The price of a free opinion
A CEO is staring at a hard call. Enter a new market or hold. Launch the product now or wait a quarter. Cut the price or defend the margin. She does what any of us would do. She asks around.
Everyone has a view. The board member has a view. The VP who has been pushing this idea for a year has a view. The consultant on retainer has a view, delivered with a slide number attached. Ask ten people and you get ten confident answers, most of them articulate, some of them contradictory, and all of them free.
That is the part worth sitting with. The opinion costs nothing to give and nothing to be wrong in. Nobody loses anything if they are wrong six months later, because nobody is tracking who said what, and even if someone were tracking it, there is no bill that comes due. A survey takes all those free opinions and averages them, and averaging treats them as equals. The person who has spent fifteen years watching this exact market moves the average by the same amount as the person who skimmed a headline on the way to the meeting. Confidence is not evidence. Neither is a job title. A free opinion tells you what someone is willing to say. It does not tell you what they actually believe, and it never tells you how much they believe it.
This is not a new problem, and an entire industry has already run the experiment at scale. Companies that build tools for aggregating belief tried the version with no money on the line and watched what happened. Cultivate Labs sold this kind of system to enterprises and governments for years, and moved away from market mechanics toward scoring rules where nothing changes hands. Metaculus runs on the same logic: forecasters build a reputation score, not a balance. Both are respectable tools, and scored forecasting is genuinely useful for tracking a large number of small, independent probabilities where reputation alone is enough friction to keep people honest.
But watch what a reputation score cannot do. It cannot force anyone to update. If a forecaster made a call in January and the facts underneath it changed in March, nothing happens to that forecaster if they just leave the call sitting there. Their score does not drop for staying quiet. Their reputation is not at risk for indifference, only for being loud and wrong. So the number on the board can go stale and nobody pays for the staleness. It sits there looking authoritative while the world underneath it moves on.
Now hold a different situation next to it. Someone holds an opinion, and holding it costs real money. Not reputation, not a hypothetical score that resets next quarter: money, today, out of an account that is theirs. The moment the facts change and their position no longer makes sense, they feel it, because the cost of being wrong is not deferred to some future evaluation. It is immediate. So they update. The incentive is pointed at the right target, and that is the whole difference. Conviction gets rewarded. Guessing gets expensive. A number built this way does not sit still while the world moves, because the people behind it cannot afford to sit still either.
This is what I built Assay to price. The number a CEO gets from us does not come from a survey of opinions or a leaderboard of forecasters. It comes from the company's own equity holders, the people who already have a stake in how the decision turns out, putting real money behind their own read of it. No one trades more than a paycheck's net pay either way, so holding a view never costs more than that. No trade shows a name on any screen, so the honesty in the number does not depend on anyone being brave in public. And the read is advisory: it informs the call, it does not make the call. The CEO still decides. What she gets is a number that people were willing to pay to stand behind, which is a different thing entirely from a number people were merely willing to say.
Free opinions are not worthless. They are a starting point, a way to hear where the room stands before you decide. But a starting point is not a decision, and a hard call deserves more than the average of what everyone was willing to say for nothing. It deserves a number that someone had a reason to keep watching, because they had something riding on getting it right.
An opinion that costs nothing is worth exactly what you paid for it. A decision worth making is worth pricing properly: with a number someone paid to stand behind, and kept paying to keep standing behind it, for as long as the question stayed open.
Sources
Cultivate Labs, on moving from market mechanics to scoring rules: https://www.cultivatelabs.com/posts/why-we-stopped-supporting-prediction-markets
Prediction Markets Reviews, on Metaculus being reputation based with no financial stakes: https://predictionmarketsreviews.com/reviews/metaculus