Occam's razor in science: all-time practice or modern fashion?

[Aidyan]

When I went to college, more than 30 years ago, as far as I can remember, nobody was talking about the use of Occam's razor in science. Reading the works of past scientists, I rarely see them invoking principles of parsimony, let alone cite Occam's razor. Yes, Newton, Einstein, and few others sometimes hinted at principles of parsimony, but overall, I don't find them talking about "Occam's razor". Nowadays, however, many consider it a cornerstone of science or a fundamental methodological rule that everyone is supposed to apply as if this were the case since the times of the inception of science. Even though the principle dates back to Aristotle and has been applied throughout the history of science, it seems to me that, at least in physics, it had no or scarce relevance until very recently. My understanding is that it has become nowadays something that we like to present as an inherent principle of science. But that was not always the case.

However, since this is only my personal subjective perception, and I have no historical research backing this claim, I'm wondering if Occam's razor has always been an all-time practice in science, or if it has become a modern fashion?

 

[Dale]

Scientific practice does indeed change over time. For example, in the early 1900’s it was not even common to report error bars or uncertainty. Now it is effectively mandatory. It is not as though earlier scientists didn’t know about uncertainty, it just became more prevalent in scientific writing and more central in scientific thinking. Same with Occham’s razor.

By the way, in my opinion both Occham’s razor and Popper’s falsifiability are properly subsumed in Bayesian inference

 

[Bystander]

Pooper’s

?

 

[Motore]

?

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

(I hope you are not fixated on a small spelling error from @Dale)

 

[Dale]

No problem, it is a pretty funny spelling error!

 

[fresh_42]

By the way, in my opinion both Occham’s razor and Popper’s falsifiability are properly subsumed in Bayesian inference.

And all three suffer from the same aspect of vulnerability: mostly isn't always, Popper is philosophy and hardly applies to scientific practice, and Bayesian is more of a personal opinion. They all cannot be decided in the end.

 

[Dale]

And all three suffer from the same aspect of vulnerability: mostly isn't always, Popper is philosophy and hardly applies to scientific practice, and Bayesian is more of a personal opinion. They all cannot be decided in the end.

I am not sure what you mean by “mostly isn’t always” as a vulnerability nor am I sure how Bayesian inference suffers from it. Can you clarify specifically regarding Bayesian inference?

 

[fresh_42]

I am not sure what you mean by “mostly isn’t always” as a vulnerability nor am I sure how Bayesian inference suffers from it. Can you clarify specifically regarding Bayesian inference?

With "mostly isn't always" I meant Occam's razor that "declares" the most probable answer the correct answer.

It has been for too long now to remember it in detail. I only remember that the discussion about Bayesianism took place, mainly between Dr. Brooks Fereby and Prof. Dr. Dinges at the stochastic department when I was studying. I still consider the two sides of the discussion as a worldview (Weltanschauung) rather than a decidable problem. I also took a course in decision theory in economics, and they used the concept of subjective probability which at least in principle would have been to be measured by experiment.

 

[Bystander]

No problem, it is a pretty funny spelling error!

Agreed, totally.

 

[Dale]

With "mostly isn't always" I meant Occam's razor that "declares" the most probable answer the correct answer.

Hmm, maybe it translates different in German. In English the sense is that the simplest answer that matches the facts is the best answer. Here “best answer” is more like “preferred” or “most useful” rather than “correct”.

It has been for too long now to remember it in detail. I only remember that the discussion about Bayesianism took place, mainly between Dr. Brooks Fereby and Prof. Dr. Dinges at the stochastic department when I was studying.

Ok, without details I cannot comment on a specific argument.

The nice thing about Bayesian inference in this context is that if you have a given set of data and two competing models the more complicated model will always fit the data closer. The Bayesian inference approach allows you to decide in principle when the closer fit is worth the additional complexity of the model. In other words, it tells you when to use Occham’s razor and when to discard it, at least in principle.

 

[fresh_42]

The nice thing about Bayesian inference in this context is that if you have a given set of data and two competing models the more complicated model will always fit the data closer. The Bayesian inference approach allows you to decide in principle when the closer fit is worth the additional complexity of the model.

There are two sides fighting in me. My nature is probably Bayesian. E.g., if I watch documentaries about airline crashes I tend to make a quick decision about the major cause. It usually turns out to be right, but, of course, there are many more reasons that all together led to the catastrophe. I usually simply get the initial fail. I think Bayesianism is similar, if it may be compared at all: you get the major cause but not the complete picture.

