I mean we'd do more than twice as well with one question than with two, and more than twice as well with three than with two. Usually, diminishing returns leads us to learn less from each additional question, but not here. How do I express that?
when you have only two data points.
I have zero data points, I'm comparing hypothetical situations in which I ask aliens one or more questions about their technology. (It seems Dawkins' scenario got inverted somewhere along the way, but I don't think that makes any difference.)
I mean we'd do more than twice as well with one question than with two, and more than twice as well with three than with two. Usually, diminishing returns leads us to learn less from each additional question, but not here. How do I express that?
That's actually a claim of superexponential growth, but how you said it sounds ok. I'm actually not sure that you can get superexponential growth in a meaningful sense. If you have n bits of data you can't do better than having all n bits be completely independent. So if one is measuring information content in a ...
Here's the new thread for posting quotes, with the usual rules: