I'm working on a distributed multi-user puzzle solving game.
The puzzles are generated from a random seed and left for users to solve.
Once a puzzle, or more usually a sequence of puzzles are published they are assigned the ownership of the publisher and can be bought or sold as produce, digital craftmanship or whatever.
Puzzle engine will be running on a public site, with all the old data available for browsing. All solutions and ownerships, however, will be stored on the block chain.
If a user finds a more optimal solution for a puzzle, that can then be published alongside the previous solutions, but only better solutions than pre-existing are allowed in.
Solutions will thus form a tree, where they are interlinked. Each solution will produce another random seed to a new problem.
I expect there to be one or several metrics for the quality or score of the solution, simplifying trading of these puzzles as commodities or even currency on a market - manual, automated or hybrid. These metrics will be mathematically defined and not dependent on subjectivity, votes or similar.
There is no ban on using computers for the games, basically because it would be impossible to uphold. It is up the the author to make something sufficiently difficult to involve humans, computer-AI or a freestyle combination of the two. Games that are obviously to simple will be sorted out by a flooding of that specific market via bots.
It would also be possible to lock a certain seed for some time, thus enabling work offline. As long as your seed is locked you will not get back with your solution(s) finding that someone has already made a better one. After the publishing, that specific seed becomes available for others to improve on but your original work remains.
There are two reasons this need ethereum, one is for the trading of the produce and the other is for verifying that each solution fits into the previous.
For the test/demo (see below) I expect the payload storage requirement of a single solution to be about 192 bits or 24 bytes, (8 bytes for the seed and 16 bytes for the solution). The original seed is derived from a nothing-up-our-sleeves function, such as pi, e or sqrt(2). Verification of a certain solution in this demo puzzle inside ethereum would involve some hashing and simple geometrical tracking and score counting.
I don't want to set the puzzles in stone at this stage, in fact I expect there to be a plethora of different games for different skills and interests, but here is a demo link of what it might look like:http://brage.ath.cx/fold