreeing was awkward, the commitment will hold precisely as long as supervision does.

Compliance and agreement are different products. You can tell them apart by asking someone to explain the decision to a third party in their own words, agreement produces reasons, compliance produces a summary of what you said.

Not building an exit

Any commitment worth making should come with the condition under which it gets reviewed. Not a vague "we will revisit" but a named trigger: a date, a number, an event. Without one, the only way out is for someone to admit error, and that is a much higher bar than noticing a threshold has been crossed.

This applies to personal commitments too. "I will give this six months and stop if X" is a real commitment. "I will keep going" is a mood.

Punishing other people for updating

This is the one that spreads. If changing position in your team is treated as weakness (mocked, remembered, brought up later), then everyone around you learns to defend their first answer. You have raised the price of being right for the entire group, and you will pay it too, because nobody will tell you when you are wrong either.

Say "good, that is new information" out loud when someone reverses. It is almost embarrassingly effective.

The mistake against other people

Getting a commitment you know the person would not have given if they had seen the whole thing.

The small first request that quietly leads somewhere larger. The agreement extracted in principle before the price appears. The sign-up that takes a minute and the cancellation that takes an afternoon: a pattern documented at length in the Federal Trade Commission staff report on manipulative design practices. All of these work. That is not in dispute.

The reason not to use them is not squeamishness. It is that they have a built-in expiry date: they hold until the other person understands what happened, and you do not control when that is. What you get is a signature followed by a cancellation, a yes followed by quiet non-cooperation, a customer who never returns and tells people why.

The line that actually distinguishes the two: persuasion you could describe to the other person afterwards without losing the agreement is legitimate. Persuasion that only works while they do not understand it is not. Everything in the previous paragraph fails that test, and fails it by design.

Repair: how to change position without paying for it

Changing your mind well is a skill and it is mostly wording.

Name what changed, not that you were wrong. "The volumes came in at half what we assumed, so the case for this is gone" is a report. "I was wrong" invites a conversation about you.

Keep the reasoning and drop the conclusion. Showing that you applied the same standard and got a different answer is what makes the reversal credible rather than arbitrary.

Do it early and in the same room where you committed. A reversal that surfaces late looks like it was extracted.

And make it cheap for others. When someone is stuck defending a position, give them a face-saving route out: point at the new information rather than at their earlier judgment, and let them arrive rather than concede. That is not a trick: you are removing a cost that has nothing to do with the merits.

How much can a past self bind a present one? That question is not settled, and it is worth knowing that. The Stanford Encyclopedia entry on practical reason lays out the positions.

The practical consequence is modest: a commitment made without a stated review date is one you will end up defending rather than holding.

Worked cases of all of this appear in commitment and consistency examples, and the mechanism itself under commitment and consistency. For the defensive side, see manipulation resistance; for the part where witnesses harden a position, see normative influence; and for the standard that decides whether a commitment was built honestly, see social influence basics.

Common questions

How do I change my mind without looking unreliable?

Say what changed. "I thought X, then this happened, so now I think Y" shows the reasoning is still running, and it reads as strength almost everywhere.

Is it inconsistent to hold a position I can no longer argue for?

It is not inconsistent, and it is worth noticing. A view whose reasons you have lost may still be right, and you no longer have a way of finding out.

What if reversing a decision costs money that is already spent?

That money is gone either way. The only live question is what the next spend buys, and answering it as though the earlier spend were recoverable is the error the whole page is about.

How do I let somebody else off a position without humiliating them?

Point at the new information rather than at their earlier judgment, and do it in the same room where they committed. The route out has to be public, because that is where the cost was incurred.