Hard Cases is a game, and it is also a research instrument.
Every case in it is real: a genuine dilemma that a court somewhere had to decide. Now, you will decide. When you do so, your answer joins everyone else's. What accumulates is something that does not otherwise exist. This something is a record of how the public — ordinary people and legal professionals alike — judge contested legal questions, on identical facts, across many countries, against what the courts actually held.
That last part is the unusual bit. There is a great deal of research on what people say they value in the abstract. There is very little on what they decide when handed a hard case, or how that compares to what actual courts do.
Work on aligning artificial intelligence with human values keeps running into the same problem: to align a system with human judgment, you need to know what human judgment actually is — not in the abstract, where most people agree, but in the contested middle where real decisions live. Law is the one domain where societies have spent centuries deciding exactly those questions and writing down the answers. Hard Cases is a large, continuously collected record of how the broader public judges those same cases.
Your answers, the day you gave them, and a small amount of context: the country you are playing from, how long you took to decide, and, if you choose to tell us, a few background details such as age range and whether you have legal training. No names, no case-by-case profiles for sale, and nothing that identifies you. Anything published or licensed is aggregate — group-level patterns, never individual records. The Privacy Policy sets out the detail.
The game is free and will stay free. There are no advertisements.
Hard Cases is built and maintained by Anton Moiseienko. Questions, corrections, and requests from researchers: [email protected].
Last updated: 5 September 2026