Josh Wardle built it for two people. Within three months, two million people were playing it daily. What the green-and-yellow grid got exactly right about the internet.
On October 1, 2021, Josh Wardle released a word game on the internet that 90 people played. By January 2, 2022, 300,000 people were playing it. By the time the New York Times acquired it three weeks later, the number was in the millions. The game had no app, no push notifications, no advertising budget, and no monetization strategy. It had five letters, six guesses, and one puzzle per day — and it became the most talked-about word game since the crossword craze of the 1920s.
Understanding why Wordle worked the way it did tells you something interesting not just about puzzle design, but about how human attention and social behavior interact with constraints.
Josh Wardle is a software engineer who had previously worked at Reddit and was known internally for creating community experiments — the collaborative art canvas Place, the button with no explanation. He had a habit of building things that were about collective experience rather than individual consumption. In 2021, working from home during the pandemic, he built a word game for his partner Palak Shah, who loved word games. He called it Wordle — a pun on his own surname.
The game was private for most of 2021. Wardle and Shah played it together and refined the word list by filtering out obscure words that felt unfair. The list of 2,309 valid answers was carefully curated — no plurals ending in S, no obscure scientific terms, no words that would stump a reasonably well-read adult. This curation, which Wardle and Shah did by hand, is part of what made the game feel fair rather than punishing.
When Wardle released Wordle publicly in October 2021, several specific decisions set it apart from the hundreds of word games already available online. First: one puzzle per day, the same puzzle for every player worldwide. This was borrowed loosely from the New York Times crossword model — a shared experience rather than an infinite supply of on-demand puzzles. Second: no account required, no app needed, no data collection. You visited a web page, played, and left. The simplicity was radical in an era of engagement-maximizing apps. Third: the shareable result grid.
The emoji grid deserves its own analysis. When you finish a Wordle, you can share your result as a pattern of colored squares — green for correct position, yellow for wrong position — that shows how you solved the puzzle without revealing the answer word. This is a stroke of social design genius: it is inherently interesting (how did you solve it?), inherently sharable (it fits in a tweet), and non-spoilery (your friends can still play). Twitter and group chats filled with these grids in December 2021, and each one was free advertising for a game that had never bought an ad.
The single daily puzzle is not a limitation — it is the entire product design. By giving players exactly one puzzle per day and making it the same puzzle for everyone, Wordle created a shared cultural moment. At its peak, office workers compared Wordle results the way previous generations compared last night's television. The constraint that looks like a missing feature — why can't I play more? — is the reason Wordle didn't become just another word game that people burned through and abandoned.
This is counterintuitive for internet product designers trained to maximize engagement. Wordle's engagement strategy was the opposite: minimize time-on-site, maximize daily return visits. Give people something they can finish in five minutes and then make them wait 24 hours for more. This scarcity model is as old as the serialized newspaper puzzle, and Wardle had either rediscovered it or intuited it from his observation of how the NYT crossword worked.
The New York Times acquired Wordle in January 2022 for a sum reported to be in the low seven figures. Wardle had already committed to keeping the game free, and the Times honored that commitment — Wordle remained free to play even as the Times used it as a gateway to its paid Games subscription. The transition was not entirely smooth: some players noticed the word list had been quietly edited, removing a handful of words that had become politically charged or unusually obscure. The community noticed, because Wordle players are extraordinarily attentive to the word list.
The acquisition also triggered an extraordinary flowering of Wordle variants. Quordle made you solve four Wordles simultaneously. Nerdle replaced letters with arithmetic equations. Worldle was a geography puzzle using country silhouettes. Heardle was a music identification game. By mid-2022, there were hundreds of Wordle-adjacent games, most of them using the same core mechanic of limited guesses plus color-coded feedback. The format had proven itself as a generative template, not just a single product.
Wordle attracted serious analysis from mathematicians and computer scientists interested in optimal play. The question — what starting word maximizes your expected information gain from the first guess — is a classic information theory problem. 3Blue1Brown's YouTube analysis of Wordle using entropy calculations reached over ten million views, introducing a generation of casual players to concepts from Claude Shannon's foundational work on information theory. Words like CRANE, STARE, TRACE, and SALET emerged as near-optimal openers because they covered the most common letters in the English five-letter word space.
This unexpected crossover between a casual word game and graduate-level mathematics says something about the puzzle community's appetite for rigor. Given a simple, well-defined problem, puzzle enthusiasts will analyze it exhaustively and share their findings with everyone. The best puzzles invite this kind of engagement rather than resisting it.