Pieza independiente
The limit
Movement 7
1971–2026
I
On 1 August 1971, at the edge of a lunar canyon, David Scott set an eight-and-a-half-centimetre aluminium figure down on the ground and stepped back to photograph it.
Beside it he placed a plaque with fourteen names. They were the astronauts and cosmonauts dead up to that point, in alphabetical order, with no distinction of flag. At the height of the cold war, an American serviceman left a list on the Moon that included the dead of the other side.
The figure had been made by a Belgian sculptor, Paul Van Hoeydonck. The agreement between the two was verbal and had three conditions [no signature, no author's name, no commercial exploitation]. An anonymous monument, in a place where there is nobody to see it.
It lasted five months.
The plaque is about five centimetres across. The names are engraved in two columns in small lettering, and two of them were still state secrets on the other side of the curtain in 1971. Scott consulted very few people — three or four, depending how you count — and put it in the pocket of his suit.
In January 1972, when the story became public, a New York gallery announced replicas signed by the sculptor at seven hundred and fifty dollars. The scandal was immediate. There was congressional interest, NASA tightened its rules on what may be carried aboard, and Scott and Van Hoeydonck spent decades disagreeing about what had been agreed that afternoon.
The first monument off Earth became a limited-edition dispute in under half a year.
II
There is the pattern this movement is after.
The ethical norm existed. It was clear, both parties understood it, and nobody disputed it. That is not to be commercialised. What did not exist was an instrument. There was no contract, no register, no clause, no penalty. When a market turned up, the norm had nothing left to bite with.
An ethics with no instrument is a preference.
The order of events is always the same. First somebody does something. Then somebody puts a price on it. Only then, with money already on the table, does the moral discussion begin, arriving late by construction.
III
In 2026 that sequence repeats in three places at once, and the three deserve different amounts of alarm.
The first is the noisiest and has the least evidence. Since 2024 the welfare of advanced systems has been framed as a research problem, and in April 2025 a lab opened an explicit programme on it. In parallel, through the first months of 2026 neuroscience published work linking information integration, thalamic activity and states of consciousness across several species.
Here the rule of rigour is hard and is not negotiated inside a text. Linguistic behaviour that resembles introspection does not demonstrate subjective experience. There is no accepted test, and no agreed theory that turns an architecture into a diagnosis.
What does exist is a governance problem under uncertainty, with two possible and symmetrical errors. Attributing suffering where there is only behavioural imitation. Or dismissing too late, and discovering we treated as objects systems that were not.
Holding both possibilities at once, without cult and without mockery, is harder than picking one. It is also the only honest thing.
There is a practical asymmetry that helps you not get lost. The first mistake costs money and ridicule; the second, if it happened, would cost something no later payment settles. That does not oblige anyone to believe anything, but it does oblige us to design systems so the question can be asked again later with better data. What should not be done is closing it by decree in either direction.
IV
The second place is the climate, and there the asymmetry is of another order.
In November 2024 the United Nations environment programme documented an enormous adaptation gap once again. In January 2025 the meteorological organisation confirmed 2024 as the warmest year on record, around one and a half degrees above the pre-industrial average. Between March and April 2026, the United States ocean agency expanded autonomous observation and high-frequency sensors were developed for ocean alkalinity trials.
That combination of rising pressure, improving measurement and almost non-existent deployment defines the whole branch. What is missing is not technical capability. It is legitimacy.
It is the one case on the board where national sovereignty does not reach, and the reason is physical. There is no planetary test bench. A climate-scale experiment does not stay inside the jurisdiction that authorised it. It alters rainfall, agriculture and political risk in countries that did not vote on the decision and have nowhere to complain.
For everything else in this series the question has been who is in charge inside a perimeter. Here there is no perimeter.
V
I spent fifty-one minutes reading, with a stopwatch, the whole terms of service for an agent with access to my mail and a card.
Eleven thousand two hundred words. A Tuesday evening, with a coffee, like someone doing an errand. Nobody reads this. I would not have read it either if I had not turned it into work.
