Why sacred texts, intelligence tests, and our newest machines all draw their power from seeming unchangeable — and why that is the very thing that hollows them out.
The keyboard you are reading this on is badly designed, and everyone knows it. The QWERTY layout was arranged for the mechanical typewriters of the 1870s, and the reasons it won stopped applying generations ago. Faster arrangements exist. Almost no one switches, because the cost of everyone relearning at once is higher than the cost of typing a little slower forever. So we keep it — a frozen accident, openly arbitrary, harmless.
Notice what QWERTY never does. It never claims it had to be this way. Nobody teaches children that the layout is a law of nature, that the letters fell into their slots by necessity rather than by the jam-prone linkages of a Victorian machine. Its arbitrariness is worn on its face. And precisely because it admits it was chosen, it can sit there indefinitely without rotting. You can replace it the day the switching cost drops, and nothing sacred breaks.
Now consider the forms that cannot afford that honesty — the ones whose whole power depends on not admitting they were chosen. A founding scripture. A measure of human intelligence. The algorithms now deciding what a billion people read and watch. Each of these draws its authority from seeming more than a decision: from appearing discovered rather than decided, inevitable rather than assembled by particular people at a particular time with particular interests. That appearance is not a flaw. It is the engine. A rule that everyone believes was found rather than made can coordinate millions, because there is nothing to argue with. You do not negotiate with a fact of nature.
But the world a form was built for never holds still, and here is the trap — though the trap is narrower and stranger than it first looks. Some forms can change in full view and keep their authority; we will come to them. The ones caught here are those that staked everything on a particular answer and built no legitimate way to revise it. Such a form cannot openly change that answer without dissolving the very thing that made it strong, because there is no sanctified channel through which the change could pass as anything but a confession that the answer was never fixed. The moment people watch the answer edited, outside any process they hold sacred, they see the hand doing the editing, and the spell — this was discovered, not chosen — breaks for good. So these forms face an impossible choice as their world drifts away from them: change openly and lose their authority, or hold still and lose their fit with reality. Mostly they take a third path, which looks like neither and is worse than both. They change in secret.
How a form hollows out
The secret change has a consistent shape. The founding thing — the text, the rule, the measure — is held fixed and declared untouchable. But the interpretation of it, the practice that grows up around it, quietly shifts to keep pace with the moving world. The words stay; what they are taken to mean migrates. For a long time this works beautifully. This is, in fact, how every living tradition adapts: not by rewriting its scripture but by reinterpreting it, not by amending its founding measure but by adjusting how the measure is read. The interpretation is a flexible layer that absorbs the change while the fixed core is pointed to, untouched, as proof that nothing has changed at all.
The problem is the endpoint — but only one kind of interpretation reaches the bad one, and the difference is the whole game. When the reinterpreting is done in the open, by a process that can be argued with and that holds itself accountable to evidence and to its own past rulings, the interpretation thickens: it accumulates real constraints, a denser web of precedent and method that genuinely shapes what comes next. That is not hollowing; that is a tradition staying alive. But when the reinterpreting is hidden — done quietly, to preserve the appearance that nothing changed, accountable to no one and to nothing outside the present moment’s convenience — it hollows. The interpretation becomes a patch layer doing whatever the present wants, while the founding thing is pointed at to bless it.
The hollowing path has a consistent end. As the world keeps moving, the gap between what the founding thing literally says and what people actually do under its name keeps widening. The hidden layer stretches to cover the gap, and stretches, and stretches. Eventually the original constrains the practice in almost nothing. It is still there. It is still revered. But it has stopped doing work — it survives as a token of legitimacy, a thing you point at to justify what you were going to do anyway, rather than a thing that tells you what to do. The form has been hollowed from the inside. The shell is intact and the inside is gone.
