I fail to see how Stockfish is being childish here. Sounds like ChessBase was given ample opportunity to comply (release the actual source) and tried to weasel out of it (by releasing a modified copy of their new engine's source which hides the secret sauce). Even so, Stockfish only tried to revoke the license when they came to find out ChessBase had ignored the GPL requirements of Stockfish with not one, but two of the engines they sell.
How far should you let someone fuck around with your license terms before you decide to exercise the nuclear option?
How far should you let someone fuck around with your license terms before you decide to exercise the nuclear option?
Game theory (specifically the iterated prisoners' dilemma) says: once. Many results from psychology - especially parenting - support the same conclusion. Both punishment/retribution for bad behavior and reward for improving behavior should be swift and unambiguous. It's ChessBase's turn to cooperate and show some good faith. Words alone won't do it.
This isnt at all related to rhe original topic but if you like game theory, you'll love this paper[1]. (Tl;dr tit-for-tat isn't the optimal strategy after all).
Also, in repeated prisoner's dilemma with random errors, it has been long known that tit-for-tat isn't the best strategy either.
Comments
I fail to see how Stockfish is being childish here. Sounds like ChessBase was given ample opportunity to comply (release the actual source) and tried to weasel out of it (by releasing a modified copy of their new engine's source which hides the secret sauce). Even so, Stockfish only tried to revoke the license when they came to find out ChessBase had ignored the GPL requirements of Stockfish with not one, but two of the engines they sell.
How far should you let someone fuck around with your license terms before you decide to exercise the nuclear option?
Game theory (specifically the iterated prisoners' dilemma) says: once. Many results from psychology - especially parenting - support the same conclusion. Both punishment/retribution for bad behavior and reward for improving behavior should be swift and unambiguous. It's ChessBase's turn to cooperate and show some good faith. Words alone won't do it.
This isnt at all related to rhe original topic but if you like game theory, you'll love this paper[1]. (Tl;dr tit-for-tat isn't the optimal strategy after all).
Also, in repeated prisoner's dilemma with random errors, it has been long known that tit-for-tat isn't the best strategy either.
[1]: http://www.pnas.org/content/109/26/10409.full