A curator in a regional art gallery is completing the wall text for a group show. She has already drafted the sentence that would place the work inside its actual argument rather than beside it. It’s the one connecting three canvases to a history the artists themselves have spoken about openly. She reads it back silently and then deletes it. Nobody asked her to do that. Nobody is likely to ever know of the existence of that sentence. She has simply understood subliminally, across a decade of applications, panels, and discreet corrections from those whose approval she needs, which sentences are safe to submit and which ones cost the next grant.
Multiply that lone curator by several thousand, spread her across galleries, universities, broadcasters, and funding bodies, and a logic larger than her private caution begins to take shape. No policy sits behind it. There’s no directive from above. Just a pattern, built one small accommodation at a time, until it acts like an intention nobody had really held.
Henry Ergas made a version of this argument recently about the Australian art world, describing what he referred to as artivism: the drift of cultural institutions towards a single acceptable political posture, achieved not through any stated rule but through those who get appointed to panels, whose applications succeed, and which curators are trusted with the larger budgets. His target was narrower than the pattern itself.
Artists have always brought political conviction to their work; that was never the complaint. The criticism, once you follow it past the specific grievance, is about the machinery beneath the complaint: an institution that exists to protect free expression can end up rewarding one form of it and slowly starving the rest, without anyone inside the institution experiencing themselves as an enforcer of anything whatsoever.
Ergas’s own instrument deserves a moment under the same light. A columnist diagnosing an entire cultural sector as captured by one political posture is not exempt from writing inside a moral vocabulary of his own, one in which suspicion of institutional zeal has become almost as reflexive, in some newsrooms, as the zeal it targets. His diagnosis valid. But it’s also partial: one side of a channel rather than a survey of the riverbed.
This is one of the least understood properties of any sufficiently complex system. It can produce collective behaviour that no individual within it intended or would recognise if described to them plainly. Give recognition to people who use a certain lexicon. Seat them on the panels that really matter. Let their networks recruit the next round of staff... Done patiently enough, the institution starts to reproduce itself along a single axis, the way water finds the same channel through soft ground long after the flood that first carved it has receded.
The people inside the process are very often sincere. That’s what makes the pattern so difficult to detect from within it. They believe they’re defending justice, or protecting the vulnerable, or simply funding the better work. In my personal experience they are rarely insincere. What they have lost, usually without noticing that loss, is the capacity to imagine that a colleague who disagrees with them might be reasoning in good faith rather than failing a moral test.
Once an institution has absorbed enough of its own moral vocabulary, disagreement stops registering as divergence. It reads only as a symptom, evidence of something wrong with the person rather than any claim that it might be true or false. That’s the precise point at which a pluralist institution stops being one, while continuing to describe itself, in good faith, exactly as it always has.
This same collapse can be seen far outside any single gallery or country. Wherever a small number of gatekeepers control access to funding, publication, or professional advancement, the same channel tends to cut itself, whatever language the gatekeepers happen to speak. It might include a state broadcaster that’s learned which stories bring discord. A donor-funded network of civil society groups that’s realised which grievances the foundations reward. Or a publishing house that’s fathomed which manuscripts a distributor will actually support. The lexicon differs by region and by regime. The mechanism, essentially a market for legitimacy narrowing itself around a small set of acceptable positions without anyone ever declaring it closed, does not.
The argument since October 2023 over Israel and Gaza has become one of the clearest instances of what happens once that narrowing sets in. Criticism of an Israeli government’s conduct is not, on its own, evidence of hostility towards Jews. Support for Palestinian self-determination is not, on its own, evidence of hostility towards Jews either. A liberal culture has to hold both of those positions as legitimate while remaining capable of noticing when criticism curdles into contempt for an entire people. That boundary asks for judgement, and judgement is precisely what an institution loses once it has decided in advance which side of an argument counts as respectable.
Under enough pressure, individuals stop being read as individuals. A Jewish colleague becomes a proxy for a government she never voted for. A Palestinian becomes a symbol of resistance whether or not resistance was ever his subject. A critic of either position becomes either a bigot or a coward depending on the audience. Ideology needs this simplification, because a category is so much easier to judge than a person. Once the category has replaced the person, exclusion barely needs any kind of justification.
Institution, capture, pluralism, liberal: all useful words, but categories nonetheless. And the same simplifying instinct that turns a colleague into a proxy or a curator into a case study can turn an entire society’s argument with itself into a diagram with two sides. Somewhere between a deleted sentence in a regional gallery and a war fought partly in the language of genocide and self-determination sits an enormous range of ordinary, contradictory, half-formed human motives with which no diagram can possibly survive contact.
Something in this pattern is now moving faster and further than any single institution could manage on its own. The large language models trained on the accumulated output of universities, newsrooms, and publishing houses inherit whatever homogeneity those institutions have already settled into. What used to spread at the pace of a career, through hiring committees, prize juries, and the occasional withdrawn invitation, can now be encoded into a system consulted by hundreds of millions of people a day, each one receiving an answer that sounds neutral because it has been trained to sound that way. The old architecture of cultural influence moved slowly enough to be noticed and debated. The new one answers before an argument can start.
What makes the pattern so difficult to resist is that it rarely needs to be enforced twice. The first refusal, the first quiet correction from a respected colleague, is usually enough. After that, the curator edits herself. The academic softens the paper before a reviewer ever sees it. The producer drops the segment before a commissioner has to say no. Power that has to keep intervening is power that’s still visible, and visible power invites resistance. Power that has moved inside the people it governs no longer needs to intervene at all. It is invisible.
A society can absorb a great deal of foolishness and survive it. What it struggles to survive is losing the ability to tell the difference between someone who is wrong and someone who is dangerous. Once that distinction goes, an institution built to protect thought has become one more mechanism for deciding, in advance, what may be thought at all, and the people running it will very likely be the last to know, since by then the correction happens before it ever reaches their desk.