Like 2 million other individuals these days, my mornings now start with a round of Wordle, a game in which players must guess a 5-letter "word of the day" in 6 attempts or less. For each incorrect guess, the game reveals information about which letters from the guess are in the correct position or show up elsewhere in the mystery word.

Strategies to Minimize Guesses

There are many ways to enjoy this game, but I initially focused on minimizing the number of guesses needed to find the word of the day. Turns out, there are already countless blog posts written on just that topic.

Bertrand Fan explains that SOARE is the best starting guess. He takes advantage of the fact that the full lists for both acceptable guesses and for upcoming words of the day are hard-coded into Wordle's source code, and he uses these lists to determine the most frequent letters to appear in each of the five positions and then choose a starting word based on that.

Tyler Glaiel takes a different stance. He claims that using Bertrand's strategy leaves players vulnerable to a worst-case scenario where 8 guesses are needed. Tyler instead focuses on choosing guesses that filter out as many of the remaining words as possible. Using this strategy, he recommends starting with ROATE, which finds the correct word of the day in an average of 3.49417 guesses. That or RAISE, if you're willing to take the hit down to 3.49546 guesses on average in order to have a chance at guessing the word of the day in one try.

Mr. Excel claims AROSE is best. He bases his choice off overall letter frequencies instead of using Bertrand's strategy of analyzing frequency per position. However, since SOARE is just an anagram of AROSE, I don't understand why both aren't equally good options. Eh. I'm mostly impressed that he uses Excel instead of a typical programming language to do the analysis.

Yet others recommend looking at sets of starting words rather than a single word. For example, starting with IDEAL and SOUTH immediately reveals information about all 5 vowels (sorry, Y!). Using SIREN, OCTAL, and DUMPY tests 15 out of the 26 possible letters.

All the above strategies prioritize positive information, i.e. using guesses to gain information about letters that appear in the word of the day. But it's possible to make use of negative information too. For example, if we know that U does not appear in our word of the day, then we can be 100% certain that Q does not appear either. Because of this, there's very little incentive to use starting words that contain QU.

Also notable about the above strategies – they don't work in hard mode. In hard mode, players must use all the information that they already know about the word of the day in their subsequent guesses. At first glance, this may not seem so bad, but it can actually make it impossible to guarantee that the word is found within 6 guesses. For example, if you first guess BARES and then learn that the letters ARES are all in the correct position, you now have only 5 guesses remaining but 13 potential words of the day to choose from:

CARES, DARES, FARES, GARES, HARES, LARES, MARES, NARES, PARES, RARES, TARES, VARES, WARES

Thus, exploration of letters in hard mode is done by picking words whose letters occur less frequently. Words like XYLYL (an aromatic functional group in chemistry) or HYPHY (short for "hyperactive") suddenly become more appealling starting words.

Variations of the Game

Wordle is not exactly the first of its kind, as it's very similar to Guess My Word or Lingo, the TV show. It won't be the last of its kind either. Since its release, several spinoffs have been made, all taking the same original mechanics and adding their own twist. These include:

  • Dordle (guessing two words of the day simultaneously)
  • Nerdle (guessing a mathematical equation)
  • Primel (guessing a prime number)
  • Sweardle (guessing a 4-letter word in 4 guesses)
  • Squardle (guessing 6 words structured as a crossword)
  • Absurdle (guessing against an adversarial word picker)

By and far, the most popular variants modify the original vocabulary lists. For example:

  • Lewdle (guessing with a vocabulary restricted to lewd words)
  • Queerdle (guessing words of the day that are related to LGBTQ+)
  • A Greener Wordle (guessing words of the day that are environmentally-related)

Of all these, I found Absurdle particularly fascinating. In Absurdle, the player's goal is still to minimize the number of guesses, but there isn't a fixed word of the day. Instead, as the player makes their guesses, the game picks a word that satisfies all known information so far but maximizes the number of guesses needed to find it. The game's FAQ page has this to say:

What is the best possible score in Absurdle?

4 guesses. And yes, this has been obtained by many different people.

6 guesses is good for a casual player, and 5 guesses is extremely good. To find a solution in 4 guesses almost always requires some kind of computer search. 3 guesses is impossible in this game.

And that's when I realized that this is probably why Wordle gives players 6 attempts. It's enough for the casual player to find the word of the day most of the time but not enough for them to be careless with their guesses.

The Future

Wordle was just bought by the New York Times for a cool million dollars or so, and I'm sure the developer, Josh Wardle, is very happy about that. I can't know for certain how the New York Times will change the game or if it'll stay free forever, but I wonder if certain words of the day will become less likely simply because they know that starting words like SOARE or ROATE are more common.

It reminds me of the hidden phrase in the bonus round of The Wheel of Fortune. Contestants guess the phrase after picking some letters whose locations are revealed if they appear in the phrase. The most common letters to be picked were RSTLNE, so eventually, the show started revealing those letters by default and allowed contestants to pick different letters instead. But now, it's believed that the show's producers pick phrases that don't have those letters as often!

Only time will tell. For now, I've stopped thinking so much about strategy and have just started picking words that make mini-summaries of my day. I tend to find the word of the day in 3 or 4 guesses, so I'm pretty satisfied. Looks like I can guess the words of the day in an average of 3.49417 guesses too!