Re:Wordle is pretty damn smart in many subtle ways
by samhw on Wednesday February 23, 2022 @11:37AM
(#30442929)
> And then the biggest mistake I see people make is getting a lot of the letters in the right spot early and then just leaving them there on subsequent guesses.
I just heard about this game for the first time from this thread, went to play it, and realised this on row #3 after initially wasting row #2. It made me want to go off and write an algorithm (I actually returned to the thread to see if anyone had written anything along those lines).
But I just realised this won't do, because your score depends not only on the last row but on the intervening ones too, so there's a perverse incentive to repeat the successful letters. That works nicely for non-programmers, but it made me think I'd prefer a purer game where the only 'cost function' is based on the correctness of your final guess. Or at least, that game would be quite fascinating to try and write an optimal algorithm for.
I just heard about this game for the first time from this thread, went to play it, and realised this on row #3 after initially wasting row #2. It made me want to go off and write an algorithm (I actually returned to the thread to see if anyone had written anything along those lines).
But I just realised this won't do, because your score depends not only on the last row but on the intervening ones too, so there's a perverse incentive to repeat the successful letters. That works nicely for non-programmers, but it made me think I'd prefer a purer game where the only 'cost function' is based on the correctness of your final guess. Or at least, that game would be quite fascinating to try and write an optimal algorithm for.