There’s no disputing that the Wordle craze has taken social media by storm, but the best way to play — and the ultimate word to start with — is a subject of contention.
Looking for a scientific answer, US-based software developer Chris Wanek analysed the frequency of each letter among the 2,315 words the game can use as solutions.
‘As a computer scientist, I want some proof — what is the actual optimal starting word?’ he asked in reply to linguist Danny Hieber’s suggestion that ‘irate’ is the best.
‘After looking through the source file, I managed to find the entire word list.
‘I made a quick script to load all of the words, calculate the frequencies of each letter, score each word based on the frequencies of the letters in the word, and then calculate the most optimal starting word, he explained.
From the solution list, the best starting word appeared to be ‘later’, closely followed by ‘alter’, ‘alert’, ‘arose’ and ‘irate’, he revealed in a TikTok video on Thursday.
Appropriately enough for a deceptively simple game like Wordle, however, getting at the final answer proved to be more complicated than Mr Wanek first suspected.
And running through a series of alternative approaches yielded a veritable gamut of other candidates for best opener — with the coder proposing ‘soare’ as the best bet.
Looking for the scientifically best starting guess for Wordle, US-based software developer Chris Wanek analysed the frequency of each letter in the puzzle game’s pool of 2,315 possible solutions. Of these, the optimum starting word appears to be ‘later’ (pictured), closely followed by ‘alter’, ‘alert’, ‘arose’ and ‘irate’, he revealed in a TikTok video on Thursday
‘As a computer scientist, I want some proof — what is the actual optimal starting word?’ Mr Wanek (left) asked in reply to the suggestion by linguist Danny Hieber (right) that ‘irate’ is the best word to guess first in the popular puzzle game
‘After looking through the source file, I managed to find the entire word list,’ Mr Wanek explained. ‘I made a quick script to load all of the words, calculate the frequencies of each letter, score each word based on the frequencies of the letters in the word, and then calculate the most optimal starting word.’ Pictured: some of the words Wordle uses as solutions
The drawback of the original frequency-based calculation — as commenters on TikTok were keen to point out — is that Wordle accepts far more words as guesses than it uses as solutions, which are deliberately confined to less obscure words.
Based on this feedback, Mr Wanek revised his analysis of letter frequencies to consider all of the 12,972 words you might possibly start a game with.
Instead of ‘later’, this calculation suggested that the equal best guesses are ‘aeros’ and ‘soare’ — whereas the worst first guess you could run with is ‘xylyl’.
(‘Aeros’ is the plural of an informal term for aircraft, while ‘soare’ is a young hawk and ‘xylyl’ is a term from organic chemistry for a univalent radical derived from xylene.)
Mr Wanek has also presented three other analyses, two looking at the position, rather than the frequency, of the letters, but using different approaches (sum versus product) — and another looking at their ‘entropy’.
(In information theory, entropy is the average level of uncertainty inherent in a variable — or, in this case, Wordle letter, given all of its possible outcomes.)
Both calculations based on letter positioning put ‘saree’ (a variant spelling of ‘Sari’, an Indian garment ) and ‘sooey’ (a cry used to attract pigs) as the best guesses.
The entropy calculation, meanwhile, joined the letter frequency approach in putting forward ‘soare’ as the optimum starter.
According to Mr Wanek , looking at entropy provides ‘the actual best scoring algorithm’ — thereby making ‘soare’ the best bet. ‘Soare also seems like a very good candidate because it shows up highly in all three lists, and has five different letters,’ he concluded
So — at the end of all that — which is actually the best choice if you want to up your Wordle game?
According to Mr Wanek, looking at entropy provides ‘the actual best scoring algorithm’ — thereby making ‘soare’ the best bet.
‘Soare also seems like a very good candidate because it shows up highly in all three lists, and has five different letters,’ he concluded.
***
Read more at DailyMail.co.uk