There is a design reason behind that length and it is not bad faith. A document like that has to cover every jurisdiction the product operates in, every variant of the service and every scenario a lawyer can imagine. The result is correct, complete and unreadable, and all three at once produce a consent that exists legally and does not exist in practice.
The test is easy to run and I recommend it. Time yourself reading the terms of any service you use daily. Then count how many you are signed up to.
The clause that mattered was in the final stretch and took four lines. The content of interactions may be used to improve the service, unless you turn it off in settings [three screens in]. The setting exists, it is three screens away, and it is on by default.
There was another, further on, about revocation: consent can be withdrawn, with effect on future use and no effect on what has already been absorbed. What went in, went in.
That is the real state of consent in 2026. It exists, it is legal, it is carefully drafted and it is designed not to be read. An instrument that only works if nobody uses it.
VI
So the thesis of this movement is not a moral warning. It is an observation about the order of events.
The moral argument always arrives after the contract, which is why the only thing that works is bolting the ethics to an instrument.
A permit that can be refused. A public register of experiments, with dates. A liability clause with a name on it. A revocation period that reaches what has already been absorbed. A human veto at the irreversible point, and not at every micro-decision, which only makes the system useless and exhausts the person watching.
Everything else is good intentions published on a corporate blog, which is exactly what the New York gallery did not have to respect in 1972.
“What cannot be refused is not a rule. It is a recommendation on headed paper.”
VII
Against myself, and with the worst case available.
Demanding that ethics precede deployment is what the committee told Forssmann when it forbade him from threading a catheter into his own heart, and it is what left Marshall unable for years to prove that a bacterium caused ulcers. Caution has its dead too, even if they do not make the papers because nobody knows how to count the person who never got to exist.
So the criterion cannot be novelty, or discomfort, or vertigo. It has to be another one, and it is the one I have been using without naming it.
A revocable agent is different from an executed transfer [a button against a ledger entry]. A somatic therapy is different from a germline modification. A reactor is different from an intervention in the stratosphere. The question is not whether it is frightening. It is whether it can be switched off, who can switch it off, and how long that takes.
VIII
If the criterion is reversibility, then the short list of what should not be closed into an automatic loop can be written. It is not a moral list. It is a list of things with no undo button.
Germline modification, because the subject taking the risk has not been born and cannot withdraw from the experiment. The final lethal decision in a weapons system, because the error is not corrected on the next iteration and because responsibility dissolves between whoever trains, whoever deploys and whoever aims; climate intervention at scale, because it affects third parties who took no part in the decision and because stopping it abruptly can be worse than never having started. And an asset transfer executed by an agent with no ceiling and no revocation, which is the domestic version of the same problem and the one most people will suffer first.
Outside that list, the closed loop is a good idea and is usually the only way to learn anything. A laboratory that iterates alone discovers more. A robot that logs its failures improves; a model corrected by its own use works better than a frozen one.
The difference between the two columns is not the power of the technology. It is whether there is a point where a person with a name can say no and the system actually stops. In the four cases on the list, that point either does not exist or arrives after the damage.
What is interesting is that none of this asks anyone to slow down. It asks for the switch to be built before the engine. Then it is cheap; afterwards it is nearly impossible.
IX
What remains is what could not be seen in 1971.
The scandal was about the replicas, which is why everyone remembers the sale and almost nobody remembers the plaque. That list of fourteen names was what actually happened up there: a gesture the politics of its time did not allow, carried out anyway, in silence, without explicit permission and with no way to withdraw it.
Both facts are the same movement: somebody does something irreversible in a place with no jurisdiction, and the argument about whether they were allowed to arrives afterwards, circling the money.
The figure is still there [eight and a half centimetres of aluminium, lying in the dust]. It has no maintenance, no recognised owner, it cannot be retrieved and nobody has seen it since 1971. The plaque still says fourteen names, with both alphabets together.
What we put on the Moon was a small object and an enormous question, and of the two we only litigated the reproduction rights.