michael_nielsen

seems like a team from Uber won this competition over teams from universities. But I dont know... https://twitter.com/spyrosmakrid/status/1012952827348570112 …

If a scientific problem can be cast as a competition, it's not science. Science is not prediction. Science is explanation, understanding, meaning. These are not empty words. Getting ten percentage points improvement in a forecasting exercise is not science.

Do you really have to be doing causal inference to be “doing science”? Surely much scientific progress can be made by building models that predict phenomena 10% better than current state of the art?

I didn’t say science was causal inference—I said it was explanation, understanding, meaning.

A good bit of science can often (not always!) aid prediction. The converse is rarely true.

Not to be a dick about this but it’s a common misunderstanding.

Here is a recent Kaggle competition that caught my eye...surely participating in competitions like this would qualify as science? https://www.kaggle.com/c/trackml-particle-identification …

So I want to be very careful here. There are many ways to participate in science.