Hacker News Reader

Comment by moomin | Hacker News Reader

I don’t think people are understanding the subtlety of what she’s saying here. She’s saying that since there are no predictions to be made, there’s nothing falsifiable in any of the many theories. Worse, if the theories _do_ make falsifiable predictions that still will only reduce the imaginative space, not reduce it down to one theory, and that instead we choose between theories on the basis of how many “constants” they require.

This argument isn’t invalidated by possible future discoveries. It’ll still be possible to generate competing elegant mathematical models that only differ in ways that are unobserved.

View on HN

Yes, but this is a generic rubric to apply to all frontiers of science: falsifiable theories are good; unfalsifiable theories are bad; even when you falsify a theory, there will always be room to fudge and rescue it by making it more baroque; the principle of parsimony should be applied to give less credence to baroque theories.

I see no reason to place the frontier of "what happened right at / before the Big Bang where vanilla GR breaks down" in a different epistemic category than all other frontiers of science, as the blog post seems to do. i.e., I see no reason that we should not build models and collect data in order to expand into this frontier, as we profitably do in other areas of science.

There’s a couple of easy explanations for this 1) It’s her field of expertise 2) There aren’t any other frontiers of science where so much publicity and money is being spent on stuff that doesn’t look like it will ever advance science.

A concise shorthand for "non-falsifiable theory" is "faith".

Only if you choose to believe that a non-falsifiable theory is actually true. Otherwise it's not faith, merely a hypothesis.

As anyone who’s watched Dinosaur Train could tell you, a hypothesis is “an idea you can test”.

Dinosaur Train doesn't get everything right. An untestable hypothesis is still a hypothesis. Even an unfalsifiable hypothesis is a hypothesis. But testable hypotheses are better then untestable (and unfalsifiable) ones because, well, you can test them to see if they are false.

Reply on news.ycombinator.com