A daily game about what the world reads next.
Every day, millions of people look things up on Wikipedia. What they read is a live signal of collective attention: a film opening, a match, a discovery, an anniversary. Fomorrow turns that signal into a small daily puzzle. Can you tell which way attention will move tomorrow?
How questions are made
Questions are generated automatically from real pageview data, with no human picking winners and no AI writing the prompts.
- An approved catalog. Questions draw from a curated list of 270 English Wikipedia articles across nine categories: space, nature, gaming, music, film, sport, technology, science and geography.
- Complete data only. An article qualifies only with seven complete days of data and an average of at least 1,000 views a day. If no fair question can be built, the slot stays empty rather than being filled with made-up data.
- Fair comparisons. Trend battles compare articles in one category whose audiences are within 4:1 of each other; head-to-heads use pairs within 1.5:1.
- Always fresh. An article used in any question is not used again for 14 days, and the same combination does not return for 60 days.
- Frozen and auditable. Once published, a question’s choices and baseline never change, and a settled result is never rewritten.
Question text only states observed numbers. Fomorrow never claims why an article spiked.
The data
Pageview counts come from the Wikimedia Analytics API, published under CC0. Fomorrow uses human (non-automated) daily views of English Wikipedia across all access methods. Wikimedia usually publishes a day’s data within hours, but it can take longer; Fomorrow waits rather than guessing.
Who runs it
Fomorrow is an independent project by The Fomorrow team. It is not affiliated with or endorsed by the Wikimedia Foundation. It is free to play and supported by ads. Read the privacy policy and terms of use.
Ideas, bugs or a topic you would like to see? Write to wnstjd4504@gmail.com.