As more readers have made WordleBot part of their day, we’ve updated it with new features and improvements. If you haven’t used the bot before, it analyzes how skillful and lucky you are in solving each Wordle. Many of our updates are based on your suggestions, so thanks for sharing your comments. Here’s what’s new:
See more people’s guesses, in more detail
Part of the fun of WordleBot is seeing how other players solve the puzzle, and now you’ll be able to see even more of that. The bot’s charts will show you more guesses from other players each round and display how common each guess was. You’ll also see more guesses from players who faced the same situation as you each round, and we’re analyzing a larger sample of Wordles for each situation, too.
View and compare your past WordleBot scores
One of the most common reader requests is to view the bot’s analysis for past Wordles. Now, for up to the past 90 days, you’ll be able to review each of your past WordleBot scores, and you can ask the bot to analyze past Wordles it hadn’t analyzed before. (The bot is always learning, so your score may be slightly different than when you first saw it.)
For the bot to view your full Wordle history, you’ll need a New York Times account that’s linked to your Wordle statistics; if you haven’t done that yet, link it now to start recording your games. If multiple people play Wordle or use WordleBot on a single Times account, the bot will list only the most recently played or analyzed game for each previous day.
WordleBot also now calculates a rolling 14-day average of your recent scores, so you can see how your luck, skill and steps have compared with other Times readers and the bot itself. If you have a hunch you’ve done better than other players and even the bot lately, now you’ll have the proof.
A new top starting word in hard mode (just barely): SLATE
For players who compete in hard mode, the bot now ever so slightly prefers SLATE as its most efficient starting word, replacing LEAST. The new version of the bot analyzes word frequency over a longer period and has a more fine-tuned understanding of obscure words or unlikely ones like proper nouns.
LEAST is still an outstanding opening guess in hard mode, and the difference between the best opening guess and second best is essentially negligible. (This is also true of the top openers in standard mode.) The bot’s preferred starting word may also continue to change as it keeps learning.
The screenshot upload option, where available, now recognizes letters more accurately, especially in dark mode.
WordleBot will continue to use the approach to solving Wordle that was introduced in August — in which it doesn’t know the solution words — and you can read more about that here.
Common reader questions about the bot
Where can I find Wordle and WordleBot?
You can find Wordle on the web or in the “Games” tab of the Times app. You can find WordleBot here, on the Upshot’s home page or on the menu of Wordle itself.
I selected an English word and an accepted Wordle guess, so why is the bot telling me it’s probably not a solution?
There are almost 15,000 five-letter words in the Wordle dictionary. Some of these — SYBBE, YCOND, IMMEW — are rather obscure. Others could be considered obscene. Still others are just in bad taste.
For its suggestions to readers as part of its analysis, WordleBot considers only roughly 4,500 relatively common words. This, of course, is a judgment call, and many words are open to interpretation. As a guideline, we’ve aimed to include as many words as possible that we think a normal Times reader might have encountered before. It sounds straightforward, but words like ADOBO, BRONC, ALEPH, PISTE, ASPER and DUOMO might seem common to some readers and utterly foreign to others. (And, important note: Neither the bot nor its minders have a say in any word that appears as a Wordle solution.)
We then assign each of roughly 3,150 words a probability of being a solution, based in part on word frequency in New York Times archives dating to 2000 and on what the bot has observed about the words that have been solutions so far.
Wordle solutions don’t repeat. Does the bot take that into account?
No, it does not. And yes, this is a disadvantage for the bot against any people who keep track of which words have already appeared as Wordle solutions.
The bot will say something like, “Here’s what 1,000 other readers did who faced the same scenario as you.” What does that mean?
The “scenario” is defined by which solutions are remaining and what turn the player is on, rather than by which particular path the player took to get there.
For example, for Wordle No. 417, after playing CRANE and then LEGIT, there would have been only one solution left: CLING. The 1,000 readers who faced the same scenario would be those who’d narrowed it to that one remaining solution after their second turn, no matter how they got there.
When calculating the average NYT result for WordleBot users, what score do you give to those who didn’t solve the puzzle in six turns?
The bot uses the expected steps needed for WordleBot to solve the puzzle, based on the solutions left after a reader fails. If a reader eliminated every solution but one, the score would be 7. If more than one solution remained after six tries, the score would be somewhat higher.
I play in hard mode, but WordleBot is scoring me as if I were playing in standard mode. How do I fix this?
For the bot to analyze your Wordle applying the rules of hard mode, go to the settings on Wordle itself and enable hard mode.
Alternatively, you can upload a screenshot of your result and tell the bot — after you’re prompted — that you played in hard mode.
Can you explain what “expected solutions after guess” is, as opposed to actual solutions left? What are you averaging over?
Suppose you start by guessing SLATE. If the solution that day is STEAL, you’ll get one green “S” and the rest of your letters will be yellow, leaving only one possible solution: STEAL.
If, on the other hand, the solution is MOOCH, you’ll get five gray squares, leaving 305 possible solutions remaining.
WordleBot makes this calculation for each of the words it uses in its analysis, averaging the number of possible solutions remaining to come up with a number for the “expected solutions after guess.”
Is the bot dumber now that it doesn’t know the solution list?
Only slightly. It solves Wordles in 3.5 turns on average (instead of 3.4) in standard mode, and in 3.6 turns in hard mode (instead of 3.5). So far, it has solved all past Wordles in standard mode, and it has failed a few times in hard mode, and we expect it to fail several more times in hard mode.
I have another question
You can leave questions in the comments section. We’ll respond to as many as we can, and you can see feedback and advice from other readers.