Build like kindergartners

There's a design challenge I keep coming back to. Teams get raw spaghetti, tape, and one marshmallow. Twenty minutes to build the tallest structure with the marshmallow on top.

Recent business school graduates do the worst. They spend the first minutes planning, negotiate who's in charge, design something ambitious, and then build it for nineteen minutes. In the last minute they place the marshmallow. The structure buckles. The marshmallow is heavier than it looks.

Kindergartners do the best. They don't plan. They put the marshmallow on a couple of sticks in the first minute, watch it fall, and try something else. They get ten attempts in the time the adults get one, and by the end they have something that stands.

I think about it a lot now, because when we hand ops people the ability to build, the MBA move is the natural one. You have a picture of the finished workflow: the intake, the logic, the approvals, the output to the right system. You want to test that, not a piece of it. Testing a piece feels like admitting the picture isn't ready.

There's fear in it too. If you run the small version in week one and it fails, you know in week one. If you build the big version, you get eleven weeks of believing it works.

And there's the stakeholder. Demo something tiny and someone asks what the business value is. I could do that by hand in five seconds. Nobody wants to hear that twice, so builders go quiet for two months and come back with the tall tower.

Going months without testing risks building something that collapses at the finish line. The kindergartner approach is better: test a rough version by day three, improve it by day ten, and have it doing real work by week four.

The marshmallow is your first real run with real data. Put it on now.