And hollow forms have a strange way of dying. They do not die of the gap. They can carry an enormous gap between word and practice for centuries, looking entirely healthy, because the hidden layer keeps absorbing the strain and the token still commands respect. They die when someone exposes them — when a rival authority, with its own interest in the matter, says that was a choice loudly enough to be heard, and the hidden adaptation becomes visible all at once. The hollowing is the slow, invisible vulnerability; the exposure is the fast, public trigger. This is why the fall of a long-standing institution always feels so abrupt, so unforeseen. It looked fine the day before, because the disease lived entirely in the hidden distance between what was said and what was done — and what finally killed it was not that gap but a competitor who profited from pointing at it.
Three examples show the mechanism at three stages: one we can see whole, one we are living inside, and one we are now building at speed.
The mechanism seen whole: a sacred canon
Someone chose which books are in the Bible. This is not a provocation; it is history. For the first few centuries, Christian communities circulated dozens of gospels, letters, and apocalypses with no agreed list. The twenty-seven books of the New Testament were fixed as a set comparatively late — the bishop Athanasius named exactly that list in a letter in the year 367, and regional councils ratified it over the following decades. The choosing was contested, and it was never fully resolved. To this day the Catholic, Protestant, Orthodox, and Ethiopian churches keep different canons; the Ethiopian church preserves a Bible of more than eighty books, including ancient texts like Enoch that the others let go. Those surviving disagreements are the fossil record of the original choice — proof, frozen in the divergence, that a decision was made where the doctrine says only discovery occurred.
Once the list was fixed and declared the unchangeable word, the text could no longer move — but the world did, ceaselessly, for two thousand years. So the adaptation went entirely into interpretation. The same fixed verses were read to permit and then forbid slavery, to subordinate and then elevate women, to condemn and then tolerate lending at interest. The words never changed. What they were taken to mean traveled enormous distances, and the traveling was always described as deeper understanding of an unchanging truth rather than what it plainly also was: a fixed core kept fixed while the practice around it adapted to survive.
And the Christian tradition contains, in the Reformation, the clearest example of what happens when someone finally says the gap has gone too far. The reformers’ charge was essentially that the interpretation had drifted so far from the text that the church’s practice was now determined entirely by accreted human tradition, with scripture reduced to a token invoked to bless it. Their fix — scripture alone, drag the practice back to the founding words, discard the centuries of drifted interpretation — is the act of declaring a form hollow and trying to refill it. It did not heal the church. It split it, permanently, because the only way to expose the hidden gap is to say out loud that authority had been doing something other than what it claimed, and that sentence, once spoken, cannot be unspoken.
The mechanism live: the intelligence test
Every so often, the organizations that produce IQ tests quietly reset them. Raw scores across the population have risen steadily for most of a century — a real and well-documented effect — and because the test is defined so that the average score is one hundred, the scale has to be periodically pulled back down to keep that average where it belongs. Sit with how odd that is. Nobody recalibrates a thermometer to keep reporting the same temperature as the room warms. A measure of a genuinely fixed quantity does not require a committee to re-center it each generation. The recentering is the tell: it is the secret adaptation, the flexible layer doing its work, the scale being adjusted to track a moving population while the claim — this is a fixed, culture-free measure of a fixed, biological thing — is preserved untouched.
Be fair to the science, because the unfairness would be a cheap and different essay. The thing IQ tests measure is real and unusually consistent; the general factor they tap is among the more replicable findings in all of psychology, and it predicts real outcomes. That last fact matters more than it first appears, because it marks a genuine difference from the canon. A sacred text has no check outside itself — nothing in the world can confirm or refute it, so its only validators are internal authority and consistency. The intelligence test does have an external check: its scores correlate with outcomes it did not author, which means reality pushes back on it in a way it never pushed back on scripture. By this essay’s own logic, that is what keeps the construct alive — it stays in contact with a verdict from outside. The trap is not in that grounded core. It is in the social form built on top of it — in the move from this is a useful, particular measure built by particular people for a particular kind of reasoning to this is an objective, timeless reading of human worth. The first is honest and survivable, and it is the part reality keeps honest. The second is the permanence claim, and it is the one that hollows. The grounded measure and the hollow verdict ride on the same test, which is exactly why the hollowing is so hard to see: the part that works vouches for the part that doesn’t.
