Notice-and-comment rulemaking

Why must an agency answer objections it is perfectly free to reject before its rule can stand?

Here is the asymmetry worth staring at. An agency must publish what it proposes, must let anyone file an objection, and must address the serious ones before the rule takes effect. And then it may do exactly what it first proposed. No majority of commenters defeats a rule; nobody is counted.

So we have a procedure with no binding force, made a condition of the rule's validity. Either that is ceremony, or the force is doing something other than deciding the outcome. Which?

Dispose of the plebiscite reading first. If comments were votes, agencies would tally them and courts would check the tally; neither happens. Mass campaigns generating identical filings by the hundred thousand rarely move a rule's substance, while a handful of technical objections from affected firms often do — I state that as the direction the empirical work points, not a measured rate. What separates the few comments that matter is not who sent them but what they carried. So ask what the agency actually lacks.

A rule is a prediction: this requirement will cost about that much, will be feasible by that date, will push behaviour one way rather than another. Where do those facts live? Almost entirely inside the firms, hospitals and local authorities that will have to comply. An agency cannot subpoena a forecast. It can guess, or find some way to make the people who know tell it.

Now the awkward question. Why would a regulated party hand over information that helps the agency regulate it? Two reasons, and the second is the machinery. The first is that silence is worse: a rule written on the agency's guesses binds you just as tightly as one written on your figures. The second is that an objection not raised during the comment period is generally treated as forfeited when the rule is later challenged in court. Speak now, in writing, on the record — or lose the point permanently. That converts commenting from an opportunity into something closer to a compelled disclosure.

But extraction alone would let the agency pocket the information and publish whatever it had always intended. Hence the duty to respond. Notice what the duty is and is not: not a duty to agree, only to state a reason. Why should a mere reason matter?

Because a stated reason is a testable object: it can be laid beside the record and checked for whether it engages the objection or evades it. That is what makes review possible without making judges the policymakers. A court asked "is this rule wise?" has no competence and no warrant. A court asked "did the agency confront this evidence and give a reason connected to it?" is asking something it can answer.

So the pieces lock together. Notice must be specific enough to argue with — which is why a final rule too distant from the proposal counts as never noticed and has to go round again. Comment extracts private information under threat of forfeiture. The response duty turns it into a record. Review makes the record consequential, and the prospect of review is what makes every earlier step happen.

There is a commitment effect on top: having published its intended position before deciding, the agency must now argue in public against its own earlier reasoning, whether it changes course or refuses to.

Three refinements, and one neighbour to keep separate.

The nearest neighbour is regulatory-ossification.md, which works through what all this procedure costs — how the stages multiply, and how an agency under that load substitutes into guidance and enforcement. The fixed point of difference is the question asked: that piece asks what the procedure costs, this one asks what it is for. Both are true at once, uncomfortably so, since the very record requirement that extracts information from people who would rather withhold it is the requirement that makes promulgation slow. regulatory-capture.md explains why the informative comments come disproportionately from concentrated interests — with the twist that a comment sits on a public record and must be answered there, which is more exposure than a private meeting gets. persistent-rules.md concerns why a rule survives once made.

The refinement most worth stating is uncomfortable. "Public participation" describes the form, not the function: a person with no information the agency lacks and no legal argument to preserve contributes nothing the machinery is built to use. That is not a criticism of anyone. It follows from what the procedure is doing.

And here is the test. If the mechanism is information-forcing plus record-building, then changes between proposal and final rule should track the content of comments — specific, technical, from parties who will bear the cost — and should be nearly indifferent to their volume. The refuting observation runs both ways. Observe agencies revising in proportion to how many comments arrive, regardless of what they say, and this is a plebiscite after all. Observe courts letting rules stand where a material objection went unanswered on the record, and it is ceremony after all, because nothing then hangs on the answering.

• Why systems that publish rules without any comment stage issue them far faster, and what that speed is bought with.

Every term the collection defines is gathered in the glossary.