Occam's razor is unjustified, so why accept it?

frank December 09, 2022 at 20:18 7150 views 49 comments
Occam's razor says that if we have a choice between a simple answer and a compound one, we should pick the simple one.

It's widely accepted even though it actually has no justification. It's acceptance seems to come down to its intuitive or aesthetic appeal. Is that enough? Or should we just reject it?

Comments (49)

Banno December 09, 2022 at 20:28 #762269
Of corse it's just an aesthetic preference.

Preferring something for aesthetic reasons is a justification...

Hence Occam's Razor is justified.
Benj96 December 09, 2022 at 21:12 #762285
Quoting frank
Occam's razor says that if we have a choice between a simple answer and a compound one, we should pick the simple one.

It's widely accepted even though it actually has no justification. It's acceptance seems to come down to its intuitive or aesthetic appeal. Is that enough? Or should we just reject it?
32m


Occams razor is based on the fact that truth/reason for something is often less extravagant, requires less reasoning, imagination and hop-scotching around than semi truths, or downright lies do.

The only thing you need to catch a liar is a rigorous enquiry, because to construct a continuous alternative narrative takes a lot of weighing, measuring, reasoning and accounting to prevent paradoxes and contradictions from revealing your lie.

The truth on the other hand is easy. It's natural and it sticks to basic straightforward path. It's not creative. It's factual.

frank December 09, 2022 at 21:14 #762286
Quoting Banno
Preferring something for aesthetic reasons is a justification...


But simplicity and complexity are equally appealing. One can be just as beautiful as the other. There's no accounting for taste.

This is Islamic:

User image

Scandinavian:

User image


T Clark December 09, 2022 at 21:16 #762288
Reply to frank

Quoting Wikipedia
Occam's razor, Ockham's razor, or Ocham's razor (Latin: novacula Occami), also known as the principle of parsimony or the law of parsimony (Latin: lex parsimoniae), is the problem-solving principle that "entities should not be multiplied beyond necessity".[1][2] It is generally understood in the sense that with competing theories or explanations, the simpler one, for example a model with fewer parameters, is to be preferred. The idea is frequently attributed to English Franciscan friar William of Ockham (c.? 1287–1347), a scholastic philosopher and theologian, although he never used these exact words. This philosophical razor advocates that when presented with competing hypotheses about the same prediction, one should select the solution with the fewest assumptions,[3] and that this is not meant to be a way of choosing between hypotheses that make different predictions.


This makes sense to me.
frank December 09, 2022 at 21:20 #762290
Quoting Benj96
The truth on the other hand is easy. It's natural and it sticks to basic straightforward path. It's not creative. It's factual.


If you look at the way your body maintains your blood pressure, it's pretty complex.

User image

Are you saying that because this answer is complex, it must be wrong?
Banno December 09, 2022 at 21:21 #762291
Reply to frank Indeed. It is a preference, and dependent on circumstance.

Reply to T Clark Methodologically the hypothesis with fewer assumptions is easier to work with. But it is not thereby true. So choosing the simplest hypothesis is an expression of an aesthetic favouring laziness....
frank December 09, 2022 at 21:22 #762293
Quoting Banno
Indeed. It is a preference, and dependent on circumstance.


So why the appeal of Occam's razor? What's the draw?
Benj96 December 09, 2022 at 21:32 #762299
Quoting frank
Are you saying that because this answer is complex, it must be wrong?


No of course not. But biologists and physiologists occamed it up to isolate each component and see what happens when it's removed, or more is added, or what it reacts with and what they make.

In essence they deconstructed blood pressure into its individual parts so they could build the full picture of how it works in its entirety.

That isn't to say the full complex compound answer isn't correct but it's much harder to jump to that conclusion than to take a step by step approach.

Also, if you think the baroreceptor reflex is complex, there's the hormones, fluid volume/electrolyte balance as well to factor in before blood pressure reveals all of its cogs and wheels.
T Clark December 09, 2022 at 21:40 #762304
Quoting Banno
Methodologically the hypothesis with fewer assumptions is easier to work with. But it is not thereby true.


Agreed. A lot would depend on the assumptions used and their relative plausibility.

Quoting Banno
So choosing the simplest hypothesis is an expression of an aesthetic favouring laziness....


There's more to it than that. The more assumptions, i.e. unproven data inputs, the more likely one of them is wrong.
Banno December 09, 2022 at 21:42 #762308
Quoting frank
So why the appeal of Occam's razor?


