Sometimes You Just Have to Risk Looking Silly

Imagine you're a soccer player deciding on where to aim your penalty kick. What's the best approach? Upper left? Lower right? Nope.

It's a statistical fact (which I believe I learned from one of Malcolm Gladwell's essays, but I'm not sure which one) that your best bet is to aim straight for the centre!

That's because the goalie will almost certainly jump for one of the corners, leaving them unable to get to the ball.

But it's also a statistical fact that most players do not aim straight for the centre. Why is that? Because if you kick straight and the goalie holds it (by not moving at all), you look mighty stupid.

The mere fear of looking stupid is a powerful driver that makes us pick overly complicated solutions without a higher chance of success.

Next time you're...

• grabbing for that GPT-4.1 integration, ask if a finetuned smaller model (BERT?) wouldn't do the trick

• setting up a neural network dozens of layers deep, see if random forests, logistic regressions, or a gradient-boost model wouldn't actually work better

• spreading your app over hundred microservices, ask if a single instance and a monolithic architecture wouldn't serve you better

An old Latin (I believe...) quip advises to "better remain silent and be thought a fool than to speak and remove all doubt." Maybe it's time for an update:

"Better be thought a fool for keeping it simple than be a fool for making it complex."