Seven things we learned from analyzing 515 million words

Millions of people play Wordle every day and share, discuss and debate how they tried to win.

For the first time, we analyzed player performance in half a billion of these Wordle games over the past year and compared their results with the strategies recommended by our WordleBot.

Here are seven things we learned:

1. Among the first 30 starting words, ADIEU is the most popular but the least effective.

Many words have been written about best opening words for Wordle. Answering this question was actually one of the motivations behind developing WordleBot. In his robot brain, a handful of words — SLATE, CRANE, TRACE – receive the robot’s seal of approval as having led to the solution with the fewest guesses on average.

But for human Wordle players, the most popular opening word, by far, is FAREWELLwith AUDIOanother four-vowel word, not far behind.

The strategy seems logical: figure out the vowels and the other letters will fall into place. The bot has never favored starting this way, but it is possible that the best strategy for a computer is not the best strategy for a human.

Our new analysis, however, shows that from FAREWELL Or AUDIO actually puts human players at a disadvantage. How many one?

On average, players who started with FAREWELL It took them about a third of a turn longer to resolve their words compared to players who started with SLATEtotaling up to 132 additional rounds over the course of a year.

Worse still: while 1.7 percent of SLATE beginners failed to solve a Wordle on average, the risk of failure with FAREWELL more than doubled, to 3.6 percent.

We cannot say with certainty that starting with FAREWELL causes players do worse. It could be that players who start with FAREWELL are more casual players or are less concerned with solving the smallest number of guesses. After all, not everyone cares about raw efficiency when playing. But we can say that, among the 30 most popular Wordle openers, FAREWELL achieved the worst.

In general, the average WordleBot user solves a puzzle in about a quarter turn less than the average Wordle player.

2. People love holiday words: PARTY, HEART, RABBIT, GHOST.

The bot starts with SLATE every day, which can be boring. Human players, freed from the burden of mathematical precision, can and do change their Wordle openers throughout the year. (Players often explain their daily selection in the Wordle Review.)

Here are the main opening words that gained popularity on certain days:

• Christmas Eve (December 24, 2022): HAPPY

• The day of Christmas: HAPPY, GIFTS, PEACE

• New Year’s Eve: TO PARTY, YEARS

• New Year’s Day (January 1, 2023): YEARS, TO PARTY, HAPPY, COSTS

• The Super Bowl (February 12): GREAT

• Valentine’s Day: HEART, LOVER, CUPID

• St. Patrick’s Day: LUCKY

• Easter: RESURRECTED, RABBIT

• Coronation of Charles III and Camilla (May 6): CROWN, ROYAL

• Labor Day: WORK

Another fun pattern: TO PARTY there is a small peak every weekend.

3. More people solve Wordle on their first guess than can be explained by chance.

In the list above, we have excluded the first guesses which constituted the Wordle solution of the day.

Indeed, in about one in 250 games, a reader gets the answer right the first time. This is much more often than you would expect if you left things to chance, although not all cases are necessarily outright cheating.

Some may use a new window to make additional guesses after failing to solve the puzzle in six attempts. Some may reintegrate a solution found on another device to maintain a sequence or to test a technical problem. Others may have had the answer spoiled or, yes, may have looked it up.

4. SLATE and STARE are on the rise, while CRANE is becoming less and less popular.

Here you can see which opening words have become more popular over time. FAREWELL remains dominant, and its popularity as an opening guess has even increased.

Popularity peaks due to external events – when “FAREWELL » was featured in the mini crosswords in August – or from the moment a word appears as a solution – AUDIO August 30, CRANE on June 21 – are obvious.

5. Hardest words? Keep an eye on J.

The most difficult words to solve started with J, ended with Y, or had a double letter somewhere. The hardest of all last year’s words, JAZZ, has all three. The five most difficult words of the past year, measured by number of rounds to solve:

• JAZZ (June 1): 5.82 rounds to solve

• JOKER (April 25): 5.69 rounds to solve

• NANNY (June 3): 5.68 rounds to solve

• JUDGE (December 26): 5.57 turns to solve

• WALL (February 22): 5.52 turns to solve

If you measure by solution rate, the most difficult Wordle puzzle was JOKERwith only 71 percent of players finding the solution.

6. The words that surprised the robot the most were ANODE and MATEY.

For over a year, WordleBot has not been able to view any list of solutions. Instead, the robot assigns almost every five-letter word in the English language – all 14,855 accepted Wordle guesses – a probability of being a solution. (The robot has no role in choosing which words appear as Wordle solutions; if it did, this task would be much easier.)

Most of these words — AAHED has ZYMIC – are considered too obscure to be a likely solution and have no chance. For the rest of the words, the more common a word is (measured by its frequency of appearance in the New York Times since 2000), the higher the chance that the robot assigns to it.

Most of the words that appeared as solutions — since CONFORM has YOUTH – had the maximum possible chances. But from time to time, a solution surprises the robot. His solution model is not perfect! This is one area where a savvy human player can outperform the bot.

Here are the solutions from last year that surprised the bot the most, along with what it thought were their likelihood of appearing, relative to the most common words.

• ANODE (August 5): 67.9%

• MATEY (January 22): 73.2%

• BORAX (April 12): 81.2%

• GUPPY (May 4): 86.0%

• DEBUG (February 11): 89.4%

7. The bot learns from your guesses.

Human players are free to choose any of the approximately 15,000 accepted Wordle words when playing. The bot is not so lucky.

Deciding which words are common enough for the bot to recommend was a matter of judgment. We want the bot to be as competent as possible. On the other hand, it wouldn’t be helpful to have a bot that scolds players for not guessing words like AWDLS, LAARF Or YOKE.

Incorporating data from player guesses will allow us to be more deliberate about the words the bot recommends. In the coming days, we will remove a handful of words from the bot’s list and make room for around fifty additions. A new batch of scientific terms — MALIC, CILIA, BORIC And NODAL – with many words related to food – CABBAGE, PANKO, MOREL, CHILE, UMAMI, JAMMY, BENTOS And BREAM – will make an appearance. We assume the robot spent more time reading the Science and NYT Cooking section.

Note: In cases where we averaged guesses, failed puzzles counted as 7.5 guesses.

gn EN headline