Прогнозные рынки делают климатические ставки, но прогнозы Hantavirus показывают панику, а не науку

Polymarket's climate predictions land within 6 percentage points of Berkeley Earth's real-time projections — that's tighter than most pundits. But its hantavirus pandemic market cratered from 19% to 5% in two weeks as the initial news cycle faded, revealing what crowds actually track.

Climate Markets: Surprisingly Tight

One Polymarket market gives a 34% chance of 2026 being the hottest year on record, and 60% for second hottest. Kalshi lands at 32% for hottest. Compare that to Zeke Hausfather's projections, which use Copernicus data: 28% chance of hottest, 67% for second. The differences are single digits. Hausfather says these science-based prediction markets are "reasonably in line with experts' estimates." That's not luck — it's because temperature records are updated weekly, giving traders a steady signal.

Disease Bets: Fear Over Fact

Hantavirus on a cruise ship spiked Polymarket's pandemic probability to 19%. Then it sank to 5%. Vaithi Arumugaswami at UCLA says the real chance of a global hantavirus outbreak is "very, very low." The CDC rates the risk to the US public as extremely low; WHO says human-to-human transmission is uncommon. Bill Hanage at Harvard nails it: the market reflects "anxiety surrounding pandemics," not epidemiological models. Prediction markets here become sentiment thermometers — useful for gauging public fear, useless for actual risk assessment.

Quantum Wagers: Long Odds, Real Signals

Polymarket gives a 3% chance of a quantum computer cracking a Bitcoin private key by end of 2026, and 16% by end of 2027. That's not crazy — breakthroughs this year pushed timelines forward. But researchers still need years to assemble a large-enough fault-tolerant machine. The market is pricing in real progress without the hype. That's the sweet spot: slow-moving technical milestones with clear, measurable events.

Richard Borghesi from University of South Florida sums it up: prediction markets are "potentially helpful forecasting supplements" but not "substitutes for models, peer review or expert judgement." When the crowd lacks specialist knowledge, the signal degrades. When the event is tradable and facts update frequently, the crowd can beat experts. Prediction markets will improve as more scientists trade them — not as oracle replacements, but as one more data stream feeding the same question.

Source: How prediction markets could forecast the future of science
Domain: scientificamerican.com