Watch what the permanence claim makes impossible to ask. A large share of humanity is now spending its formative attention on smartphones and short-form video, plausibly training a different cluster of mental skills than the abstract, decontextualized reasoning the test was built to measure. Is that a decline in intelligence, or a shift to a different kind of it? The test, by its own self-description, cannot tell you — because answering honestly would require admitting that it is a snapshot of one era’s preferred competence rather than a window onto a timeless quantity. A form that has staked its authority on measuring something fixed cannot say the thing I measure may be going out of date. So it reports a decline and calls it a decline, the way QWERTY would report a fast Dvorak typist as a bad QWERTY typist — measuring the new skill with the old instrument and recording the mismatch as deficiency. And the pressure to keep calling the recentered test a fixed natural measure does not come from the evidence. It comes from the institutions that use the number to sort people, who need it to seem like discovered fact rather than a revisable human tool, because a discovered fact is so much harder to argue with.
The mechanism industrialized: the machines
We are now handing machines the job these forms always did — the job of grading, ranking, selecting, deciding what counts. An algorithm chooses what you see; a model scores the essay, screens the résumé, flags the loan. And these systems are presented, almost universally, with the oldest move in the book: as objective. Not as particular tools built by particular people on particular data with particular blind spots, but as neutral readers of how things really are. The output is true because the machine is impartial.
That is the permanence claim, rebuilt at industrial speed. An algorithmic ranking that presents itself as objective is making exactly the gamble the canon made and the intelligence test made — borrowing authority from the appearance of having discovered rather than decided — except now we can manufacture new such forms by the thousand, faster than anyone can check whether they actually track reality. And the hollowing here does not need anything exotic to get started. A machine learning system is trained to maximize a measurable stand-in for the thing you actually want: clicks instead of value, watch-time instead of worth, a test score instead of the ability the test was meant to detect. The stand-in is the hidden interpretive layer, and it drifts from the real goal the moment optimizing it pays better than serving the goal — which is immediately. The system gets relentlessly better at the proxy while the thing the proxy was supposed to represent quietly empties out, all while the form is pointed at as a neutral measure of the real thing. That is hollowing with no secret adaptation required and no fix from cleaner data, because the gap between proxy and goal is built into the act of optimizing a proxy at all.
The exotic failure is the acute version of the same disease. When a model is trained on the world’s text it is grounded, however loosely, in something outside itself; when it is increasingly trained on text that earlier models produced — its own output fed back as input — the grounding thins and quality degrades in a documented way researchers call model collapse. That is the fast death: a form cut off entirely from the reality that kept it honest, hollowing in a few cycles instead of a few centuries. But the ordinary death needs none of that. It needs only a proxy, an incentive, and the word objective. A generation of rankings and recommendations that look authoritative, optimize a stand-in, drift from what actually serves people, and are deferred to anyway — because the algorithm said so is the most efficient legitimacy token ever built. We are mass-producing forms whose entire authority rests on seeming objective, and calling the production progress.
What the survivors did differently
Not every long-lived form hollows. Some have absorbed two thousand years of a moving world without going hollow, and the reason is subtler than refusing to claim permanence. They make permanence claims too — large ones. The difference is where the permanence sits.
