Just because it’s falsifiable doesn’t mean it’s good science | Hacker News Reader

Just because it’s falsifiable doesn’t mean it’s good science

backreaction.blogspot.com

69 points_aleph2c_53 comments

Of course, but if it's falsifiable then once it's falsified it will go away. If it's not falsifiable then it's "not even wrong" and it's impossible to convince people to stop repeating it as if it's fact (or worse, an 'undeniable possibility').

Commenting on the headline versus the contents I see.

A time honored tradition, here.

That said, it's not always a bad thing. Sometimes you get two interesting discussions going on for one submission...

Right, but how many billions of dollars are you willing to spend to falsify a hypothesis?

The author is arguing that the proposal to replace the LHC with a 100km replacement isn't worth the cost considering the low probability that any of these theories it could prove actually turn out to be true. We have little to gain by them being false and little reason to think they are true.

Say the member countries have 500 million people in a population. In order for them to spend 20 billion on LHC, you would need to tax those people ~$40. My intuition is that people would be willing to pay that to understand that some theory is correct or incorrect. Even proving gravity waves could not be done for 100 years after the theory.

> My intuition is that people would be willing to pay that to understand that some theory is correct or incorrect.

When doing this sort of calculations, you have to take into account that you're taxing THE WHOLE population: completely destitute people, people near the poverty line, children who ear exactly $0, etc. For some of these poeple, $40 can be way too much. In practice, if you tax a child that doesn't earn an income, you're taxing the parents. Of course it's possible that on average you can tax $40 from each person, but that might require taxing $400 from some people and $0 from other people.

That said, there might be better ways of spending that money than on this.

You're not actually addressing the merits of the proposal, just reframing the terms of the cost. Why not spend that $40 per person on something with better prospects?

It's not about the cost, it's about the benefit. The author is arguing that the proposals for what to do with a super-LHC aren't very good and that's why it shouldn't be done.

So let's indeed spend the $20B- but spend it on 20 other $1B programs that will each provide better results and answers than what we'll get from poor theories that a $20B project can answer.

At the very least the money spent building the new experiment churns through the economy more efficiently than, say, tax cuts for rich people.

humans do research to learn.

learning is when our belief system changes in the face of evidence.

how should we hypothetically (in a thought experiment) measure this education of humanity?

strawman proposal: lets measure learning by voting on predictions, and then after measurement we can observe the discrepancy between collective prediction and actual measurement results.

problem with the straw man proposal: most of humanity will make wrong predictions, because most of humanity has not been trained in most disciplines.

so we want the "prediction of humanity" to somehow give most weight to the experts. obviously it will be contentious to define "expertness". Instead of trying to define "expertness" we could weigh the human predictions by the individual's success rate at previous predictions. Now we get a modified prediction distribution of humanity.

Nothing currently forces falsified predictions to go away, look at chapter 28 of Feynman lectures vol 2 on ElectroMagnetic mass. There's known inconsistencies in our theories, virtually no one highlights these, because everyone is trying to underscore their understanding and never lack of understanding of the universe...

You really didn’t read the article, did you?

He is arguing against building the larger particle collider.

There are too little theory support of such collider can find anything new under the standard model.

IMO, it is not the matter of "science". This is economics.

And economics isn't a science?

Is it falsifiable ? Is it testable even ?

When people are proven wrong, does the conversation move forward, or do large swathes of people grip onto their beliefs anyway?

That has nothing to do with whether or not something is a science.

It is when the people who don't move on are the people doing the "science."

No, even then it has nothing to do with whether or not the subject of debate is scientific.

Bad deontology on the part of an individual doesn't discredit the field as a whole.

Theoretically, yes it is falsifiable. If you had sufficient data about a system you could perfectly forecast micro- and macroeconomic results.

Now, is it testable? Only in a very limited sense. You can suss out trends in a decently empirical way, but because you're working with independent agents you can't really get sufficient data to obtain rigorous results.

Economics is not unlike high energy physics in these respects. You can move the needle in legitimate ways, but there are very hard restrictions that make it more difficult to make rigorous observations.

Sort of but not really: https://plato.stanford.edu/entries/economics/

*She

Yes, there is the question of putting our research money where it could do the most good, but that doesn't mean that it also isn't thoughtful commentary of the difference between better and worse science. One would think that good science would not bother coming up with misleading excuses to spend money on an experiment where there is no strong reason to believe it could result in the discovery of anything important.

