Writing thirteen game pages in one sitting
Every game we'd made was a canvas element and nothing else. Sitting down and writing a real page for each of them was supposed to be an SEO chore. It turned into the most useful design review we've done.
Until this week, a game page on this site was: a title, a canvas, a leaderboard, a link home. That's a perfectly good game. It's a terrible page — there was nothing on it to read, nothing to find it by, and nothing to tell someone who'd just arrived what they were looking at.
So the job was to give all thirteen a written section: what the game is, how to play it, some strategy, a short FAQ. About 900 words each. We expected a long, boring day.
Writing "how to play" finds your design holes
The surprise was how often the writing refused to come out cleanly, and how reliably that turned out to be the game's fault rather than the writer's.
If you can't explain a control in one sentence, the control is too complicated. If the tips section is just "try to survive longer", the game doesn't have any strategy in it yet. If the FAQ has to explain something that isn't visible on screen, that thing needs to be on screen instead.
The clearest case was writing about the tether in Tether. The rope pumps itself as you swing — it reels in through the bottom of the arc and pays out at the top, the way you build height on a playground swing. Writing that sentence made us realise the game never communicates it. Players felt themselves accelerating and assumed it was a bug or a difficulty ramp, when it's the single most important mechanic in the game. The prose fixed something the interface should have.
The trap of the identical template
Thirteen pages in one sitting produces a very strong pull toward a formula, and we gave in to it: every page ended up with the same four headings in the same order — About, How to play, Tips & strategy, FAQ.
Individually each page is fine, and the prose in them is genuinely specific to each game. But laid side by side, thirteen pages with identical scaffolding read as generated, even when every word was written by hand. It's a real cost we didn't anticipate: the structure says "template" loudly enough that people stop crediting the content.
We're unwinding that now, letting each page's shape follow its own game — the daily puzzle gets a section on how its answer is chosen, the driving game gets a proper reversing explainer, the echo-location game gets a note on its sound design. Same effort, completely different fingerprint.
Where the writing goes matters
One practical thing that took a couple of attempts. The written section has to sit below the game, not above it, because the reason someone clicked was to play. A wall of text between a visitor and the play button is a page that gets closed.
But "below the fold" has to mean below one screen, not below three. The pattern that works is the game filling the first screen exactly, with a visible hint that there's more underneath. Players who want to play never see the text. Players who want to know how the scoring works scroll once and find it.
The FAQ that answers real questions
Our first drafts had the FAQ nobody needs: "Is it free?" "Do I need an account?" Useful, boring, and identical across thirteen games.
The better questions came from the reviews players had actually left. "Does it work on mobile" is a real question with a different answer per game, and for some of ours the honest answer is no. "Does it save my progress" matters enormously for a long game and not at all for a three-minute one. "Can I cheat the leaderboard" turned out to be something people genuinely wondered about, and answering it properly became a whole guide.
An FAQ built from questions you were actually asked reads completely differently from one built by imagining questions.
The one we missed
Thirteen games got a page that day. There are fourteen games on this site.
Strata launched four days later, through a slightly different build process, and shipped with a page containing eight words. Nobody noticed for nine days. That one has its own post, and it's the reason we're now building the check into the deploy pipeline rather than trusting ourselves to remember.
Worth doing?
Entirely, and not for the reason we started. The search-engine argument is real but slow. The immediate payoff was that describing thirteen games in plain language, back to back, is the best comparative review you can give your own work — you find out which ones you can explain in a sentence, which ones need a paragraph of apology, and which ones have a brilliant mechanic that the game itself never mentions.