Users searching Google News for coverage of geopolitical events this week found something unusual alongside Reuters and The Guardian. A search for news about the Strait of Hormuz returned a Polymarket bet on the exact number of ships allowed to pass through it. Google called it an error and removed the results — but understanding why it happened in the first place reveals something important about the nature of prediction markets.
Table of Contents
Why the algorithm got it wrong — logically
Google News uses automated criteria to identify news sources: publication freshness, topical relevance, update frequency, and structural markers resembling news articles. Polymarket meets almost every one of those criteria. It publishes new markets continuously, ties them to real-world events actively in the news, updates them constantly as odds shift, and writes titles that read like headlines.
A market asking “Will ChatGPT Be the Number One Free App in the US Apple Store by April 10?” is, structurally, almost indistinguishable from a news article tracking the same question — at least from a content indexing system’s perspective. The algorithm was not fooled by something random. It was doing exactly what it was designed to do.
Why Google was right to remove them
The mistake was understandable. The correction was still correct. Journalism and betting are fundamentally different things. Google News is designed to surface reporting — researched, fact-checked, and editorially accountable. Prediction markets reflect where money is flowing, shaped by speculation, positioning, or insider knowledge. Presenting both in the same interface misleads users about the nature of what they are reading.
It is worth noting that Google already partners with both Polymarket and Kalshi for its Finance product — a far more appropriate context, where probability estimates sit alongside stock prices and market data, and users can evaluate them accordingly.
The bigger question
The incident raises an uncomfortable question for the prediction market industry. If these platforms are genuinely information infrastructure — as they have argued with growing success in regulatory and media circles — it is hard to explain why their presence in a news aggregator is an error. If they are something categorically different from journalism, which is the right answer, the industry needs to be more careful about the analogies it invites.
The algorithm made a reasonable mistake. But the fact that it was reasonable should prompt some reflection about what prediction markets actually are — and what role they are playing in the broader information environment.
