Cracking Wordle
Michail Strijov · Last Updated: 20 Feb 2022
“SAUCE”
is the perfect Wordle starting word, as far as I know.
If you’ve ever gone online in the past six months, chances are you’ve heard of Wordle, an online word game as simple and as good as they come. Even Masha Gessen, one of my favourite authors in the past years, has taken to posting her (impressive) Wordle results on Twitter.
While it’s essentially easy to “win” the game simply by looking up the Wordle of the day, one (or at least I) can’t help but wonder if there’s an ideal word to start the game with, especially since Wordle gives you important bits of information:
- All words are five letters long
- There are no inflected forms, and almost no plurals
- Getting as many single letters as possible in the right place with the first word gives you a good shot at solving the Wordle with the six attempts the game gives you
Based on my primitive understanding of statistics and a GitHub repo of the 20,000 most common English words (built from Google’s Trillion Word Corpus), I’ve written some Python code that:
- Ingests the word list
- Creates a new list with only five-letter words (excluding any words ending with an ’s’ since that includes almost all plurals and skews the statistics; this results in a corpus size of 1980 words)
- Slices the words and calculates the probability of a certain letter occuring in a certain position (i. e. the likelihood of the first letter being an A is 6.7%)
- Sums up those probabilities for all five letters of each word in the word list
- Sorts the list by probability sum and prints it in the console
According to this script, something like “Saiey” would be a great starting word for Worlde – if only it were a word.
And the one word with the most likely sum of letters occuring in their respective position would be “Sauce”.
Sweet! 🙂