Exploring Wordle
Hercule Poirot makes an astute observation in ABC Murders
“When do you notice a pin least? When it is in a pincushion.”
The story of Wordle is similar — “where do you store all answers? Right on the page — in the source code — in plain sight”.
There are 2309 words listed in the source code as of 12th Feb. When I compared this with another list similarly pulled out from the source code from a few weeks ago by Owen Yon, I noticed that 6 words have been knocked out.
A tiny bit of poking around on Wayback Machine and I can see that this change happened on 1st Feb 2022. I would like to think that this is the NYT editorial team filtering out words originally input by Josh Wardle or It may just be that Mr Wardle himself removed these words. The last one of these deletions is before Wordle no 400 (At the time of publishing, we are at Wordle #239). Around 5 months of Wordles have been screened.
I am pretty confident that we can expect more changes to the source code and words to be edited out soon.
The words which have been removed fall into a few themes –
· Offensive (WENCH)
· Insensitive (SLAVE, LYNCH)
· Americanness (FIBRE — WORDLE uses American spellings. Remember HUMOR?)
· Phonetically or Structurally Similar (AGORA was removed as it was preceded by a word which it resembled)
· Difficult (PUPAL)
I am not sharing the master list or a tutorial on how to see this — I am trying not to be like the guy who told a bunch of primary school kids that Santa is not real.
Having said that, I spent a few hours on Excel, just doing basic cuts on the list, doing my best to not peek at subsequent answers. While the analysis is rudimentary, I do believe it answers some interesting questions.
5 simple observations, which you can put to use when you play Wordle:
1. Over 50% of the words begin with one of the following alphabets: A, B, C, P, S or T. Don’t waste time starting with J or Z
2. If you want to think of letter frequencies, forget EAOIDH NRSTUY (Discussed in Edgar Allan Poe’s Gold Bug) or ETAOIN SHRDLU (Made famous in Sherlock Holmes’ Dancing Men). Remember EOLART SCNFID— the letter frequencies for Wordle.
3. There is a healthy distribution of 2-letter groups (e.g. ER, ES, IF) and doubled letters (e.g. OO, EE). In the past, we have also seen alphabets being repeated 3 times (SISSY).
2-letter groups: the most common one is ER (as seen in SERVE, PAPER) — I found it 189 times in the 2309 Wordles. The next 3 most common pairs are IN, ST and RE. I looked at 3-letter groups too — only one stood out — ING (appears just 31 times).
Doubled letters — we pretty much have 3 winners — LL, EE and OO — eerily appearing 58 or 59 times each 😊
NOTE: I did not interrogate the list looking for pairs — but based it on Herbert Zim’s top 2-letter groups and 3-letter groups from his “Codes and Secret Writing”
- 2 Letter groups: “TH HE AN RE ER IN ON AT ND ST ES EN OF TE ED OR TI HI AS TO”
- 3-Letter groups: “THE ING AND ION ENT FOR TIO ERE HER ATE VER TER THA ATI HAT ERS HIS RES ILL ARE
- Doubled letters: “LL EE SS OO TT FF RR NN PP CC”.
4. For the middle spot (i.e. 3rd alphabet), go with the vowels — it is one of the vowels for over 50% of the cases.
5. The last alphabet — go with EYTRLH in descending order of letter frequencies — these 6 letters make up over two-thirds of the Wordles.
A part of me was not comfortable doing the analysis. It just felt like I was creating cheat codes for a game that is giving people all over the world endless hours of joy. I think that there is something pure, innocent, and even poetic about the game. I was somehow relating how I feel about the game to Archibald MacLeish’s line “A poem should not mean. But be.”
Curiosity has its own reason for existing!
PS: at the time of publishing, I noted that someone else has written about the words removed. I had a different take and hence leaving it as is
Image Credits : Markus Spiske (Pexels) and Andrea Piacquadio (Pexels)