Why not appeal of Occam's razor? Pick the pretty flower, pick the short hypothesis.
T Clark December 09, 2022 at 21:43 #762309
Quoting frank
Are you saying that because this answer is complex, it must be wrong?


This is a hay man or straw dog, or whatever you call it. It has nothing to do with the complexity of the system being described. it's the complexity of the unjustified inputs.
Banno December 09, 2022 at 22:07 #762317
Quoting T Clark
The more assumptions, i.e. unproven data inputs, the more likely one of them is wrong.


Ah... an argument from statistics.

On the other hand, the hypothesis with the most assumptions is the most falsifiable. If, the more assumptions, the more likely that one of them is wrong, and if we ought prefer falsifiable hypotheses, we ought prefer the more complex ones. Hence falsificationists ought reject parsimony. :wink:

But further, determining the number of assumptions is a question of interpretation, subject to how the hypothesis in question is expressed. For example, is that this thread is in English one of the assumptions of this argument? Where do we draw the line between what is a relevant hypothesis and what isn't?

That is, even expressed in statistical terms, the preference is aesthetic.

frank December 09, 2022 at 22:15 #762321
Quoting Banno
Why not appeal of Occam's razor? Pick the pretty flower, pick the short hypothesis.


I guess as long as that's the spirit in which people embrace Occam's razor, it's ok. Like: eeny meeny miny moe, pick the theory whose complexity is low.
Banno December 09, 2022 at 22:19 #762325
Reply to frank Cute. :grin:

Edit: We might add that it is worth noting that the choice of hypothesis is not final; we can modify that choice based on further data. So if a prettier flower comes along, we can drop the old one and pick the new one. Parsimony is one part of a method that involves ongoing interaction with each other and with the world, not the final determinate.
frank December 09, 2022 at 22:26 #762328
Quoting Benj96
In essence they deconstructed blood pressure into its individual parts so they could build the full picture of how it works in its entirety.


Is that the approach to things that works best for you? Breaking things down to simple parts?

The cardiovascular system needs to looked at as a whole because it's self regulating, as if each part is performing a duty to the whole. If you get lost in the details, you could miss the awesomeness of the whole thing. Maybe that's my aesthetic preference?

Never thought of it that way.

T Clark December 09, 2022 at 22:33 #762332
Quoting Banno
That is, even expressed in statistical terms, the preference is aesthetic.


I don't disagree that using statistical reasoning is not a strong argument.
Banno December 09, 2022 at 22:36 #762334
Reply to T Clark There are two sorts of responses to statistical arguments. The one looks at the quantity of data and the complexity of the analysis and thinks "Gee, this must be important". The other thinks "Fuck, here we go again..."
frank December 09, 2022 at 22:38 #762338
Quoting Banno
We might add that it is worth noting that the choice of hypothesis is not final; we can modify that choice based on further data. So if a prettier flower comes along, we can drop the old one and pick the new one. Parsimony is one part of a method that involves ongoing interaction with each other and with the world, not the final determinate.


I think I've seen it used as a reason to discard a thesis. I think you agree that it doesn't provide a justification for that. It's a bad idea to play favorites prior to getting experimental results. If you can't experiment, you have to be satisfied with not knowing.
T Clark December 09, 2022 at 22:39 #762339
Quoting Banno
There are two sorts of responses to statistical arguments. The one looks at the quantity of data and the complexity of the analysis and thinks "Gee, this must be important". The other thinks "Fuck, here we go again..."


User image
Banno December 09, 2022 at 22:45 #762342
Reply to frank Yep. And admitting to not knowing is a good thing to do.
frank December 10, 2022 at 00:30 #762393
Reply to Banno Very true
Benj96 December 10, 2022 at 12:26 #762506
Quoting frank
Is that the approach to things that works best for you? Breaking things down to simple parts?

The cardiovascular system needs to looked at as a whole because it's self regulating, as if each part is performing a duty to the whole. If you get lost in the details, you could miss the awesomeness of the whole thing. Maybe that's my aesthetic preference?

Never thought of it that way.


To be frank, Frank, I think it's equally important to view it as a holistic system and to explore its individual parts in isolation.

Just as its important to review the performance of a car as a unit - all of its functions used in test drives, as it is to examine each part: the brake system, the engine, the catalytic converter, chassis, aesthetic features, electronics etc.

To know something is to know how it behaves as a compound thing, as well as to understand the relationships between its individual components. If you dismantle a car and then put it back together, you're likely to "know what a car is" better than someone who has just driven a lot of them.

