Can you come up with a strategy?


I made this game as a tool for AI research, and am happy to have received the following review from an online forum.

dcsan 14 hours ago  from what i see you have to guess a word from the list given yellow squares are how many tries you have

as you put a word in, it will get added to the list of guesses

the closer words are at the bottom (thanks @ukv) so the goal is to extrapolate, eg is it an animal name, or an object? based on the order of your previous tries.

the font-size thing is a complete misnomer as its not really perceptible. but the insert location for guesses is more useful (but not documented)

there may be sort algorithm strategies for playing this, eg try two most distant words first... but the nature of the multi-dimensional word-distance being flattened to a single dimension makes it really hard to guess, imho.

when you get it right it deletes all word context and history.

interesting idea but a front-end dev could help make it playable!

If you are a front-end dev wanting to extend this game for a larger audience, it is now possible to purchase the source code for $10.

In the mean time, I am curious what the strike rate of an expert player could be, how it might compare with GPT-4, and if humans can outperform by discovering a strategy. Try it out if you are a genius streamer, and send me a link!