The Argument for Prediction: Why Fans Started Keeping Score

Fandom has always involved prediction. "They'll collapse in the second innings." "This one's going to flop." "He's finishing the season as top scorer." These sentences have been said in living rooms and group chats forever.

What they had in common was that nobody wrote them down. The confident wrong call evaporated by Monday; the lucky right one got retold for a decade. Fan prediction was a performance, not a record.

The shift worth talking about is small and consequential: the calls started being recorded. Not monetised, necessarily — recorded. Timestamped, priced, and settled against an outcome. That changes the activity more than it sounds like it should.

What a scoreboard does to an opinion

An unrecorded opinion is free. You can hold it as strongly as you like, and revise it silently afterwards, and remember yourself as having been broadly right. Hindsight bias is not a character flaw; it's the default setting of human memory.

A recorded opinion behaves differently. You have to commit to how confident you are, in advance. And when it resolves, the record disagrees with your memory if your memory has been generous.

Almost everyone discovers the same thing when they first do this: they are substantially less accurate than they believed, and the gap is concentrated in the things they care most about. That discovery is uncomfortable and genuinely valuable — it is the beginning of calibration, which is the only forecasting skill that transfers to anything else.

Why a market, rather than a poll

Once you decide to record predictions, you need a way to combine them, and a market is a better aggregator than a poll for one reason: it makes people put something behind the answer.

In a poll, saying "definitely" is free, so a lot of people say it. In a market, taking a position at 95 costs you nearly everything you'd gain if you're right and hurts badly if you're wrong. That asymmetry filters out cheap talk. People with genuine conviction commit more; people who are performing certainty commit less.

The result is a number that tends to beat individual experts — not because crowds are wise in general, but because a market weights each opinion by how much its holder is willing to stand behind it, and errors in different directions cancel.

It is worth being precise about the limits, though. Markets aggregate what participants know. They do not manufacture knowledge. A market where nobody has looked at the screen count is a market with no information about the screen count, however confident its price looks. Thin markets on obscure questions are opinion polls with extra steps.

What it does to watching

The most common thing people report after their first month is that they watch differently — and more closely.

A match where you've made a call on the first-innings total is a match where the powerplay matters to you even if you don't support either side. A release weekend where you've priced an opening is a weekend where you notice the showtime density at your local cinema.

This is the honest appeal of the format, and it's got nothing to do with what you win. It converts passive consumption into an argument you've made and can be checked on. The stakes are attention, not money.

The parts that don't work

Three honest limitations, because an article that only argues one way isn't worth reading.

Prediction can crowd out appreciation. If every film becomes a number and every match becomes a position, something is lost. The people who get the most out of this treat it as an occasional layer over things they'd watch anyway, not as a lens over everything.

Short windows are mostly noise. A week is not enough time for skill to show. Any weekly ranking is substantially a variance contest, and mistaking a good week for evidence of skill is the most common self-deception in the format. The honest measure is a season.

Fandom markets are biased in a specific direction. The crowd overprices what it loves. That's an opportunity for a disciplined player, but it also means the aggregate price on an emotionally-charged question is a worse estimate than the same crowd would produce about something they didn't care about.

Where this goes

The interesting frontier isn't more markets on more things. It's better resolution — questions written precisely enough that reasonable people agree on the answer, settled against sources anyone can check, with the reasoning published rather than announced.

That's an unglamorous engineering problem, and it's the one that decides whether a prediction platform is a community or a slot machine with commentary. A platform that writes sloppy questions and settles them opaquely will produce a leaderboard that measures nothing. One that writes precise questions and shows its work produces a record that actually means something about the people on it.

Fans were always going to keep making calls. The only question was whether anyone would keep score honestly.