The Hackathon Insight That Changed How I Build

CareSRE, an AI powered outpatient queue system for government hospitals, won 1st place at the Google Developer Groups Vertex AI Hackathon. Roughly 2,400 teams competed.

Almost none of them lost because their code was bad. They lost because they were playing a different game than the one being scored.

Two games, and most people play the wrong one

In a 48 hour hackathon there's the game everyone assumes they're playing: build the most technically impressive thing possible. Elaborate architecture, every feature you can cram in, the newest model you can wire up.

Then there's the game that's actually being scored. Does this address a real problem, in a way a judge can understand in ninety seconds, that a real hospital could plausibly use next week.

We built the simpler system. We spent the time we saved on the user journey and the business case instead: who uses this, what does their day look like before and after, why would a hospital administrator say yes.

Spotting the gap is the actual skill

This shows up well outside hackathons. The game everyone is visibly playing and the game that determines the outcome are very often two different games. Founders lose to competitors with worse products. Engineers lose promotions to people who ship less impressive code.

The pattern is the same every time: optimising for the visible contest instead of the one that counts.

The trick isn't working harder at the obvious game. It's noticing when the obvious game isn't the real one.