I believe in markets, the freerer the better. But I do not think that any free exchange of goods is necessarily good.
Stewart Brand and Vitalik Buterin have prediction markets down as better than regular markets. Prediction markets reward being right, they are bounded between 1 and 0, reducing mania likelihood, and they are better than the most apt comparison, social media. Those are worthwhile points, but miss the most salient question about prediction markets, which goes to form. The most important question is whether prediction markets should be compared to gambling or market makers.
Gambling is a big part of a small, trivial thing. Market-makers are a small part of a big, useful thing. For example, credit card networks takes about 15 basis points on trillions of dollars of real world economic transactions. In steep contrast, casinos rake in billions of dollars facilitating card shuffling and ball rolling.
Gambling drives non-economic activity, zero-sum outcomes, and reckless risk-taking. When you walk into a casino, you walk into a depressing world with an iron law: the collective always walks out of a casino with less money. In a casino, nothing matters in the real world. Whether the dice lands on red or black affects the prediction payoff, but is utterly meaningless in the real world. In exchange for the degradation of the social and cultural fabric, we may pick up a few points of economic growth.
On the other hand, market-makers facilitate economic activity, positive sum outcomes, and productive risk taking. Market-makers have usually been built on the foundation of real economic activity, and function as a lubricant in a larger system. For example, America harvests 16 billion bushels of corn per year. CME, the dominant market maker in corn futures, trades about 638 million bushels of corn every year.
Some parts of prediction markets offer compelling promise. In politics, the rewards of winning office will usually be much higher than predicting the winner. If the reward for predicting the next president ever got to be larger than the reward for becoming president, then prediction markets would lose their predictive power. This is why a sense of proportion really matters for figuring out if prediction markets are good or not. Back to the casino. The real world implication, absent the bets riding on either side, of a card coming up aces or not are completely trivial.
If the payoff for a prediction can be larger than the payoff for the underlying real world action, then prediction markets function as another gambling den.