In March 2026 we asked an Instagram audience to comment one word: YES. Nothing was offered for it. No prize, no reason, no account of what YES agreed to. The same instruction was posted four times over four days. The first three attempts failed. The fourth, with the rule tightened to "YES, once", passed.

The thread that carried the round holds 154,480 comments from 113,180 distinct people, counted when the round closed. 101,709 of those comments are the word YES and nothing else. 46,970 ignored the instruction and wrote anyway. And the single most liked comment in the whole round, ten minutes in, was a refusal.

That is one reel, on one day. The round was five posts: four attempts and the pass, and by the count the platform reports on 31 August 2026 they have drawn 5.2 million views and 494,000 comments. Attempt 1 alone drew 1.67 million views and 198,064 comments, more comments than the thread the pass was declared on. The exact-YES analysis runs on attempt 3 because the archive holds that thread whole (154,480 comments, the same number the ledger filed) and holds only 8,882 of attempt 1. Every post in the round is tabled at the end of the paper.

This paper is about two things the count shows. A crowd will do a small exact thing for no stated reason, but only once the thing has become the obvious move. And the same crowd that complies will put the one person who refused on top.

The posted pass condition was one word, once, from enough people to matter. The hypothesis underneath it was simpler still: will strangers perform the cheapest possible public act, exactly as asked, with nothing offered in return?

• Instruction: "Comment YES. Once." Attempt 4 tightened the rule to YES once, with repeated entries discarded.
• Where: Instagram. Four posts on four consecutive days, 15 to 18 March 2026. The thread that carried the round is the attempt 3 post, which ran to 154,480 comments.
• Who: the ordinary audience of the account plus whoever the platform showed the post to. 113,180 distinct people commented on the carrying thread.
• How we counted: every comment on the thread, replies included (40,411 of the 154,480 are replies). A comment counted as exact if its text was YES and nothing else. Comments that contained YES with other words were counted separately. Everything else counted as ignoring the rule.
• Why one thread carries the exact count: exactness is a reading of each comment's text, and the platform reports totals, not text. Only the comment archive has text, and it holds attempt 3 whole (all 154,480 comments at round time) and slices of the other four posts. Table 1 is the whole thread; Table 3 runs the same rule on every post's slice with the coverage stated.
• Two clocks: the exact-YES count is the carrying thread as it stood when the round closed, which is the count the ledger filed and the pass was declared on. The platform totals in The whole round are from the Instagram API on 31 August 2026 and have kept growing since the round.

The first three attempts did not fail because people would not comment. They failed because people would not comment exactly. Writing YES costs nothing; writing only YES, once, requires reading the rule and accepting that the rule is the point. On the fourth day the rule had become the obvious move, and the crowd made it so by watching itself not make it three times.

The second result is the one the count was not designed to catch. The crowd that complied crowned the refusal, and then, in the replies, told the refuser to delete it. Compliance, applause for disobedience, and a demand to erase it came from the same hands, in the same hour. A comment section is a voting machine as well as a reply box, and the two votes do not have to agree.

The 46,970 off-rule comments matter for a different reason. To the platform they are activity, and activity is what the platform rewards. So the same thread reads as a pass or a failure depending on which number you quote. The record prints both.

• Four attempts is not a clean repetition test. The audience grew between attempts and the fourth post tightened the wording, so repetition and the word "once" cannot be separated in this data.
• The count includes replies. 40,411 of the 154,480 comments are replies to other comments. The exact-YES count is a count of comments, not of people, though 113,180 distinct people is on the record.
• The thread kept growing. The archive now holds 114,069 top-level comments on the carrying reel and the platform reports 186,251 comments in total. The exact-YES share of everything written since the round closed is not in the ledger and this paper does not estimate it.
• We do not know why the refusal was liked. It may have been the joke, the emoji, or the refusal. The count only shows that it was.

Every round after this one narrowed the instruction until exactness itself was the thing being measured. By Round 8, one wrong character on one comment could cost the whole wall. Those rounds are the follow-ups, and their papers sit beside this one.

The round this paper reads is 5 posts, not one. Every post is below with the platform's numbers and what the comment archive holds of each.

The round is filed in the Chronicle as THE YES ROUND, ⟡029. Its law, participation is cheap and precision is dear, is the rule every later round was built to test.

The instruction as it went out: the reel frame beside the caption, captured from the operation's own signed-in account. What the paper counts is the answer to exactly this.

Native captures of the comments this paper leans on, taken from the operation's own signed-in account on 31 Aug 2026. Handles and like counts are as they stood that day; the numbers in the paper are the round-time count.