Pop Sanders

The game is one button. You claim a display name, you tap, your count goes up, and a leaderboard of the top players updates in front of everyone. At the end of the campaign it freezes and the final standings are published.

The mechanic is Popcat, and we won't pretend otherwise. That was the right call for a campaign: a format people already know needs no instructions, and instructions are the last thing you want standing between an ad and a player.

So the game was the easy part. All of the difficulty is in the fact that everyone plays it at the same time.

A national campaign doesn't arrive gradually. It arrives when the ad runs, and it arrives all at once. We built for two hundred thousand people clicking simultaneously, because the version that handles a thousand and falls over at the peak is the version everyone sees fail.

The usual way to lose a game like this is a single scoreboard that every click has to queue up behind. So the counting is split across many independent counters, and the leaderboard everyone reads is a copy served from the edge, refreshed continuously. Reading the board never waits on someone else's tap.

Names are the other quiet problem. Two people claiming the same name at the same instant has to resolve to exactly one winner, in Thai or English, every time. That's handled where it can actually be guaranteed rather than hoped for.

The campaign ran to its scheduled end, at which point the game froze itself, published the final leaderboard, and then checked once more for late taps before closing the book.

Nothing about the last day needed a person standing by. That's the outcome worth having on campaign software: the interesting part was the marketing, not the infrastructure.