Open utterances-bot opened 2 years ago
I think the ideal strategy for the first word is to maximize information. b/c most words have an 'a' in them, knowing that letter is part of the word doesn't reduce the solution space much -- it doesn't give you a lot of information. On the other hand, learning if the word contains 'c' almost evenly divides the set of possible solutions -- maximum information.
Still, getting the most green boxes as you can early on feels good. And I find it easier to reason about possible words given letters that are in the word rather than their absence (see "The dog that didn't bark"). It's an interesting intersection of information theory, information people can easily use, and fun.
Edit: I should have mentioned I came to this post b/c I ran a similar analysis using the same maybe-misguided optimization, also factoring in most popular letter by position. I came up with 'tales' which I still like to use as a first word (followed adding 'cr' to whatever other letters I've learned. e.g. when only 'l' is a hit from 'tales', my second guess is 'clour')
Edit: I should have mentioned I came to this post b/c I ran a similar analysis using the same maybe-misguided optimization, also factoring in most popular letter by position. I came up with 'tales' which I still like to use as a first word (followed adding 'cr' to whatever other letters I've learned. e.g. when only 'l' is a hit from 'tales', my second guess is 'clour'
Yes this honestly is probably much better. Do you have a link to your cost function?
Min-Maxing Your First Wordle Guess
Personal blog and resume for Jan-Hendrik Ewers.
https://janhendrikewers.uk/min-maxing_your_first_wordle_guess.html