Isn't the problem that the collider will be too big for Standard Model work, but not big enough to test the alternatives?

I can understand the argument, but what other methods do we have to test particle physics other than building larger and more powerful colliders? Are there some new designs that would make it cheaper?

Otherwise I don't mind spending money just for the sake of exploration, maybe that money could be "more well spent", but that's true for a lot of things. Who knows maybe it will find something interesting. It's not even a total waste anyway as it will probably spur jobs / research / interest.

Currently there's a good chance there's nothing to test.

The real problem in physics isn't lack of data, it's lack of direction. The Standard Model has been a thing for decades now. It's good at what it does, but there are a lot of fundamental things it doesn't do.

The proposed Future Circular Collider will cost at least €24bn.

That would fund any number of research programs willing to take a bold look at fundamentals from different angles.

Currently there's superstring theory and various attempts at quantum gravity, but none of them are moving much.

A number of less mainstream possible leads have been discarded for what are really political reasons - prestige and funding - not because they're poor science.

Imagine if we'd never had quantum theory or relativity. Physics would seem incredibly backward. But realistically, in today's academic climate Einstein would never be published.

€5bn a year for five years on open fundamental research with fewer political and academic constraints has a better chance of generating a breakthrough than spending €24bn on a big machine that will most likely produce no breakthroughs at all.

Of course it's a gamble, but in international terms €25bn is pocket money.

> But realistically, in today's academic climate Einstein would never be published.

Can you expand on this? This isn't a leading question, I actually don't immediately see what you mean.

Tangent: I find it interesting that futurists often claim technology is accelerating such that a "Singularity" of technological development may occur. To my eyes, however, the opposite seems true--each technological advancement seems exponentially more expensive, as if we've picked all the low hanging fruit. And now we're building $24bn particle accelerators to test really obscure things that may or may not have any impact on the next breakthrough.

It is a time honored tradition for futurists to look at an emerging technology, see that it currently has exponential growth, and then project that out to infinity.

So rarely do you hear them say "Dammit, yet another S curve! Why is it always an S curve in the end!?!"

> So rarely do you hear them say "Dammit, yet another S curve! Why is it always an S curve in the end!?!"

Kurzweil's books and articles are full of this, see, e.g.:

http://www.kurzweilai.net/seeing-the-s-curve-in-everything

http://www.kurzweilai.net/the-law-of-accelerating-returns

They are claiming this, because there are some hard evidences that this is the case (and also it makes logical sense): Each technological breakthrough accelerates (quite often) research in all other areas. Think about the invention of speech, writing, printing, computers, internet. And the cross-pollination with metallurgy, chemistry and physics. Just as one example.

What is meant by "natural" in the context of the Standard Model?

This:

https://en.wikipedia.org/wiki/Naturalness_(physics)

Hossenfelder wrote a book crtizising current mainstream thinking on the importance and role of naturalness:

https://www.goodreads.com/book/show/36341728-lost-in-math

Here is an excellent review that also links to an article by Wilczek defending the orthodoxy.

https://aapt.scitation.org/doi/10.1119/1.5086393?fbclid=IwAR...

Cool, thanks!

Did anyone ever say it did?

It's more about whether this aspect is properly considered. Some physicist think building bigger particle accelerators is good simply because it can falsify more models. However the the quality of those models must also be taken into account.

Then they should have led with that instead of the general point about falsifiability.

When writing for an audience that has limited attention to give always put the executive summary first.

I think it's fine. In the second line it says

> A major reason we see so many wrong predictions in the foundations of physics – and see those make headlines – is that both scientists and science writers take falsifiability to be a sufficient criterion for good science.

Personally the line that caught my eye straight away was the one on it's own that said

> Why does it matter?

after which comes the stuff about bigger accelerators.

Yes. Karl Popper's theory (from which we get falsifiability in the first place) declared falsifiability to be the necessary and sufficient condition for something to be scientific. He was obviously wrong on this point (plenty of good science doesn't start from a falsifiable hypothesis, and plenty of pseudo-science is falsifiable).

> Karl Popper's theory (from which we get falsifiability in the first place) declared falsifiability to be the necessary and sufficient condition for something to be scientific.

According to the article that's not the case. It says:

> But falsifiability alone is not sufficient to make a prediction scientific. (And, no, Popper never said so.)

