Every durable company begins with a disagreement. Not a disagreement about execution, which is cheap, but a disagreement about reality. The founder sees something the market has priced wrong, and the mispricing is the entire opportunity. If the truth were already legible, the returns would already be arbitraged away. You are not paid for being correct. You are paid for being correct before the consensus catches up.
The failure mode is obvious once you name it. Founders optimize for ideas that sound defensible in a pitch meeting, which means ideas the room already half-believes. This is the opposite of the job. An idea that a sophisticated investor immediately understands and approves of is an idea whose premium has already been competed away. The best founders sound slightly wrong at first. They are defending a position the room has not yet arrived at, and the room resents the work of catching up. That resentment is the signal.
Getting this right looks like a founder who can articulate, in one sentence, a specific thing the world believes that is false. Not a trend. Not a vibe. A falsifiable claim. "Enterprises will never run their critical workloads on someone else's servers" was such a claim in 2006. "Nobody will get in a stranger's car" was such a claim in 2010. The claim has to be specific enough to be wrong, and the founder has to have a reason, grounded in something they have seen directly, for why the consensus is about to break.
Getting this wrong looks like a deck full of TAM slides. If your argument is that the market is large and growing, you are not starting from a mispriced truth. You are starting from a legible one. Large legible markets are crowded with competent operators. You will not win by being slightly better at something obvious.
Legibility is collapsing faster than ever. A thesis that takes six months to write up and circulate is already priced in by the time the memo lands. Models are reading the same research, summarizing the same podcasts, and pattern-matching to the same precedents. The half-life of a public insight is shorter than the fundraising cycle. This means the only mispricings left are the ones that require proprietary exposure: a founder who spent five years inside an industry, saw a workflow no outsider can see, and now has an unfair view of what will break. AI compresses the value of general intelligence and raises the value of specific, lived knowledge. The edge is no longer in synthesizing information. The edge is in having information that has not yet been synthesized.
The sharpest version of this argument is that the valuable truths in any era are not hidden, they are unfashionable. Most of what looks like a secret is just something the consensus has decided is embarrassing, boring, or already settled. The work is not discovery in the romantic sense. It is the willingness to take seriously an idea that intelligent people have collectively agreed not to take seriously, and to keep taking it seriously after they have told you why you are wrong. This belongs under the first rule because conviction without this quality is just stubbornness attached to a fashionable opinion. The founders who build anything that matters are the ones who can distinguish between a contrarian pose and an actual disagreement with the world, and who can sit inside that disagreement long enough for the evidence to catch up with them.
Signing this rule means you commit to backing disagreement before it is socially safe, and to defending it when the room has not yet caught up.