If you aren't using this word to start your Antiwordle game, you're not getting your best score.
In case you missed it, a while back (when Wordle was rising in popularity), I made a bot that can consistently solve Wordle puzzles in 3-4 guesses. Although I thought that would be the end of my word puzzle solving programs, I recently discovered the Antiwordle game where the goal is to use as many words as possible while playing on hard mode. Hard mode means you have to include any correct letters on subsequent guesses, and you can’t include any letters that you already ruled out.
I had already created a hard mode for my wordle bot, so it could technically play this new Antiwordle game without breaking the rules. However, it would get the word as fast as possible, which was not my goal. Therefore, I needed to make a few changes to get this bot to play Antiwordle well.
The first change I made was in selecting a word to play. Previously, the bot went through a list of possible words, and figured out every possible response the game could give it. It would then pick the word where the possible words for each response were as spread out as possible, meaning it would (on average) eliminate the most words.
I decided to modify the program to do exactly the opposite of that - it would go though the list of words, and find the word to play that would eliminate the least number of words, making it more likely to include as many words as possible in the result. The system is not perfect, if the word happens to line up with the answer it might be eliminating a lot of possibilities, but it’s the best I could come up with.
So, what word did it pick as the best starting word? It highlighted a lot of double-letter words, including civic
, vivid
, and mimic
. However, after running it for a while (my code was not very fast), it ended up picking the word immix
. Now, you might be thinking, “immix? That’s not a word”. Surprisingly, it is, it means to “mix or blend” and is a valid choice for Wordle and Antiwordle.
My score, without the bot, was a measly 6 guesses for today’s Antiwordle, but the bot was able to milk the system and get a total of 9 guesses before guessing the correct word. Confident of my bot’s ability, I’ll call it good for today and see how it does in the next few days.
Also, if you missed it, here is a link to my post about the best Wordle and Absurdle starting words. If the NYT ends up taking down Antiwordle, here’s a link to an archived version.