> I'm trying to understand what the criticism is here
You're correct in your understanding of prediction markets with respect to traders using insider information. There are a couple things going on here. One is the subtext from most news media now that Technology Bad. New technologies are treated as guilty until proven innocent, because that is a more engaging narrative for readers. So in this case, those covering this stuff immediately latch onto the rich get richer, insider trading viewpoint, and that gets reported without any analysis of why that might actually be desirable.
Second, prediction markets, in trying to become broadly accessible to "normal" people and desiring liquidity, need a marketing strategy that is understandable. They can't put out a Robin Hanson article as marketing material. So they market by appealing to something people do already understand, which is gambling. The public has this idea now of prediction markets as a way to make money, not as a tool for learning information. So the default perspective on insider trading is now one of unfairness: somebody used their privileged position to make money. The correct perspective is, in fact, that prediction markets are providing users with value by eliciting information from those insiders, information that the public would not otherwise have. The latter perspective is mostly foreign to degenerate gamblers, and the marketing campaigns of Kalshi and Polymarket aren't helping.
> they are just statistical machine outputing whatever they training set as most probable.
How is this sentiment still the top comment on an article about AI on HN in 2026? It's not true with today's models. They undergo vast amounts of reinforcement learning optimizing an objective that is NOT just predict the most likely next token given the training corpus. I would say even without the RL the "predict the next token" objective doesn't preclude thinking and reasoning, but that's a separate discussion. Generative sequence modeling learns to (approximately) model the process that produced the sequence. When you consider that text sequences are produced by human minds, which most would consider to be thinking and reasoning, well...
You're correct in your understanding of prediction markets with respect to traders using insider information. There are a couple things going on here. One is the subtext from most news media now that Technology Bad. New technologies are treated as guilty until proven innocent, because that is a more engaging narrative for readers. So in this case, those covering this stuff immediately latch onto the rich get richer, insider trading viewpoint, and that gets reported without any analysis of why that might actually be desirable.
Second, prediction markets, in trying to become broadly accessible to "normal" people and desiring liquidity, need a marketing strategy that is understandable. They can't put out a Robin Hanson article as marketing material. So they market by appealing to something people do already understand, which is gambling. The public has this idea now of prediction markets as a way to make money, not as a tool for learning information. So the default perspective on insider trading is now one of unfairness: somebody used their privileged position to make money. The correct perspective is, in fact, that prediction markets are providing users with value by eliciting information from those insiders, information that the public would not otherwise have. The latter perspective is mostly foreign to degenerate gamblers, and the marketing campaigns of Kalshi and Polymarket aren't helping.