How Wordle Was Made in 5 Days (Josh Wardle’s Story)

Crop faceless child in sneakers sitting on floor and playing with wooden blocks

How Wordle was made is one of the cleanest stories in modern web gaming. No team, no funding round, no roadmap. A Brooklyn-based software engineer named Josh Wardle revived a word-game prototype during the pandemic as a quick side project for his partner, and shipped it to the open web in October 2021. The famous “five days” figure is a popular retelling rather than a number Wardle is on record giving — but the build really was that small. By January 2022, the New York Times had bought it.

Key takeaways

  • Josh Wardle first prototyped Wordle in 2013, then rebuilt it in 2020 as a gift for his partner, Palak Shah; the “five days” figure is folklore rather than a documented timeline.
  • The original was a static web page — no backend, no accounts, with the answer list shipped in the page’s JavaScript.
  • Wardle launched it publicly in October 2021 at powerlanguage.co.uk/wordle/.
  • The emoji share grid, added in December 2021, was the feature most credited with triggering virality.
  • The NYT acquired Wordle on January 31, 2022, for a “low seven figures” sum.

The five-day build

The “five days” in countless retellings is shorthand, not a quote. We couldn’t find Wardle giving a precise day count; what he has described is a small side project he picked back up in 2020, played privately with Palak Shah for months, and only then released publicly in October 2021. The share grid came later still.

Whatever the exact number, a short build tracks with the technical surface area. Wordle has no server logic in its original form. The daily answer is selected from a fixed array using the date as an index. The grading is a single function that compares the guess to the answer and returns a color for each letter. The UI is a 6×5 grid and an on-screen keyboard. A competent JavaScript engineer could build the core in a matter of days.

The starting point

Wardle first built a version of the game in 2013 — almost a decade before the public launch — drawing on the color-coded feedback of the code-breaking game Mastermind. The format also echoes the pencil-and-paper game Jotto and the TV game show Lingo. That early prototype let you play puzzle after puzzle from an unfiltered word list, and he shelved it. The pandemic, and his partner Palak Shah’s love of word games, brought him back to it. Shah helped sort the dictionary into words she knew and didn’t know, which became the basis for a curated answer list of roughly 2,000–2,500 words — enough for about five years of daily puzzles.

That earlier prototype matters because it explains why the rebuild could move quickly. Wardle wasn’t designing the rules from scratch; he was implementing a format he had already thought through. The rebuild was about refining a format he already knew, not inventing one.

The tech stack

The original Wordle was static. A single web page with its JavaScript and CSS, no server logic, and nothing beyond static file hosting. The word list lived in a JavaScript array in the page source — players could pop open DevTools and read the full upcoming sequence if they wanted to.

This minimalism is part of why Wordle scaled. There was no database to crash, no auth flow to fail, no API rate limit to hit. Every player ran the same JavaScript locally against the same date-indexed word list. The marginal cost of an additional player was the cost of serving a few kilobytes of static content.

The daily-word constraint

One puzzle per day. No way to play yesterday’s puzzle. No way to play tomorrow’s. This wasn’t a monetization hook — Wordle had no monetization. Wardle has said the one-a-day limit was deliberate, creating a sense of scarcity, and it fit the original use case. Wardle and Shah were playing the same word together. The daily cadence was the cadence of two people doing a shared activity.

Scaled to millions, that constraint turned into scarcity engineering. Players couldn’t binge. They could only play once and then talk about it. The conversation that followed each daily puzzle is what made Wordle a social object rather than just a game.

The share grid

The share grid was not in the original build. Wardle added it in December 2021, about two months after launch, after noticing players in New Zealand sharing their results as emoji squares. He built a one-click “Share” button that copied a spoiler-free emoji grid of the player’s run to the clipboard.

That feature is the inflection point. On November 1, 2021, Wordle had 90 players, by widely reported figures; by early January 2022 it had more than 300,000, and it reached millions soon after. The grid is the example most game designers cite when they talk about coordination mechanics — a way to share an experience without spoiling it.

The viral arc

By late December 2021, the emoji grids were everywhere on Twitter. By early January, Wordle was a topic on cable news. The New York Times published a profile of Wardle and the game in early January 2022. From there, coverage cascaded across every English-language outlet that runs a culture desk.

Wardle made no money from this. The game was free, no ads, no upsells. He has said publicly that the scale of attention was overwhelming for a one-person side project, and the NYT acquisition arrived only weeks into that spotlight.

The NYT handoff

The New York Times announced its acquisition on January 31, 2022. The Times described the price only as “in the low seven figures,” and no exact figure was disclosed. The NYT moved the game to its own domain under NYT Games and kept the daily puzzle free.

Wardle has kept a low profile about the game since the deal closed. The handoff is one of the cleaner exits in indie game history — a creator who shipped, scaled accidentally, and handed the game to a publisher that kept it free.

What the build teaches

The Wordle build is often cited as evidence that small games can succeed on the open web. The lesson is real but specific. Wordle worked because the format was already well-defined (Mastermind and Jotto had the core feedback mechanic decades earlier), the prototype was already in Wardle’s head, the constraints were tight enough that scope creep was impossible, and the launch surface was free and frictionless.

The lesson is not “ship anything in a week and you’ll get a NYT acquisition.” It’s that when you have a clear format, a clear audience (in Wardle’s case, one person), and a willingness to keep the surface tiny, the build itself is the smallest part of the project. For a deeper profile of the creator, see our piece on who made Wordle, and our history of Wordle traces the format’s roots back through the 20th century.

Frequently asked questions

Did Josh Wardle really make Wordle in 5 days?

Probably not in any literal sense. The “five days” is a popular shorthand, not a timeline Wardle is on record giving. The real story spans a 2013 prototype, a 2020 rebuild for his partner, months of private play, an October 2021 launch, and a share grid added that December.

What language is Wordle written in?

The original Wordle is a client-side web app written in JavaScript, with HTML and CSS. It had no backend; the daily word list lived in the page source and was selected by date.

Who was Wordle originally built for?

Wardle built Wordle as a private gift for his partner, Palak Shah, who loved word games. They played it together, and shared it with family, for months before he released the public version in October 2021.

How did Wordle go viral?

The breakthrough was the emoji share grid Wardle added in December 2021. Players started posting their colored-square results on Twitter without spoiling the answer, which turned every daily puzzle into a public conversation.

How much did NYT pay for Wordle?

The New York Times said only that it paid “an undisclosed price in the low seven figures.” No exact number has been confirmed. The Wikipedia Wordle entry tracks the public reporting in detail.

The takeaway

How Wordle was made is the story of a small build by a thoughtful engineer, sharpened across months and released without expectations. A 2013 prototype, a 2020 rebuild, and about three months from public launch to NYT acquisition. The technical surface is the smallest of single-page web apps; the design taste behind it is the rare thing. For a similar minimalist browser experience with a different rhythm, the Chrome Dino game waits at the top of this site — one input, one mechanic, no upsell.

🔌 Connect any AI assistant to dinogame.gg — we run an MCP server: https://dinogame.gg/mcp