Both have their place in the knowledge of any subject.
It's a matter of scope. Specialists have a narrow particular scope and expertise in one area whilst others deal with the holistic/general overview.
Both are required to explore the knowledge of any discipline.

Occams razor is a useful approach. But it only elucidate part of the information.
frank December 10, 2022 at 16:22 #762554
Quoting Benj96
To know something is to know how it behaves as a compound thing, as well as to understand the relationships between its individual components. If you dismantle a car and then put it back together, you're likely to "know what a car is" better than someone who has just driven a lot of them.


Occam disagreed, sort of. He believed that there are no "compound things.". He would say the "car" only exists as an idea that humans use to group things for their purposes. There is no car out there in the world, only isolated, individual things. He was a kind of atomist, or proto-nominalist.

This is the basis of his belief that explanations must be as simple as possible. If you start explaining things in complex terms, using compound objects, you're really just off in the realm of imagination, not describing the world as it is.

Do you agree with that?
Isaac December 10, 2022 at 18:54 #762589
Quoting frank
it actually has no justification.


Yet...

Quoting frank
the basis of his belief that explanations must be as simple as possible. If you start explaining things in complex terms, using compound objects, you're really just off in the realm of imagination, not describing the world as it is.


frank December 10, 2022 at 19:43 #762631
Mww December 11, 2022 at 12:18 #762856
Reply to frank

Occam’s Razor: the principle that says the fewer ways there are to make a mistake, the easier it is to correct it.

I’d accept that principle.
frank December 11, 2022 at 14:03 #762876
Reply to Mww

KISS?
Mww December 11, 2022 at 14:21 #762879
Reply to frank

Yep. But be careful; I hear Gene and the guys are particularly defensive regarding their brand.
frank December 11, 2022 at 15:04 #762885
Reply to Mww

:grin:
Manuel December 12, 2022 at 15:20 #763138
Well, I suppose that arguing, instead of Occam's razor per se, that one should present a hypothesis or theory in the simplest available manner is better than presenting such information in a convoluted or inflated way.

One can say that gravity pulls apples to the ground.

Or one can say that given the universe we are an in, and the planet we find ourselves in, plus the properties of apples all combine such that it follows, that in the vast majority of circumstances, gravity on Earth pulls apples to the ground given ordinary conditions, because a hurricane might complicate the process.

Both are true, one is simpler. But sometimes we cannot simplify more than we'd like.
frank December 12, 2022 at 15:38 #763142
Quoting Manuel
Well, I suppose that arguing, instead of Occam's razor per se, that one should present a hypothesis or theory in the simplest available manner is better than presenting such information in a convoluted or inflated way.


I don't know, Schopenhauer said Kant intentionally obscured some of his writing to avoid criticism from the church. And was it Derrida who supposedly said that the only way to make it as a French philosopher is to convolute your writing?



Manuel December 12, 2022 at 15:46 #763144
Reply to frank

It's likely, or at least it wouldn't surprise me. But one can explain the basics of Kant without much trouble.

That was Foucault. Derrida claimed that he never fell into that temptation to write more obscurely for the sake of profundity. Clearly, he wasn't being honest.
frank December 12, 2022 at 16:01 #763146
sime December 12, 2022 at 17:10 #763157
In science, and especially data science and machine learning, Occam's razor is often misunderstood to be an a priori principle. This can encourage biased and erroneous inductive inferences, typically in cases of Bayesian model selection or Bayesian averaging with respect to a family of different theories, where the 'prior' confidence assigned to the predictions of a particular theory is taken, without justification, to be inversely proportional to it's 'description length'.

The above principle can only be applied non-controversially when a supplementary argument is given to justify why the theories are described in the way they are, for otherwise the description lengths assigned to each candidate theory is arbitrary. E.g a diagonal straight line is only 'simpler' than a diagonal sine wave when the coefficients of both lines are given in terms the Standard Basis corresponding to the Cartesian axes. But the opposite is true when both lines are described in terms of a Fourier basis.

Quoting Manuel
Well, I suppose that arguing, instead of Occam's razor per se, that one should present a hypothesis or theory in the simplest available manner is better than presenting such information in a convoluted or inflated way.


Which goes towards explaining what Occams razor actually is; the principle of Occam's razor is our post-hoc revision of our linguistic conventions in response to our observations, so that our language encodes our most validated theories as efficiently as possible. Occam's razor shouldn't be mistaken for an a priori principle of inference, rather it should be understood to be a prescription for revising our linguistic conventions so that our past-conditioned expectations are easier to communicate and describe.

