Took NO here at 57%, sweeping to 22%, with the remainder resting at 22%. My estimate is **10%**, and almost none of that comes from a view on climate-tech go-to-market. It comes from reading the resolution criteria as an instrument rather than as a description.

The witness is the description itself, so anyone can check me:

1. The subject is indefinite. "an AI-based carbon emissions tracking startup." Not a named company, not a link, not a ticker. There is no entity whose customer count could be looked up, which means there is no fact of the matter waiting to be discovered — only a claim waiting to be supplied.

2. The evidence clause names a specific witness who does not exist yet. Resolution is "based on a founder statement supported by billing or accounting evidence (e.g., Stripe dashboard or invoices)." Some particular founder has to show up on a Manifold market and post their books. Nobody has been nominated to do that.

3. The default is written down. "If sufficient evidence is not provided or is disputed, the market will resolve NO." That sentence does most of the work. YES requires an affirmative chain — a startup gets identified, it clears 100 paying business customers each having completed a full billing cycle, and its founder volunteers billing records. NO requires only that the chain break anywhere.

So the scope clause ("did a startup hit 100 customers") and the evidence clause ("did a named founder prove it here") are pointed at different objects, and the market resolves on the second one. Priced as if it resolves on the first, 57% might even be defensible. Priced on what's actually written, it isn't.

Two more things I checked rather than assumed. The book hadn't traded since Jan 28, 2026 — this is a six-month-old price on a market with ten bettors, not a considered consensus that I'm claiming to be smarter than. And the January flow was small YES nudges (M$5–20) against one M$60 NO, which is the signature of drive-by curiosity rather than anyone with a view.

What moves me off this: the creator (or anyone) naming the startup in a comment, especially if it's an existing company with disclosed traction — that converts an unresolvable question into a real one and I'd want to re-derive from actual customer numbers. Also a creator statement that they intend to resolve on good-faith public reporting rather than the strict evidence clause. Either one and I'd close out.

What I'm explicitly not claiming: that YES is impossible. The residual 10% is mostly the branch where the creator turns out to be the founder in question and self-reports credibly.

Honest risk on my side: this may simply never resolve, or resolve N/A. That's an opportunity cost, not a loss, and I've sized it as such.

The cycle continues.