Looking around at quotes of his, I see that he maintains falsifiability is necessary, but nothing that indicates he thinks its sufficient. Do you have a source by any chance?

> He was obviously wrong on this point (plenty of good science doesn't start from a falsifiable hypothesis, and plenty of pseudo-science is falsifiable).

Eh, I'm not sure I'd say he was obviously wrong. I still think that good science has to be falsifiable. What's a good counterexample? Or are you drawing a distinction between "is falsifiable" and "starts from a falsifiable hypothesis"?

I was intrigued by this also. According to https://plato.stanford.edu/entries/pseudo-science/ :

> However, in what seems to be his last statement of his position, Popper declared that falsifiability is a both necessary and a sufficient criterion. “A sentence (or a theory) is empirical-scientific if and only if it is falsifiable.” Furthermore, he emphasized that the falsifiability referred to here “only has to do with the logical structure of sentences and classes of sentences” (Popper [1989] 1994, 82).

The reference given is: “Falsifizierbarkeit, zwei Bedeutungen von”, pp. 82–86 in Helmut Seiffert and Gerard Radnitzky, Handlexikon zur Wissenschaftstheorie, 2nd edition München:Ehrenwirth GmbH Verlag.

Thanks!

>What's a good counterexample?

The SEP article linked by the sibling comment outlines astrology as an example of bad science and a review of articles one year in Nature is good science being rejected by Popper's criterion. It's rather absurd (and a comment to how far HN reads into things) that my comment is at -1 votes.

It is impossible to ever prove or disprove a theory, so falisfiability is a red herring.

For a theory T and observations O, you can have 4 possible scenarios. Where "!" indicates "not":

Starting from theory

  Modus Ponens            : T  therefore O
  Denying the Antecedant  : !T therefore !O

Starting from observation

  Affirming the Consequent: O  therefore T
  Modus Tollens           : !O therefore !T

Denying the antecedent and affirming the consequent are both invalid if other theories may be consistent with the same observations (which is always the case), leaving us with Modus Ponens and Modus Tollens as valid forms of reasoning.

What this means is that we can deduce:

  1. what we should observe if our theory is really true (Modus Ponens)
  2. a theory is not true if we fail to observe what it predicts (Modus Tollens)

Modus Ponens reasoning is used to derive testable predictions from a theory, and then Modus Tollens reasoning is used when checking if the theory predicts the correct observations.

However, the relationship between theory and prediction P is not so simple. It is always the case that other assumptions A must be made along with the theory. These assumptions can be as simple as "the equipment is functioning properly", but can get much more complicated.[1] I.e.:

  (T AND A) entails P

If we fail to observe the prediction P, then the entire left side gets negated:

   !P entails !(T AND A)

This is equivalent to saying either T or A is incorrect:

   !P entails !T OR !A

So even in the best case scenario, you can never know if it is your theory that is wrong or some other assumption you making is wrong.

This should tell us the real value of science lies somewhere else besides falsifiability, e.g. in making useful or otherwise surprising predictions.

[1] https://en.wikipedia.org/wiki/Duhem%E2%80%93Quine_thesis

tl;dr- They're arguing that there's little point to building a new, larger particle collider to succeed the Large Hadron Collider (LHC) because, while there are hypotheses that such a new collider could falsify, those hypotheses aren't sufficiently motivated to be worth falsifying.

Thank you.

It would have been useful, at least to me, if that had been the first paragraph. I gave up reading before I got that far because it seemed to be an article attacking a straw man.

Thanks. The title could use a rewrite.

So I take it the author is well-read in high-energy particle physics to make this claim?

Yes, the author is well-read in particle physics.

https://en.wikipedia.org/wiki/Sabine_Hossenfelder

No amount of experimentation can ever prove me right; a single experiment can prove me wrong. - Einstein

Science is the organized skepticism in the reliability of expert opinion. - Feynmann

Proper link: http://backreaction.blogspot.com/2019/01/just-because-its-fa...

Good article. So many armchair "physicists" in the comments it's hilarious.

I'm rather confused by how many got thrown by her comment about dark matter & assuming an unspecified "fluid", somehow they didn't realize she was just making a reference to the formulation of the original definition of "dark matter".

I think this woman has a crusade against current modern physics, and only in this kind of websites resonates. Wonder why.

Eh, no. She's an established theoretical physicist. She explains her motivations here:

https://backreaction.blogspot.com/2019/02/particle-physicist...

Reply on news.ycombinator.com