Crossword Help Forum
Forum Rules

SOARE (a baby eagle) gets in the three top vowels and two of the three most common consonants. If this only produces one on two letters, I try TUNIC - these ten letters are in the twelve most common. The problems come when you end up with *OUGH, or like today, S*ILL, and try to devise words that will identify the one letter of five+ alternatives .....

I always use TEARS as my first word and something like PIOUS as a follow-up, depending on the results of the first guess. I enjoy it and it provides 5 minutes of fun - various family members and friends compete.

Indeed, oldpoliticsstudent. It depends on your word list what the three top vowels are. Many frequency distributions favour I over O, which might give the edge to RAISE. This was analysed a while ago by computer, using the actual words used by the game (a US list, hence the FAVOR row). It marginally preferred ROATE, and gave an average result of c 3.5 guesses. More impressively, the worst result was 5 with RAISE.

I see Susie Dent says her average is about 3. I’d happily have a daily wager with her, 3 or less she wins, 4 or more she loses, minimum 100 trials 😉 Or is she more accurate than a computer simulation, albeit with limitations, which has the word list of possible answers? I’d like to know.

For those who are interested:

https://medium.com/@tglaiel/the-mathematically-optimal-first-guess-in-wordle-cbcb03c19b0a

Thanks to all who replied. To my shame I discovered this morning that ARENT was rejected by Wordle. I should have tested before posting.

Lot's of helpful suggestions above for which I am grateful.

It's like a lottery really.
I just use a random word each time. Success with two 2/6 but mostly 4/6 and I admit having a 6/6 (received a 'phew' message for that) I always use 3 consonants and random vowels

If you have the word list (or list of all allowable 5 letter words) a computer can be better than any human since they can use a search tree which eliminates as many words as possible each guess. It's like solving "Guess Who" but with extra info about letters which are correct but in wrong place, but the search space is much larger so building the tree might take a while.

Or you could train a neural net, but with those you don't really know what is happening "under the hood" as it were.

ParadigmS.....
OR get a life ! haha

Oof!

I'm kind of interested in this stuff since I work as a games programmer and have dabbled in AI stuff. I also know about stats and game theory from my maths degree :p

It's no coincidence that the best scrabble players tend to be mathematicians as well ;)

Some guy won the French scrabble championships (several times I think) despite not speaking French :)

PS You know I was joking !🤡