Deep - it aligns with more fundamental explanations of how the world works.
Broad - it can explain a range of phenomena of the same class.
Hard to vary - changing a small part of the explanation would change its predictions."
1. There's little mention of prediction or falsifiability. That's a rather extraordinary omission, compared to everyone else's philosophy of science. (The two are of course closely related, because failure to predict is the gold standard of falsification)
2. Almost everyone wants Depth .. ultimately correspondence to.reality itself. But it can't be measured directly .. that's possibly the major problem in scientific realism. And the reason philosophy of science tends to emphasise simplicity, in particular, is as a proxy for Depth/Correspondence. (Predictiveness is a virtue in its own right).
3. The substitution of Hard to Vary for Simple is another novelty of Deutsch's approach. However, it is not an improvement.
Scientific epistemology has a distinction between realism and instrumentalism. According to realism, a theory tells you what kind of entities do and do not exist. According to instrumentalism, a theory is restricted to predicting observations. If a theory is empirically adequate, if it makes only correct predictions within its domain, that's good enough for instrumentalists. But the realist is faced with the problem that multiple theories can make good predictions, yet imply different ontologies, and one ontology can be ultimately correct, so some criterion beyond empirical adequacy is needed.
Realism is "deep", instrumentalism is "shallow". Deutsch is on the realist side, because of his concern with Depth, lack of interest in prediction, and disdain for induction.
Now, in most forms of scientific epistemology, simplicity is key to obtaining realism over and above empirical adequacy. You can have a set of theories which all make good predictions, and the same predictions, but have different ontological interpretations, so they cannot all be correct representations of the reality.
There are multiple simplicity criteria, but not multiple truths. So you need the right simplicity criterion.Minimally, it needs to be able to single out a "best" theory from the candidates.
A theory is easy to vary if part of it isn't doing anything: that's a part that can be removed, or substituted. Removing redundant parts makes theories simpler, so picking hard to vary theories is picking simpler theories, all other things being equal. So the HTV criterion is a simplicity criterion. Bur the HTV criterion, while it removes N theories that have redundant parts, leaves M theories that don't -- and M is likely to be greater than one. So it doesn't hone in right on the truth -- admittedly a high bar! -- and is also less selective than most rivals.
Another problem is that Kludges, patches and epicycles, are not redundant!
A theory can be baroquely complex, yet consist entirely of elements that are doing useful work -- each epicycle in a Ptolemaic theory adds some accuracy.
In an era where nothing is what it seems, the use of Occam's razor to determine what the supposed truth and reality is, brings about a dangerous misconception, and those who use Occams razor in this manner will most definitely end up cutting themselves repeatedly.
Ah - that’s not what I meant. I’m saying that you can declare one explanation as being better than another, but the specific definition of “better” depends on the 2 explanations.
That shouldn’t stop you from being able to make a generalised statement
Thanks Martin! It was the first time I came across such a unique analogy and thought I must share it with my readers. Looking forward to reading more of your writing
Oct 8, 2023·edited Oct 9, 2023Liked by Zan Tafakari
A quote wrongly attributed to Einstein: "If you can't explain it simply, you don't understand it well enough," leads us to a well-intended, but most often misapplied concept. If we simplify enough, we will understand the problem.
Yet micro-solutions, ignoring the systems, don’t solve macro-problems.
Furthermore, a quote that Einstein did say, in response to a request for a simple explanation of his Nobel Prize winning physics solution, captures the challenge of systems thinking very well:
“If it could be summarized in a sentence, it wouldn't be worth the prize.”
What you hit on is the core challenge with Systems Thinking. Namely the difficulty in oversimplification when problems are truly wicked.
Thanks Michael! You're right - reductive thinking is often idolised as "clear" thinking but they're not strictly the same thing. I'll be sure to read your essay and let you know what I think
Thanks Dylan! Yes - I highly recommend adaptive markets. I don't agree with everything in the book, but he lays out some compelling ideas for sure. