Hacker News Reader

Comment by kloch | Hacker News Reader

> Yes, dark matter is the best way we currently have for fitting the data

This is the key problem. Since there is no theory of what dark matter is made of or how it is created and destroyed, you can always have as much or as little of it as you need to fit the data.

As for gravity, we don't have anything close to a working theory of quantum gravity. We can't even measure the classical gravitational constant to more than about 4 digits. The reproducibility by different experiments has not demonstrably improved since the 1940's! Clearly there is a lot more we have to learn about gravity.

View on HN

There's actually an embarrassingly large number of theories for what makes up dark matter. I doubt most people have ever even heard of the form that was ruled out in this article. There's not just a ton of theories, a great number of them are falsibiable with experiments we can plausibly perform.

And yes, quantum gravity is also something we don't understand. But the domain where we are missing that understanding doesn't overlap with the dark matter "problem".

Having a large number of theories is embarrassing! That means you have no idea. Gravity and QM are so great because they are clear winners – they are not easy to modify a little bit and nothing approaches them in terms of practical power.

Falsifiability comes from having specific predictions, which means one theory that's an island, not a bunch of approximations.

The fact that there's any "amount" - a single scalar quantity of a thing we already have all the equations for - that neatly explains multidimensional data, is highly suggestive by itself, no?

It's not just a fudge factor, it's a fudge factor that behaves exactly like mass.

Any fudge factor with a continuous effect on the output that covers the range of interest can be used to adjust a given output to an output of choice.

For dark matter, one of the problems is that any given instance is allowed to have whatever amount of dark matter gives the output we want. This makes the math work out great, but weakens its predictive power.

Reply on news.ycombinator.com