Look closely and the survivors are not modest. Common law insists it does not invent but finds — that it is uncovering a law that was always there, an aura of timelessness as strong as any scripture’s. Rabbinic tradition holds that its oral interpretation was given at Sinai alongside the written text, discovered and not decided. Science treats the laws of nature as eternal and out there, waiting to be read. Each carries a full permanence claim. What sets them apart from the canon and the timeless test is that the permanence is attached to the process, not to any particular answer. The eternal thing is the method — case reasoning, structured argument, the standing offer to overturn any claim with better evidence — and the answers it produces are all openly, avowedly revisable. The form says, at once, the way we find truth is sacred and fixed and every specific truth we currently hold is provisional. Revision is not a betrayal of the sacred thing; it is the sacred thing operating.
That coexistence is what the binary misses. A form does not have to choose between an aura of inevitability and the ability to change; it can have both, if the inevitability lives in the revision process and the changes are the process running in the open. This is also why visibility alone does not break the spell. People watch the Supreme Court overturn precedent and the law does not dissolve, because the overturning is the law working; they watch a theory fall and science gains authority rather than losing it, because falsification is the method honoring itself. What breaks a form is not that it changes in view, but that it changes outside the channel it claims is sacred — revision smuggled in where the form has promised none is possible. The canon had no sanctified way to revise its answers, so when its answers had to move, the movement had to be hidden, and the hidden movement is the rot.
This is the counterintuitive heart of the matter. The thing that feels like protecting a cherished form — fixing a particular answer and declaring it beyond revision — is the poison, because the world will move and the answer will have to follow it in the dark. And the thing that feels like surrendering it — building an open, accountable, sanctified way to revise the answers, and admitting that any current answer is provisional — is the cure, and it need not cost the aura of permanence at all, if the aura is relocated onto the process of revision itself. The cure is cheap if you build it in early. It is catastrophic if you wait, because by then the only way to make revision visible is the Reformation’s way: a rival declares the old form hollow, breaks the spell in public, and forks. The survivors built the sanctified channel early. The hollowed forms staked permanence on an answer and kept patching it in secret, because opening a real channel always feels like surrendering the permanence when it is the only thing that preserves it.
What to ask of anything that cannot change
There is a clean way to test which kind of form you are looking at, and a clean prediction it makes. Do not ask whether it claims permanence — the healthy ones do too. Ask where the permanence sits. Does it have an open, accountable, sanctified way to revise its own answers — a channel its believers regard as legitimate, even sacred — or does it stake everything on a particular answer and forbid the channel? A form with the channel can change in broad daylight and keep its authority, because the changing is the thing it reveres. A form without it can only change in the dark, and is hollowing whether or not anyone has noticed yet. It has simply not been exposed — no rival has stood up and said that was a choice where its believers could hear.
The condition that would prove this whole picture wrong is specific: a form with no sanctified channel for revising its answers that nonetheless revised one by open, public decision and kept its aura of inevitability — believers watching the answer change, outside any process they held sacred, and going on treating it as a law of nature. I do not think this happens, because watching a supposed law of nature get edited outside a trusted channel converts it, in the watcher’s mind, into a choice, and that conversion does not reverse. When the Catholic Church openly modernized its practice in the 1960s, it changed answers many believers had been taught were unchangeable, through a process those believers did not experience as the sacred thing revising itself — and it produced a permanent traditionalist fracture among exactly the people for whom the changelessness had been the point. (I am recalling that episode rather than auditing it, but it reads as a form paying the late, catastrophic price for revising an answer it had no legitimate way to revise.)
So notice what we are doing right now. We are building machines whose authority rests entirely on seeming objective, deploying them into every corner of decision-making, and teaching a generation to defer to them as neutral readers of reality — which is to say we are manufacturing, at the largest scale in history, forms that cannot survive admitting they were chosen. The intelligence test took a century to reach its quiet crisis. The canon took two thousand years. The machines will get there faster, because they are cut off sooner from the only thing that keeps any of these forms honest: a steady, external, unfakeable verdict from a reality that does not care what the form claims about itself, delivered slowly, over time, to anything willing to keep listening. The forms that last are the ones that never stop listening for it. The ones that hollow are the ones that decided, at some point, that they already knew.
cafebedouin.org 