[adlpz]

I believe the point being made here is that, indeed, scientific method is based on accuracy of prediction.

However, in order to reach a good accuracy, a great deal of fine-tuning and tweaking is required. E.g. high energy physics. The standard model has something like 18 free parameters.

A new theory, which could be "more correct" and provide, in the long term, better predictions, might require time and manpower for this fine-tuning and tweaking to occur.

However, as it did not provide better accuracy of prediction in its initial stages, is said to be not worth pursuing or even directly pseudoscience or quackery.

This used to not be the case because, 100 years ago, relatively simple theories workable by 1-3 solo scientists provided a large enough breakthrough in prediction power to be seriously considered.

It may be the case however that nowadays, with the amazing level of precision measurement we are able to achieve, we've optimized ourselves into a corner.

We've fitter our quite-a-lot-of-degrees-of-freedom theories into a local maximum so hard that finding a theory that predicts better is, at least, impractical.

And through this process we've blinded ourselves from any new and disruptive ideas.

Again, look at the standard model. It's got so many damn dials to tweak that no wonder it fits reality so well. And if it ever doesn't, we can just shove supersymmetry in there.

We need to be able to dedicate resources to theories that do not provide better predictions, but that provide new perspectives. A moderate amount of resources. But calling those scientists quacks does no good.

[qsort]

> We need to be able to dedicate resources to theories that do not provide better predictions, but that provide new perspectives.

You are talking as if this isn't the case already, for instance: https://en.wikipedia.org/wiki/Physics_beyond_the_Standard_Mo...

I don't think that the usefulness of alternative viewpoints is contested by anyone, the problematic part of the article is the vague, ill-informed critique of 'scientific rigor'.

What other way is there to judge models than to compare them with experimental results? Grounding theories to experiments is literally the only thing that separates science from crackpottery.

It's also pretty weird for the article to criticize "logical positivism", when the most common contemporary views on the topic of philosophy of science (e.g. Popper, Kuhn, Putnam) don't actually agree with logical positivism.

[adlpz]

I agree, this is the case already to some extent. And yes, the article is quite vague in general. I believe it's intended as thought-provoking and not really a proper argumentation. I think in that sense it's an interesting piece. We must remember to not be too myopic when knees-deep into our over-fitted math frameworks.

I'd only add that, in my opinion, the vehement rejection of non-mainstream ideas is actually more common in non-professional circles. While a physicist may find interesting considering a new idea, a physics enthusiast, in my personal experience, is much more likely to acuse of pseudoscience and crackpottery. It's sort of a validating and gregarious warm and fuzzy feeling. Look at those people with their clearly energy-conservation-violating nonsense. Hey, I've done it before.

[8note]

We already have the best disruptive model that doesn't make good predictions: string theory.

There's not really a need to come up with these alternatives when we know string theory is powerful enough to do it, and only has one free parameter. That is, unless you can come up with a theory that had no free parameters.

These theories are fundamentally math models, and I don't find it likely that we'll determine that p = np by finding just the right np problem and exploring that on its own.

In the theory department, we're fine with what we've got. What we're missing is new experiements that can actually challenge our models. We've got a couple things to push on, sure { black hole singularities, dark matter, dark energy, wave function collapse, quantum gravity } but we don't have the tools to manipulate and observe them like we do for electromagnetism or chromodynamics

[joppy]

Does string theory only have one parameter? I was under the impression that you also had to insert some very complicated underlying geometry into spacetime (the extra dimensions), which I would count as more parameters.

[dj_mc_merlin]

You would be somewhat right, this is known as the landscape problem. Technically string theory has no free dimensionless parameters (by the formal definition, values which need to be determined experimentally).

edit: to expand, the landscape problem is more about finding the right parameters for compactification (folding up the geometry) that produce the universe we all know and love, or at least a realistic one. This makes them similar to free parameters.