If 2% of things “succeed” in a field, imagination matters at least as much as prediction. What do I mean by this? The key is to understand accuracy, precision and recall - concepts from machine learning / statistics / whatever you wanna call it. I never remember the formulae so here’s how I understand them.
If something happens 2% of the time, such as startup “success”, then you could predict things correctly 98% of the time by just always guessing a startup will fail. This would be your accuracy.
But don’t pat yourself on the back too soon. From the startups that end up succeeding, you’ll have predicted failure each time. Your recall would be 0%. That’s the price you pay for high accuracy in an unbalanced case like this.
Precision is a little more subtle I think. It’s the times you guessed success correctly, as a proportion of the times you guessed success. It’s kind of like how well you cry wolf (a good wolf in this case).
Let’s say you randomly guess success for each of 100 startups you see. Then you’d be right for 49/98 of the failures, and right for 1/2 of the successes on average.
> Your accuracy would be (49+1)/100 = 50% correct predictions overall.
> Your recall would be 1/2 = 50% of successes that you caught.
Your precision would be (times you guessed success correctly) / (times you guessed success) = 1 / 50 = 2%. So you’re crying wolf way too much here, so to speak.
In an unbalanced situation like this, prediction accuracy will mean you miss out on success, with low recall. “Crying wolf” helps with recall, but comes at a cost.
The precision/accuracy/recall dynamic can be applied to so many things in real-life, and is a really powerful mental model. In the startup case, I like to analogise it as the difference between trying to predict well (as is typical in many fields), versus having the imagination to see how things could succeed.
Imagination matters at least as much as prediction.