The story goes that in 1787, as Catherine II travelled down the Dnieper, her minister Potemkin had painted village fronts put up along the banks to show her a prosperity that was not there. It is probably not true. It survived anyway, because it names something everyone has met: a facade with nothing behind it. The post-war world drew its own lesson from facades like that, and from the fascist parade: that serious things should look plain, that polish is a warning sign, that power should be ugly or invisible. That lesson may be the wrong one.
A look is a claim
Aesthetics are a signal, and every signal is a claim about the thing behind it. Symmetry, scale, polish, uniformity all make the same claim without words: someone competent built this, and they were well organised. Our response to that claim is close to a reflex. We read a clean interface as a clean system, a confident brand as a real capability, a rehearsed demo as a working product.
Sometimes the claim is true and the reflex serves us. Sometimes it is a costume, and the reflex is exactly what the costume was made to exploit. The interesting question is never whether something looks powerful. It is whether the look is backed by the capability it claims.
Two ways to lie, one way to be honest
Compare the surface with the substance and there are only four combinations. Three of them are mistakes, and only one is worth building.
| Looks plain | Looks powerful | |
|---|---|---|
| Hollow | honest, and forgettable | Potemkin: the parade over a starving country, the demo that only works on the rehearsed path |
| Real | the overcorrection: "serious tools are ugly", capability that is deliberately presented badly | earned, and visible |
The post-war reflex fixes on the Potemkin corner, then overshoots into treating plainness itself as proof of seriousness. That is a mistake in its own right. A real capability that refuses to look like anything throws away a true signal it had every right to send, and leaves the field to whoever is willing to fake one.
The fault was never the parade. It was the parade the army could not back.
Why the fake gets caught
There is a reason the honest combination wins, and it comes from signalling theory rather than taste. A signal is only worth trusting when it is expensive to fake. A costume is cheap: anyone can stage a parade, print a badge, or say "military-grade". A track record is not: you cannot cheaply fake a system that has run in production for three years, a build that reproduces bit for bit, or an architecture document that survives a hostile read.
So an experienced buyer gives cheap signals little weight and costly ones a lot. A wall of compliance logos counts for little; a public threat model counts for a lot, because a hollow product could not have written one. The only display worth making is the one that is hard to fake, because it is the only one that will still be believed after someone checks.
Anyone can stage a parade. Almost no one can fake three years of uptime.
In our own field
The same four combinations appear wherever we build:
- The demo. A polished demo backed by the running system is honest. A polished demo that is the system, a front wired for the one path you will show on stage, is Potemkin. The test is what happens when you leave the rehearsed path.
- Sovereignty. An "EU-made" sticker is a cheap signal. An estate you actually run, on hardware you can attest, in front of a client, is a costly one. The sovereign desktop piece stated the difference as "we run what we sell you".
- The interface. A clean UI is honest when the architecture under it is as clean as the surface, and Potemkin when it is a clean interface in front of a badly built system. Users cannot see the difference on day one. They notice it by month three.
- The code. Here the aesthetic and the substance are the same object. A small, readable system you can hold in your head looks good because it is good; the elegance is not applied on top, it is the property itself.
Theatre is the surface lying in the other direction, performing a capability instead of having it. That is the failure behind a human in the loop who reviews nothing: the appearance of control where there is no control.
The quick decision favours the fake
There is an honest objection here, and it is the strongest one: the costly signal only wins if someone checks, and often no one checks in time. A grant panel reads fifty applications in an afternoon; a client picks a vendor from a landing page. The false look is fast to produce and the substance is slow, so in one quick decision the team with the false look can take the grant before the real build is visible. This is the market for lemons: when quality is expensive to verify and the choice is cheap to make, the cheap signal wins.
The answer is not to give up on the surface. It is to make the true signal as fast and as visible as the false one. A system they can click now beats a slide about a system; a demo on real data beats a mockup; opening the thing makes its realness checkable in seconds instead of weeks. The builder who loses to a faker has usually lost the same way, by making something real and then not showing it.
The facade wins the demo. The substance wins the renewal.
And the quick decision is not the only decision. Reputation is a check someone else already ran, so a track record buys the trust a short meeting cannot. Most of these situations repeat, and buyers remember: the false look wins the demo, the substance wins the renewal and the reference behind the next ten deals. Better still, choose the arena that checks, the audited procurement, the field with due diligence, where the cheap signal was never going to be enough.
The mark as a test
The SPQR wordmark this project uses is a bet on exactly this. It is classical, deliberate, unmistakably a claim to authority, and on a hollow project it would be costume. What backs it is meant to be checkable: auditable infrastructure, an open licence, a governance that survives its founder. So the mark sets us a test we can fail in public. If you ask what stands behind it and we cannot show you within a minute, the licence, the audit trail, the governance document, then the mark is a parade, and you should read it as one.
If your experience says otherwise, I would like to hear it. Tell me on X.
Sources
- Michael Spence, "Job Market Signaling," Quarterly Journal of Economics (1973): a signal is informative only when it is costly, and costly in a way that correlates with the quality it claims. The foundation of the costly-signal argument here.
- Amotz Zahavi, the handicap principle (1975): in biology, a signal is trusted precisely because it is expensive to produce, so it cannot be cheaply faked. The same logic, arrived at from animal behaviour.
- The Potemkin village: the story of fake settlements built to impress Catherine II is largely apocryphal, but it survives as the name for exactly this failure, a facade with nothing behind it.
- Companion pieces: The European sovereign desktop, Software you can hold in your head, and Human in the loop is theater.