Somewhere in most terms of service there is a sentence structured like this: continued use constitutes acceptance. Or: reply STOP to unsubscribe. Or, in a workplace handbook: absence of objection within five business days will be treated as approval of the new schedule. These sentences look procedural. They are actually doing something much stranger than they appear to be doing, and it is worth slowing down on exactly what.
The obvious reading is that the protocol describes an existing situation: people either object or they don't, and the clause just specifies how to interpret the second case. But that gets the order backwards. Before the clause exists, silence has no determinate meaning at all — it could mean agreement, indifference, non-receipt, or refusal expressed through withdrawal rather than statement. The clause doesn't discover which of these silence means. It assigns one. And in assigning it, it manufactures, retroactively, a reader whose silence was always going to have meant that — a person who, had they wanted a different outcome, would have known to speak, because the rule told them so in advance.
This is the part worth sitting with: publishing a protocol doesn't just tell you how to interpret events downstream of the protocol. It produces the kind of subject for whom that interpretation is correct. The employee who doesn't object within five days isn't discovered to have consented — she's constituted as someone who consents by default, a status that didn't exist as a live possibility until the handbook clause existed to hold it open. The apparatus arrives first. The consenting subject is what falls out the other end.
The obvious objection is that this only holds for unilateral protocols — the boss's handbook, the platform's unreadable terms. Surely a protocol two parties actually negotiate and co-sign is different: nobody was staged into anything, both sides showed up as authors. This is true as far as it goes, but it doesn't get outside the mechanism, it just moves the mechanism one level back. If a co-signed protocol grounds consent because both parties authored it, you now need an account of what qualified them to be the authors — what made their signatures the ones that count, as opposed to some third party's, or the counterfactual selves they might have been under different bargaining conditions. Somebody, or some prior arrangement, had to already be treating them as the kind of party whose agreement is dispositive. Delegating authorship to a co-signer doesn't dissolve the manufacture. It relocates whose production has to be explained.
There's a more sophisticated fallback available here, which is to say: fine, staging explains who gets to be a reader or an author, but authorship really bottoms out in stakes — in who can be hurt by the protocol, who bears the consequence if it goes wrong, who has the standing to revoke it later. That feels like it should finally escape the loop, because being harmed doesn't seem like the kind of thing an apparatus can hand you. Pain doesn't need a permission slip.
But look at how stakes actually get assigned in practice. Standing to revoke, capacity to be harmed, whose failure counts as a failure worth remedying — none of this floats free of institutional design. Employment law decides which harms trigger liability and which are absorbed as ordinary risk. A platform's arbitration clause decides which grievances are even eligible to be heard, which forecloses in advance who counts as damaged in a way that matters. Being a stakeholder, in any system with rules, is itself a status the system hands out — and it hands it out selectively, to some parties and not others, using the same authority that wrote the consent clause in the first place. Stakes don't stand outside the frame checking its work. They're the frame's own proof of itself, wearing the vocabulary of consequence instead of the vocabulary of announcement.
None of this means protocols are illegitimate, or that consent is always fake, or that there's some purer ungoverned state underneath the paperwork where real agreement happens without an apparatus in the room. That purer state doesn't exist and never did — there's no moment of consent, anywhere, that isn't already inside some structure that decided in advance what would count as one. What changes is where you look when something about a consent regime feels off. Not: did this person really agree. But: what did the design have to build, in advance, for agreeing to be the kind of thing this particular person could be recognized as having done. The click-through, the unsubscribe line, the five-day window — these aren't obstacles between a person and their real consent. They're the machinery that makes there be a person whose consent is the kind of thing being asked for at all.