The office is small. Nobody in it has received an instruction. A university’s legal counsel, its head of student conduct, and someone from communications are poring over a draft policy. It treats any criticism of a foreign government as potentially subject to discipline. No court has ruled on the wording. No parliament has passed a bill requiring it. What does exist is a deadline — new anti-racism requirements taking effect from January 2027 — along with a report card, published by an office with no power to fail anyone, that nonetheless determines which universities look compliant and which look exposed.
Everyone in the room understands the stakes without anyone having stated them. A poor score costs nothing in law and everything in perception — a line in next year’s coverage, a question at Senate estimates, a donor’s quiet withdrawal. So the room drafts policy for a reader who has not yet been identified, against a judgement that has not yet been rendered, on the assumption that both are coming.
This is Australia, mid-2026. The office in question belongs to a special envoy: a three-year appointment, created by the executive without a vote, whose formal remit is simply to advise. The special envoy can’t fine a university. She can’t cancel a research grant. She can’t compel testimony, issue a visa refusal, or direct a broadcaster’s editorial line. Every substantive power touching funding, migration, or media standards belongs, on paper, to someone else — a minister, a regulator, a court. And yet the compliance apparatus assembled around her recommendations behaves, in every observable respect, as though her office held the power itself. Committees redraft policy in anticipation of judgements that have not yet been made. Institutions perform compliance with a standard that doesn’t yet exist, against a body that can’t enforce it.
None of this requires anyone to be threatened. It requires only that enough people find a threat plausible, and act to pre-empt a version of it that lives entirely inside their own risk assessment. The office issues a report card; it doesn’t issue a verdict. The distinction collapses in practice, because a report card read by journalists, by rival vice-chancellors, by a minister looking for a reason to intervene later, behaves exactly like a verdict for anyone whose funding or reputation depends on how it gets read. What can’t be enforced by the office of the special envoy gets enforced instead by everyone downstream of it, each protecting themselves against an audience that hasn’t spoken yet either.
Three years earlier, in a different setting, an entirely different apparatus went the other way. A proposal to give Indigenous Australians a constitutionally protected advisory body — one permitted only to make representations to parliament and the executive, nothing more, no veto, no binding force — was put to a national referendum and rejected by a clear majority. The mechanism that would have carried the least formal power of any body in the Constitution needed the highest bar Australian law provides: a nationwide vote, a majority of states, years of public argument, a result of sixty per cent against. The mechanism that carries no constitutional status at all needed none of that. One appointment took an afternoon.
The asymmetry is not an accident of politics. A body written into the Constitution can be sued, cited in court, tested against every future government that would rather it didn’t exist — so its birth has to survive the electorate before it can survive the lawyers. A body that only advises can’t be sued for advising badly. Nobody can take a special envoy to the High Court over a report card she has no power to enforce. That legal weightlessness is exactly what let her exist without a vote, and exactly what lets her recommendations move through universities with a force her office never possessed on paper. The apparatus that can be challenged needs consent to survive. The apparatus that can’t be challenged never has to ask.
Right and wrong are separate questions from the one under scrutiny here: which forms of power require the public’s consent to exist, and which do not. A body seeking permanent, entrenched standing had to win an argument with the entire electorate. A body with no standing at all, built to expire in three years, had only to be signed into being by a minister who liked the recommendation. The vote is the expensive road. The memo is free.
Free, but not without cost — the cost simply moves to where it can’t be itemised. Because no standard has been published, nobody drafting against it knows where it actually ends. A published law has a final clause; you can read to the bottom of it and stop. An anticipated law has no bottom, because the anticipation is being generated by people trying to imagine the harshest plausible reading a critic, a journalist, or a funding body might produce. Legal counsel doesn’t draft to the middle of that range. He drafts to the edge of it, because the cost of guessing too permissive is a headline and the cost of guessing too restrictive is nothing anyone will ever notice — a restriction that silences something has no natural way of announcing what it silenced. What emerges is a standard stricter than any rule actually under discussion, manufactured collectively by people who have never met, each responding to an estimate of what the others might do.
There’s also nowhere to take a complaint. A law can be challenged in a courtroom, with rules of evidence and a losing side. Anticipation has no forum. You cannot cross-examine a worry, subpoena a reputational risk, or appeal a decision that a communications officer made in her head about how a story might read in six months. The people whose speech, or research, or funding gets shaped by this process rarely learn that a process occurred at all — the paper trail runs through internal memos and diary meetings, not through anything a citizen has standing to request.
Nor does the fear fall evenly. A well-endowed university can afford to wait for an actual finding before it moves, and often has the relationships to help shape what that finding eventually says. A university running on discretionary international enrolment, or a research centre one grant renewal from closing, can’t afford to wait for anything — it complies with the imagined maximum immediately, because it has no reserve to absorb a bad guess. The same holds inside these institutions. Tenured professors debate the policy in seminar rooms. Casual staff and doctoral candidates on visa sponsorship simply pause. They stop raising the topic in class, because they carry the least capacity to survive being wrong about how much latitude they actually have.
And once a defensive posture is built, it tends to outlast whatever provoked it. The envoy’s appointment runs three years; the redrafted policy, once adopted, carries no expiry clause at all. Dismantling it later would require someone inside the university to stand up and argue, publicly, that the danger the policy was written against was never real — a far harder case to make than the one that built the policy in the first place, since the absence of a prosecution proves nothing about whether the caution was justified. Fear that produces no visible harm looks, from outside, indistinguishable from fear that was warranted. The policy becomes a permanent fixture. Nobody remembers to ask whether it’s still needed.
Nothing about this is unique to Canberra. A central bank’s forward guidance moves currency markets and mortgage rates without a single binding order — the words alone do the work, because everyone downstream has learned to act on the words before the mechanism that would enforce them ever engages. A sovereign credit downgrade reshapes a government’s borrowing costs, its political room to move, sometimes its cabinet, issued by an agency with no jurisdiction, no army, no seat in anyone’s parliament. In each case the formal chain of command stays intact and almost irrelevant. The real instrument is anticipation driven largely by apprehension: institutions and markets moving to avoid a consequence before the body that could impose it has uttered a single word.
Each of these systems strengthens the same way. A market that moves ahead of a rate decision, and turns out to have moved correctly, treats that as proof the signal was reliable — which makes the next signal move markets faster, before it has even finished being spoken. A university that avoids controversy and suffers no funding cut treats the avoidance as vindication, not as evidence that the danger might have been smaller than assumed. The anticipation is never falsified, because acting on it removes the only conditions under which it could have been tested. What looks like prudence from inside the room is a system quietly training itself never to find out.
Perhaps this is why law, when it finally arrives, always looks louder than it is. Law has to justify itself in public — in a chamber, on a record, against an opposition that gets to argue back. Advice needs none of that. It circulates quietly, attaches itself to funding and reputation, and produces its effects through people who were never ordered to do anything, only invited to imagine what might happen if they did not.
You have almost certainly done a version of this yourself — softened a sentence in an email before sending it, not because a rule forbade the original version but because you pictured, for half a second, how it might be forwarded. Multiply that half-second by every drafting session in an office you’ll never see, and you have the entire apparatus this essay describes. It’s not new. What has changed is only its scale, and how many rooms it now runs through at once.
Back in the small room, the redraft continues. Nobody there is breaking a law. Nobody there has been told what the law will eventually say. They are writing to a standard that exists, for now, only as the expectation that somebody, someday, might ask why they didn’t.