A new epistemological tool more powerful than falsifiability and Occam's razor

This video uses a highly edited audio file taken from this TED Talks source:

David Deutsch: A new way to explain explanation
http://www.youtube.com/watch?v=folTvNDL08A

I claim fair use.

I want to talk about Deutsch's ideas. I think his concept of "invariability" might turn out to be an epistemological tool more powerful than Occam's razor and Karl Popper's concept of falsifiability combined.

Occam's razor states that "entities must not be multiplied beyond necessity" or "plurality should not be posited without necessity," and that the simplest explanation tends to be the best one. That means that when competing hypotheses are equal in all other respects, you should choose the explanation that makes as few assumptions as possible, and eliminate those that make no difference in the observable predictions of the explanatory hypothesis or theory. They used to call it the law of parsimony, or law of economy or or law of succinctness.

Karl Popper next argued that the reason for us to prefer simple theories wasn't about appealing to practical or aesthetic considerations, as Occam himself seemed to have assumed. Popper justified simplicity by connecting it to his falsifiability criterion because simple theories had more empirical content and were more testable.

And in the end, Occam's razor is just a heuristic, a rule of thumb, and not really a law or an irrefutable principle of logic while falsifiability is a must for any scientific theory.

And just as Occam's razor is a side effect of the truth of falsifiability, falsifiability is a side effect of invariability.

That may not be the best way to state things, perhaps you can do better, please discuss in the comments section.

Category:

Education

Tags:

• David Deutsch
• Occam's razor
• falsifiability
• testability
• science
• epistemology
• philosophy
• philosophy of science
• Karl Popper
• theory
• hypothesis
• explain explanation
• explaining explanation

[Psychosmurf547]

Look up Bayes' Theorem and the Transferable Belief Model, etc. This is nothing new.

[zarkoff45]

@Psychosmurf547

I'll look into that eventually, but I have to say it sounds a bit off the wall since Deutsch's ideas imply more of a Popperian epistemology rather than Bayesian epistemology.

Deutsch's main point was that good explanations are "hard to vary" and I've not seen that idea expressed in a Bayesian epistemology.

[mm1979dk]

I agree with your interpretation of the talk, however I think you are overstretching the importance of "invariability" contribution. Variability is just the opposite of falsifiability -- the more degrees of freedom one allows, the harder it is to falsify.

Also falsifiability is attributed to Karl Popper, but the principle itself has been used in science long before him, even Charles Darwin said that one needs to find a single non-conforming species to falsify the theory of evolution.

[zarkoff45]

@mm1979dk "...opposite of falsifiability -- the more degrees of freedom one allows, the harder it is to falsify."

But falsifiability applies to predictions, not explanations. Invariability is about explanations and it doesn't necessarily contribute predictions.

"...the principle itself has been used in science long before him..."

Indeed. It seems most philosophers of science are just describing what scientists have been doing, not really telling them what to do.

[SisyphusRedeemed]

Interesting talk! I do have to wonder if it might not undermine lots of 'genuine' scientific explanations, though. The Cartesian/Newtonian debate about 'action at a distance' is an example. Newton's laws didn't actually EXPLAIN anything, they simply described. Some philosophers of science suggest that this is all science should do, just describe reality and abandon explanation entirely. This kind of instrumentalism creates havoc for quantum mechanics. Should we just 'shut up and calculate?'

[zarkoff45]

@SisyphusRedeemed "Newton's laws didn't actually EXPLAIN anything, they simply described."

And we now have AIs that can do what Newton and Galileo did faster than they could. Google: "Eureqa, Software to Replace Scientists"

Eureqa how to:

watch?v=NhC1Qb-PQ5Q

"Should we just 'shut up and calculate?'"

That's going to be a very important question now that we have Eureqa. There is no agreed upon explanation for quantum mechanics and yet it still works.

[MrShysterme]

@zarkoff45 I believe that falsification in a bayesian framework is seen as strong evidence against one hypothesis. So, perhaps falsification is a special case of bayesian analysis.

[GargamelGold]

@zarkoff45 He did a pretty good job. I never thought of scientific inquiry in that way.

[Dummweltschutz]

If you have a bad explanation/theory you can't just rewrite it, you'll have to do more research. if the greeks knew there was the tilted axis MAYBE they'd have rewote it, but after that, they'd have found more evidence of not demether causing the season change -and so on, until they got explanatipon we have.

[Psychosmurf547]

@zarkoff45 Sorry, I was trying to remember where I saw a concept very similar to invariance, I just can't remember what it's called now.

[17secret]

I'd ask for instance, what is a force, what is gravity, what is light..many concepts/facts stay without a real physical definition as far as i can remember, a force has always been an arrow on my paper, light is a wave or sometimes a particle, and gravity is something magic or is it a magic graviton??

We only create fragile 'functional' definitions that are not the real stuff, and it is very hazardous to try to build a so-called empirical world on such striking inaccuracies and fallacies.

[Torcika]

The writing interferes with the voice.hard to concentrate.should have thought it a little better