[chriswarbo]

Information is a measure of surprise: we should choose to verify our predictions; but it's disappointing when our predictions turn out to be correct.

We want our best predictions to be proven wrong, since we would learn a lot.

[dylan604]

I get where you're coming from, but at the same time, we have to KNOW things before we can continue to theorize about new things. If you keep theorizing on things that turn out to not be true, then it's just a waste of everyone's time. Find things that are true, theorize on the next step, test, prove/disprove, lather, rinse, repeat.

Just coming up with stuff and disproving it isn't always "learning a lot", it's just showing how bad we are at logically thinking about the next step.

[chriswarbo]

> Just coming up with stuff and disproving it isn't always "learning a lot", it's just showing how bad we are at logically thinking about the next step.

That's why I said "our best predictions"; as in, what we actually think is the case (and would put money on).

[notriddle]

If "information is a measure of surprise", then you don't want every prediction to be wrong. You want every prediction to have a 50% chance of being wrong. Half of your experiments should be failures.

[dylan604]

I don't understand this logic at all. At those odds, you're just guessing like flipping a coin. An inferred prediction should have a higher success rate. We look at the data we do know, we see what holes are there, and then make predictions based on all of our previous knowledge. The fact that the previous knowledge allows us to make more accurate predictions shows we have a better understanding of the subject than just random coin flips.

[matthewdgreen]

The most apt phrase here is “building castles in the air.”

[specialist]

Yes and, riffing on your clarification, maybe also make a distinction between hypothesis and prediction. Both are based on models, but a hypothesis hasn't been experimentally validated yet.