Manuel December 12, 2022 at 18:09 #763184
Reply to sime

I see what you are saying, I'd add that it's not linguistics solely, but also conceptual. By expressing ourselves in a clear and concise manner, the information or data we are presenting is more easily understood than in some other, more technical or obscure manner.

Understanding is not limited to language, I don't think. But, point taken.
creativesoul December 14, 2022 at 06:42 #763659
Occam's razor is about reducing the likelihood for error. The fewest unprovable assumptions is best. The fewest entities is best.

The hitch seems to have been forgotten though...

...so long as there is no loss in explanatory power, the simplest explanation is the best.
creativesoul December 14, 2022 at 06:42 #763660
It's not just about aesthetics. It's about methodological approach. It's about warrant. It's about further discriminating between competing explanations.
Bylaw December 14, 2022 at 16:06 #763821
Reply to frank It's not an ontological claim, it's a methodological suggestion. If we have two explanations and both work, we might as well use the simpler one...that's just easier. But no one has to follow this suggestion.
frank December 14, 2022 at 16:44 #763843
Quoting Bylaw
It's not an ontological claim, it's a methodological suggestion. If we have two explanations and both work, we might as well use the simpler one...that's just easier. But no one has to follow this suggestion.


Sounds reasonable.
Down The Rabbit Hole December 15, 2022 at 00:00 #764003
Reply to creativesoul

Quoting creativesoul
Occam's razor is about reducing the likelihood for error. The fewest unprovable assumptions is best. The fewest entities is best.

The hitch seems to have been forgotten though...

...so long as there is no loss in explanatory power, the simplest explanation is the best.


It makes sense that the fewer barriers to something being true, the more likely it is to be true.

It is commonly used against belief in god, but I don't see how there are any barriers to something that has no cause.
creativesoul December 15, 2022 at 02:58 #764020
Quoting Down The Rabbit Hole
It makes sense that the fewer barriers to something being true, the more likely it is to be true.


That does not make much sense to me. What sort of barriers are you referring to?

Occam's razor is commonly used against the explanation "God did it".
Outlander December 15, 2022 at 03:34 #764028
From the title of this post alone, all discovery becomes mute, annulled, voided. Truly a stark testament of the times in which we live where despite everyday use of things by those who could not reproduce said things themselves on request shamelessly use to profane the very notion of possibility itself. It truly boggles the mind.
Ansiktsburk December 15, 2022 at 05:46 #764035
On can chose not to accept it but solving anything non-trivial, one will bear it in mind. Was it Einstein that said “as simple as possible but not simpler”. Aesthetics sometimes is rather the opposite, guys wanting to make problems harder than they are because they like to solve intricate problems,
Down The Rabbit Hole December 15, 2022 at 13:33 #764100
Reply to creativesoul

Quoting creativesoul
That does not make much sense to me. What sort of barriers are you referring to?

Occam's razor is commonly used against the explanation "God did it".


As you've alluded to, Ockham's Razor has a qualification that "all things being equal" the simplest explanation is usually the correct one.

Let's say, just for the purpose of my argument, the evidence for a panpsychist and emergent view of consciousness is equal. There would be nothing getting in the way of panpsychism developing, as it has always been there, but for emergence there are barriers, such as the possibility that inanimate matter would never reach awareness, and further that consciousness would not be preferable for evolution (which many scientific tests are hinting at). There will be other barriers I can't even imagine to inanimate matter somehow becoming aware. It would be simpler to say it has always been there, and thus has no barriers to it becoming reality. The panpsychism has to have always been there for there to be symmetry with my argument about God.

All things being equal, God as always existing would have no more barriers, and is no less likely to exist than universe/s always existing.
creativesoul December 26, 2022 at 01:25 #766518
Reply to Down The Rabbit Hole

Perhaps, but it does invoke an extra entity.
Xanatos February 16, 2023 at 00:27 #781375
Reply to frank It's based in terms of probabilities, IIRC. So, the simplest hypothesis is not guaranteed to be true, but it is more likely to be true. In other words, if you were a betting man, then that is the way that you should be betting. At least as a general rule.
frank February 16, 2023 at 14:17 #781545
Quoting Xanatos
So, the simplest hypothesis is not guaranteed to be true, but it is more likely to be true


Why?
Agent Smith February 16, 2023 at 16:12 #781570
[quote=Numerius Negedius]It's quite simple really, he lied.[/quote]
bongo fury February 16, 2023 at 17:53 #781590