Maybe I didn't grasp Bayesianism in its entirety. To me, it looks like a personal decision rather than an objectifiable quality. It remains philosophy.

 

[apostolosdt]

[...] Nowadays, however, many consider it a cornerstone of science or a fundamental methodological rule that everyone is supposed to apply as if this were the case since the times of the inception of science. [,,,] My understanding is that it has become nowadays something that we like to present as an inherent principle of science. But that was not always the case. [...]

Interesting comments. Could you please cite some concrete references from modern literature to support them?

 

[BWV]

Isn’t frequentism just another Bayesian prior?

And all three suffer from the same aspect of vulnerability: mostly isn't always, Popper is philosophy and hardly applies to scientific practice, and Bayesian is more of a personal opinion. They all cannot be decided in the end.

There will always be a philosophical underpinning to science, those who claim to be free from philosophy are just naive about their own philosophical views. paraphrasing Keynes, practicitioners who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct philosopher

 

[PeroK]

Isn’t frequentism just another Bayesian prior?

No, it isn't!

 

[Aidyan]

Interesting comments. Could you please cite some concrete references from modern literature to support them?

That was my question (read to the end). If there is historical research that highlights if and when it was more or less in fashion? From my interaction with scientists, especially when it comes to issues of philosophical nature, nowadays, they tend to resort much more frequently than in the past to principles of parsimony to defend their ideas. But I'm willing to discover otherwise...

 

[apostolosdt]

That was my question (read to the end). If there is historical research that highlights if and when it was more or less in fashion? From my interaction with scientists, especially when it comes to issues of philosophical nature, nowadays, they tend to resort much more frequently than in the past to principles of parsimony to defend their ideas. But I'm willing to discover otherwise...

OK, but is someone aware of any published paper in an indexed, peer-reviewed journal, in which the authors claim their proposed model has been selected because it is supported by its inherent simplicity?

Whenever I come across words like `philosophy' and its derivatives in a text dealing with science, I become very cautious. I admit I understand nothing about philosophy and I strive really hard to understand a bit of physics. But then, it may be only me.

 

[Vanadium 50]

Let me start with a quote by, I think, Feynman: "Scientists need philosophers of science like birds need ornithologists".

Occam is fine for Monday morning quarterbacking, but in the heat of the moment, it is often hard to tell what "simplest" is. To take an example from my own field, there were two particles of identical mass, lifetime and spin, but opposite parity: one decayed to two pions, the other to three. Some possible explanations:

1. There really are two particles, and by coincidence they have identical mass and lifetimes to within our ability to measure.
2. There is only one particle, and parity is not conserved in its decay - but is conserved everywhere else.
3. There are two particles, but some unknown new physics, most likely a symmetry principle, forces them to have identical mass and lifetime.

Which is the simplest? I would argue #1. It's also the least predictive. The correct answer is #2. Most people at the time expected #3 to be the solution. Particularly because people were finding new exact and approximate symmetries all the time

 

[BWV]

Occam is trivial, not controversial and embodied in the fact that existing models are first used to attempt to explain some new observed phenomena, with the creation of new models as a last resort

Feynman’s core belief that the scientific method alone, without deference to some larger political or religious ideology, is sufficient to discover truths about the natural world, is a philosophical position

 

[PeroK]

Feynman’s core belief that the scientific method alone, without deference to some larger political or religious ideology, is sufficient to discover truths about the natural world, is a philosophical poposition.

I've never trusted that view. It's like saying atheism is a religion.

Or, that not having a strategy is itself a strategy.

 

[Vanadium 50]

Not to decide is to decide.

 

[BWV]

I've never trusted that view. It's like saying atheism is a religion.

Or, that not having a strategy is itself a strategy.

No it’s like saying atheism is a philosophical position, although I get that creationists like to play this card

 

[PeroK]

Not to decide is to decide.

It's common practice in government IT projects to decide not to make a decision. That's for sure!

 

[Dale]

Not to decide is to decide.

if you choose not to decide you still have made a choice. -Rush

 

[BWV]

And one cannot define the concept of a scientific theory without resorting to philosophy

The philosophy of science attempts to define the rules of the game, and while too much ink may get spilled over trivial controversies , it’s a worthwhile endeavor.

If every HS student had to take a semester course on just the scientific method, without any actual science content, perhaps the public discourse over policy issues like vaccines or nuclear power would improve

 

[Dale]

And one cannot define the concept of a scientific theory without resorting to philosophy

I think that we can just define a word however we like, and merely having a dictionary doesn’t require one to have a philosophy.

