Wordle strategy
Wordle is a game that combines the word-guessing of crosswords, anagramming of Scrabble, and positional guesswork of Mastermind. The goal is to guess a five-letter word, but the only clue you're given is after you enter a guess (which HAS to be a real word, no "AEIOU"): the game will highlight in green a letter that's in the right place, in yellow a letter that's in the word but not in the right place, and in gray a letter that's not in the word.
You get six tries.
So, how do you make each try count?
To enter all the right letters in all the right spots, it helps to know what letters are in the answer, and what spots they go in. If you're a computer, you can tackle these questions analytically, by measuring the relative likelihood of every letter in every possible word, guessing the word with the most likely letters in it, and winnowing that down as you learn what letters are impossible, confirmed, or locked. I built a bot that does this, and it wins more than 95% of the time.
But if you're reading this, you probably aren't a computer, and so you probably don't have the time to recount how often each letter appears in each possible word after every round.
That’s where strategy comes into play.
Get smart to win later
If you're a person, it helps to start by understanding the central conflict of Wordle. See, your end goal, to guess the right word, is in direct conflict with your supporting goal, to get information about what letters could be in the word and where they could go. The word that is most likely to be the answer is also the one that will provide you the least information, and the one that gives you the most information will have the lowest odds of winning.
Let's look at an example. Let's say on your first play, you guess BEARS, and— lucky you! you got four greens on the last four letters of the word! Now all you have to do is figure out what slots in for the B, and you're golden. You're almost there— but how do you get from "almost there" to "there"?
If we try to win on turn 2, we might guess something like TEARS— that word has really high odds of being right. But, what happens if it turns out that T is gray? What have we learned from guessing TEARS? We still have 20 letters of the alphabet that haven't been tested, that we know nothing about. This is really bad news for us, because we've only got 4 turns left, and there are WAY more than 4 options for 5-letter words that end in _EARS. The first character could be P, D, F, G, H, W, N, Y, or even the S or R that we've already tested, since double letters are allowed. With 10 options and only 4 guesses left, our odds of winning with this method are less than half!
Instead, what we want to do is see if we can turn any of that list of possible first letters yellow, and check as many at a time as we can. We forget about EARS, and instead test a word like HYPED or WAGON, that lets us try to turn lots of possible first characters yellow— if any of those letters turn yellow, we know we've got the winner in HEARS or YEARS, since there's only one place they could go. Even though we know for certain that HYPED can't be the answer, we still get a lot of information from playing it (four times as much as playing TEARS, which can help us figure out what IS the answer.
This is an extreme case, but the same idea applies whenever you play. If you play MUTED in the first round and all you get is a yellow E, which would you rather do: play something like EERIE, to try to lock down where that E goes, but learn little else? Or play something like CRASH that won't tell you anything else about E, but gets you information about five totally unknown letters? Would you rather guess a word that you know has E for the second letter, or one that you know has E, S, and R in it?
Of course, Wordle is limited to words that actually exist in English, so eventually you’ll have narrowed in to a point where there are only a few possibilities. That’s where strategy has to end, and your own knowledge of English has to kick in— if you’ve got a yellow I and D, and have grayed out pretty much everything else, maybe the right play really is VIVID. That’s a play that, if it’s right, well, it’s right— but if it’s wrong, you’ll learn very little from it.
It’s in the rules
If you're still not convinced that reusing your yellows and greens isn't always the optimal play, let's look at one other case: the game's hard mode. In hard mode, you have to use your clues in all later guesses: once you get a yellow A in ARISE, you have to include an A in every guess, and none of them can contain RISE. Once you turn that A green by guessing TANGY, every guess has to have A in the second slot.
Why does that make the game harder? If your goal is just to guess the word most likely to be the answer, this wouldn't change your style at all— of course you'd reuse your clues, those are the ones likely to be in the answer!
The reason this is harder is because of its effect on information gathering: every time you learn something, that makes it harder to learn more. If you have one yellow, you can only test 4 new characters in each round, rather than five. As we saw with our BEARS example, if you get a bunch of yellows or greens, your ability to figure out what the fifth letter is gets extremely limited— if you’re required to reuse them, which on easy mode, you’re not. Lucky you!
On hard mode, the best possible answer you can get is five grays, because your next guess will still let you learn about five letters, not just four or fewer.
Strategy & tactics
So, how do you balance this tension? I think it helps to target your goals one at a time, by turn.
I like to start off by gathering as much information as I can, testing a wide range of common letters, and not even taking a real stab at the answer until my fourth or fifth move. Again, playing a move that I think might be the answer on turn 3 is great if I'm right, but makes my position much worse if I'm wrong, because re-playing greens tells you nothing new at all. Better to keep gathering information right up to the end, unless you're absolutely certain.
I try to guess letters I haven’t tried yet until I have three yellows and greens, at least one of them a vowel. Then I try using my yellows, but NOT my greens, until I’ve got about three greens. At that point, I can treat this less like a game of chance or statistics, but like an anagram puzzle: what move starts with H, ends with S, and has an E and an R somewhere in the middle?
Below are the things I keep in mind when playing:
DON'T do the intuitive thing, and just replay yellows and greens because you have them! As a rule of thumb, don't replay letters of a given color unless you have 3 of that color (or a stronger one).
Don't try to win with a wild guess: the odds are not generally in your favor, and a guess that could be the answer is a guess that gives you less information.
Early on, test as many letters as possible, starting with the most common letters: pairs of words like IDEAL / SHOUT or AROSE / UNLIT can let you identify your vowels early, and rule some common consonants in or out.
Don't reuse greens until you have at least 3, or you can't think of a word that doesn't use one.
And a few miscellaneous reminders:
Doubles are allowed, so having a green doesn't mean that letter can't appear again in the word. Don’t guess words with repeated letters until you’re trying to guess the answer.
You almost never want to reuse a deconfirmed / gray letter. There are extremely rare edge cases where a computer might find it optimal to do so, but our poor meat computers will never be able to detect those opportunities on our own, so better to just rule it out.
It’s a game
The great thing about Wordle is that it’s not your job: it’s just a game. You don’t have to make the optimal play, and you don’t have to play to win. All you have to do is what feels good to you. If winning and optimal play scratch an itch for you, I hope you’ve found this helpful.
If not, you do you.
Is Wordle your job, actually? Curious about my wordle-playing bot? Want to talk about other word-related strategies?