Hacker News Reader

Comment by rezgi | Hacker News Reader

Isn't it that when you combine two extremely low prevalence phenomenons together, you can't really make any accurate predictions because the numbers are so low that the error margin is too high? I might be wrong, but I seem to remember something to that effect. Could it be what's at play here?

View on HN

The thing is, if you had a discrepancy of some order of magnitude (e.g. 6,000 or 60 instead of 600), you could talk about accuracy problems. You could look into rounding errors and margins of errors, like we do with constants in physics, and that would maybe yield some new or modified equations (models).

But the "binary" absence of even 1 single case hints at something else: it's a category thing, there's "in it" or "out of it", and it seems that being born blind somehow means you can't develop schizophrenia.

Or it shows underdiagnois or hard to find almost-never events

Eh indeed, and it could even be that schizophrenia (whatever its "base principle" is in the brain) exists in some people born blind, but presents itself in such a different way that it's misdiagnosed for something else (so categorically, that we already have another name for it).

But I think it's really worth investigating. Could yield important knowledge and may lead to new forms of treatment perhaps.

The absence of any cases is strong evidence against the null hypothesis that the both are independent (assuming the combination isn't just much harder to diagnose), but it isn't strong evidence for it being impossible. Just because something hasn't happened doesn't allow you to distinguish between it being impossible and it being very unlikely (of course the other way around does work, then this is strong evidence that it is possible).

I will yield to your logic, and thanks for taking the time to explain how my approximation was flawed. I really need to brush up my logic skills... I did say "hint" though, which really means it is hypothesis, not strict logic at that point.

Edit: wait, no, I re-read my post and clearly, I did not make the logical fallacy. You're correct, and I did see that, hence using the word "hints at", not "means that". But my wording was bad afterwards ("is" instead of "would be"). : ) so thanks for clarifying.

Reply on news.ycombinator.com