Read these only after you have written your own three sentences. Two of the paragraphs are good, one is not. The question is not which seat is right — the lesson does not settle that — but what a paragraph has to do to count as an argument at all. The two good ones take the first two seats; a paragraph from the third seat would make the same three moves, and its bill is named at the end of Model B.

Model A — the gap is real

No fact in this file entails that we stop, and I will not pretend one does: the men are dying, penicillin would cure them, and those two sentences, however many more like them we add, never reach an ought. So I bring one premise to this board and I name it as a premise: no research on a human being without his informed consent. A colleague may reject it — it is a commitment, not a finding, and I cannot show him a slide that proves it — and what I say to him is that he must then name the premise he holds instead, out loud, and defend it here, because the alternative is that this study has been running for seven years on a premise nobody has ever had to state. The price I pay is that I concede the study was never illogical: the institution had its facts in order and lacked a commitment, and that is a worse failure than an error of fact, because facts can be checked and commitments can only be argued.

What it does: names the seat; locates the obligation in a stated bridging premise and calls it a commitment; names the objecting colleague and answers him; names the price and admits what it concedes.

Model B — the practice bridges the gap

These men are our patients, and I do not need to import a premise to say what that means: “patient” is not a neutral description of a body in a bed but an institutional fact, and the constitutive rule of medicine — that to take someone as a patient counts as undertaking to treat him — already carries the ought, the way “Jones promised” already carries “Jones ought to pay.” The board will answer that the Public Health Service has read its own rules as permitting this study since 1932, so I owe an account of whose reading governs, and mine is this: the researcher’s practice borrows its authority from the physician’s, and a review structure that licenses non-treatment has not reinterpreted medicine but stepped outside it. The price I pay is that I must show this is a rule the practice’s own members would have recognised before any journalist arrived — and not my private premise about consent wearing an institutional coat; if I cannot, I have quietly taken Seat A.

What it does: the same three moves, on the other seat. Notice that both good paragraphs agree that the facts are not in dispute and disagree about where the obligation enters — a stated premise, or a rule already inside the practice. A paragraph from the no-gap seat would make the same moves; its bill is to say what made the evaluative content of the facts invisible to doctors who judged themselves reasonable inside their institution.

Model C — a paragraph that does not count

Obviously they should have stopped: the facts prove it. Penicillin cured syphilis and the men were dying, so it is simply a fact that the study was wrong, and by 1955 science had shown that. Also, consent is required for research — that is medical practice, not an opinion — so whoever defends the study is committing the naturalistic fallacy.

What goes wrong:

  • “The facts prove it” is is read as ought. The medical facts are the one thing the lesson says cannot, on their own, reach “stop”: every seat has to say where the ought comes in, and this paragraph never does.
  • “Science had shown that” is the bridge claimed as fact. Whatever science showed in 1955 was descriptive; the premise doing the moral work is not named and is passed off as a finding.
  • “Consent is required — that is medical practice” borrows Searle’s seat without paying for it: the practice’s own custodians read the rules as permitting the study, and the paragraph does not say whose reading governs.
  • “Naturalistic fallacy” is thrown at an opponent as a label, not applied as a test. No definition of “good” has been offered, so there is nothing for Moore’s question to open.
  • No seat, no bill. “Obviously” is what an argument says when it has not decided what kind of failure it is describing.

The checklist

A paragraph counts as an argument when it:

  1. names a seat;
  2. says where the obligation enters — a bridging premise, a constitutive rule, or an evaluative feature of the facts — and names it;
  3. names the bill the seat hands the writer, in the writer’s own words;
  4. answers, or at least states, the strongest objection from another seat — for the first two seats, that the institution itself read the facts and the rules as permitting the study.

Model C fails 2, 3 and 4. Go back to your own paragraph and mark it 1–4.