 

[BWV]

I think that we can just define a word however we like, and merely having a dictionary doesn’t require one to have a philosophy.

Only if the definition is arbitrary and does not reflect a consensus of hundreds of years of debate within a culture

The definition of Marxism is equally arbitrary and absent of philosophy by that logic

 

[Dale]

Only if the definition is arbitrary and does not reflect a consensus of hundreds of years of debate within a culture

So then any consensus is the result of philosophy? Having studied philosophy, my experience indicates the opposite. In my studies it seemed that philosophers never reached a consensus. Observation of consensus in a field seems to be evidence against the importance of philosophy to that field.

Sorry, but this, IMO, is attempting to simply label something ubiquitous as “philosophy” and thereby claim that philosophy is ubiquitous.

 

[BWV]

So then any consensus is the result of philosophy? Having studied philosophy, my experience indicates the opposite. In my studies it seemed that philosophers never reached a consensus. Observation of consensus in a field seems to be evidence against the importance of philosophy to that field.

Sorry, but this, IMO, is attempting to simply label something ubiquitous as “philosophy” and thereby claim that philosophy is ubiquitous.

So you are arguing there is no philosophical content behind the concept of a scientific theory?

The idea that individuals have certain rights is ubiquitous (at least in this culture) but represents a particular philosophical position

 

[Dale]

So you are arguing there is no philosophical content behind the concept of a scientific theory?

I am saying that the philosophy is unimportant.

The valuable scientific practices that we use today are a result of collective experience, not a conclusion of some philosophical debate. The things that need to be changed in scientific practice are also known to be problematic as a result of experience, not as a conclusion of some philosophical debate.

If scientists waited for philosophers to come to some consensus, then we would never do any science. And yet, we do science. Therefore the philosophy is unimportant.

 

[BWV]

I am saying that the philosophy is unimportant.

The valuable scientific practices that we use today are a result of collective experience, not a conclusion of some philosophical debate. The things that need to be changed in scientific practice are also known to be problematic as a result of experience, not as a conclusion of some philosophical debate.

If scientists waited for philosophers to come to some consensus, then we would never do any science. And yet, we do science. Therefore the philosophy is unimportant.

Yes, philosophy is more prevalent in politics. This is one reason that there is so little consensus in politics.

There is consensus in politics around philosophical proportions there is consensus on - like a certain basic level of human rights. You are presenting a different argument than I am making - if the philosophy of science is unimportant to doing science (with which I would agree ) it is because the philosophical work was done centuries ago and has a wide consensus among scientists today. The philosophy of science serves an important and useful role today not within science, but framing what is unique about scientific knowledge it’s boundaries and limits - distinguishing science from either pseudoscience or scientistism, for example

 

[Dale]

The philosophy of science serves an important and useful role today not within science, but framing what is unique about scientific knowledge it’s boundaries and limits - distinguishing science from either pseudoscience or scientistism, for example

So the philosophy of science is as important as a few definitions in a dictionary. Ok, I guess that isn’t 0 importance, but I still feel ok characterizing that as “unimportant”.

 

[BWV]

So the philosophy of science is as important as a few definitions in a dictionary. Ok, I guess that isn’t 0 importance, but I still feel ok characterizing that as “unimportant”.

The philosophy of science is only unimportant until some ideology starts encroaching on it.

Given the attacks from both ends of the political spectrum today, it’s worth some thought and attention

 

[Dale]

The philosophy of science is only unimportant until some ideology starts encroaching on it.

Given the attacks from both ends of the political spectrum today, it’s worth some thought and attention

And should the philosophers of science ever actually successfully protect the institutions of science from politicians, then I will consider that to be strong empirical evidence against my current position that it is unimportant.

 

[apostolosdt]

There will always be a philosophical underpinning to science, those who claim to be free from philosophy are just naive about their own philosophical views. paraphrasing Keynes, practicitioners who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct philosopher [...]

That's just a personal opinion and little can be done to refute opinions. The statement `The Earth is flat' is a worse opinion, but at least it is testable.

I now appreciate the comments by moderators in a post nearly ten months old and titled "Curious, Why No Philosophy Chat Allowed?".

 

[apostolosdt]

And one cannot define the concept of a scientific theory without resorting to philosophy. [...]
If every HS student had to take a semester course on just the scientific method, without any actual science content, perhaps the public discourse over policy issues like vaccines or nuclear power would improve

That last remark about scientific methods vs. science content caught my attention. Could you kindly elaborate on that? For I have been teaching such subjects for years and the syllabuses covered both.

And, please, don't just cite what the terms mean, we all know that.