The Chia Gaming Platform

What it is and current status

For roughly the last two years I’ve been working on bringing gaming to Chia. This will have:

• Playing games of skill for real money (specifically XCH)

• Real time play, without waiting for the blockchain to make every move

• Enforcement of the game rules so no cheating

• No casino so no fee paid to a casino. The only extra money you lose out on is transaction fees which are currently near zero.

• Legal in many more jurisdictions than casino gaming. There are meaningful legal distinctions between games of skill and games of chance and games played with and without and intermediary. These are intentional parts of the law and not a loophole.

• Whoever you’re playing against won’t be able to skip out on their debts. Money won or lost is transferred immediately.

Those are, to put it mildly, huge features, and it’s been a big engineering lift to make them possible. Current status is that we’re finishing up debugging the core of it (you can follow ongoing development here) and will soon start building it out into a real application which will be normal software development instead of a science fair project and that is expected to ship in a months timeframe.

At a high level the way it works is that when two people want to play a session they use an offer and acceptance to set up a state channel at the beginning, which takes about a minute for the transaction to go through on chain. (State channels are very similar to payment channels like are used in Bitcoin but can do more things like support gaming, which is a problem because Bitcoin can’t support gaming at all, even not in a state channel.) Then they play over that state channel and when they’re done they close out the session with another transaction which pays out the amount they had left at the end. If there’s a dispute in the middle (which there’s no reason for unless one player tries to misbehave or has a serious technical problem) then whatever games are pending get played out on chain. Poker is a fairly good fit for this because session is many short games instead of one long game.

The restrictions of this medium are that it’s restricted to turn based games with very few moves for exactly two players. To get an idea of why more than two players is a problem see envy-free cake cutting. There needs to be as few turns as possible to make the fallback to playing on chain not excessively slow. What exactly is possible in turn based games is a little hard to convey but there will be a suite of (fun and addictive!) games shipped with it initially which does a good job of showing what things are possible and how to implement them. Randomness can be done using commit and reveal, but supporting card replacement value like happens in Poker is problematic. The vast bulk of the academic work on supporting ‘Mental Poker’ is on handling that one not very important feature. Instead we’ll start with a Poker variant which uses an infinite deck because that makes the problem go away completely. (If you search for ‘commit and reveal’ there are a lot of crypto projects claiming they’re doing something cutting edge and amazing and that it has to be done on chain. That isn’t true. It’s trivial and can be done over state channels.)