The seams are the part worth showing

The advice is always the same. Post your journey, show the build, grow an audience, and distribution follows. It is not wrong about the audience. It is wrong about which part is the point, and it was written before the part it recommends became the dangerous part.

Because here is what building in public actually gets sold as: a highlight reel. The shipped feature, the revenue chart pointing the correct direction, the launch that went well, the testimonial with the nice headshot. All the wins, curated, in a row. That version grows an audience, and an audience is a real thing to have. It is also worth roughly what a demo is worth, which the case for an AI operating system already settled: it shows the happy path, everyone can now generate the happy path, and it tells you nothing about what the thing does when a customer does the thing customers do at 11pm on a Saturday.

The seams are the opposite of the reel. The seams are the mechanism showing, the bug you caught, the number that went the wrong way and why, the thing you tried in front of everyone that did not work. And the reason to show them is not that vulnerability is nice. It is that only the seams contain an incident anyone can grade.

That is the whole argument, so it is worth being slow about. Trust does not accrue because you were likeable. It accrues per incident, one survived failure at a time, which is the mechanism: a business hands a system the next irreversible piece only after it watched the last piece fail loud and bounded and get caught, or fail silent and get caught for that. A highlight reel has no incidents in it. It was built specifically to have no incidents in it. So it cannot accrue the thing. The seams can, because a failure that happened in the open and got handled is now a fact about how you behave under load, and a fact about how you behave under load is the one asset a rival cannot lift off your site. They can copy the feature by Thursday. They cannot copy the eighteen months of you not setting anything on fire, and they cannot manufacture the specific Tuesday your thing broke and you bounded it in forty minutes and said so.

Same mechanism runs on the builder, in the other direction. A failure you buried taught nobody, and that includes you, because nobody with a move of their own ever got to grade it. This is the graded-bet logic, the same clock. A buried failure is a verdict you walked away from without a scratch. A public one gets scored, by people who can leave, switch, or reply, and a scored bet is the only kind that improves. So the seams do two jobs with one move. They build the track record and they run the learning loop, for exactly the same reason: they are the only version that gets graded where it counts.

Now the part the advice skips, which is the reason most of it is dangerous when followed whole. Showing a blunder does not raise you in general. It raises you only if you were already, visibly, good.

There is a name and a study for this, and it is not folk wisdom. In 1966 Aronson, Willerman and Floyd ran the pratfall effect. A person spills coffee on himself. If the room already believes he is highly competent, the spill makes him more likeable. If the room thinks he is average, the identical spill makes him less. The blunder is a multiplier, and it multiplies whatever sign was already there. Read across to the builder and buyer, carefully, because this is a 1966 lab study on personal likeability and I am borrowing it to illuminate a trust decision it was not built to prove: seams raise credibility for a builder whose competence is already on the record, and they lower it for one whose is not. A new builder posting mostly failures is not being brave. He is publishing evidence for a case that was already going against him.

Which gives the strategy a built-in bill, and the honest version of this piece has to read the bill out.

One. The copy tax. Your visible roadmap and your visible mechanism are free R&D for a faster builder, and the tools that let you build in a week let him clone what he can see in a weekend. The old indie-hacker instinct was to stop sharing specifics somewhere around twenty or thirty thousand in monthly revenue. That threshold has collapsed toward zero. What used to be safe to show because nobody could act on it fast enough is now the thing somebody ships next Tuesday.

Two. The competence precondition, which is just the pratfall effect handing you a bill. If your competence is not already established, the seams do not read as candour. They read as the reason not to buy.

Three. The performance trap. The instant building in public becomes a growth tactic, the incentive flips: manufacture a photogenic failure, stage the vulnerability, film the pivot with good lighting. That is a highlight reel with a failure stapled to it, and it accrues nothing, because none of the incidents were real and a fake incident cannot be graded.

Four. The exposure cost. Some failures are flatly unrecoverable in public. A security hole on the risky core while it is still open. A live client matter, a legal one. "Show the seams" was never a licence to publish a wound that gets worse the more people see it.

The strong objection, then, built to win, because arguing with the weak one would be a highlight reel of its own. Building in public is a net negative for a serious solo builder now. It is free market research for competitors who clone your roadmap over a weekend. It selects for founders who are good at performing the journey over ones who are good at building, so the loudest wins and the substance loses. It rewards manufactured vulnerability, the staged failure, the photogenic pivot. And the exposure is asymmetric: one real failure shown at the wrong moment ends you, while a hundred wins barely move the needle. The quiet builder who ships a finished, trustworthy thing and lets the work speak beats the noisy one every time.

Granted on three of the four, and the concessions are the argument, not a dodge. The copy tax is real. That is why the rule below draws a hard line at the risky core and the unrecoverable. Performance-over-substance is real too, and it is the reason the mechanism here is the pratfall effect rather than "be vulnerable": staged seams from someone with no visible competence accrue nothing, so the strategy defeats itself in the performer's own hands, and the fix for fake seams is more real competence shown first, not fewer seams. Asymmetric exposure is real. So the rule was never "show everything," and it isn't.

What does not survive is the last claim, that the quiet builder wins. Against a rival who is equally competent, the one with a public record of the core behaving under load takes the trust decision, because the quiet builder is asking the buyer to trust the salesperson, and the open one is showing the record, and those are very different credit ratings. The objection is right that reach is a bad reason and performance is a real risk. It is wrong that silence beats the record. The record is the moat. Silence has nothing to show.

I will name the one I am building, once, and keep it honest. Enthius is my pre-launch venture, an AI operating system for a small business, held to the locked standard and no looser, and I am building it in the open by myself. It has not earned a thing yet. There is no client, no number, no price. I am arguing that the seams accrue trust faster when the failures are visible instead of buried, while standing on a track record of, so far, one contained failure I posted about and a lot of unfinished. That is a builder's wager and I have not won it. I am also, right now, exposed to my own copy tax and my own competence precondition, and I know it. Ask me in eighteen months whether showing the seams paid, because today I am the case, not the proof.

So the useful part, whichever side you are on. Not a test to run, a line to draw, per item, before you post.

Sort what you are about to publish into three buckets. Show the seams of anything already shipped and reversible: how the shipped feature actually works, the bug you caught and how you bounded it, the thing that failed and what it taught, the number that dropped and why. That is track record and it compounds. Withhold the mechanism of the unshipped thing that is your actual edge. Roadmap, fine. The crown-jewel how, no, because that is the one thing the weekend-cloner needs and cannot get from your shipped product. Never publish the live unrecoverable wound: the open security incident on the risky core, the client or legal matter, the failure whose damage grows with every viewer.

The line, in one question you can answer before you hit post: is this failure already contained, and does a competent rival reading it gain anything they could not already get from the shipped product. Contained, and no rival gain, show it. Uncontained, or the unshipped edge, hold it. Openness is per item, not a posture, and it is the seams you can afford to show that build the only thing a rival can't ship next Tuesday.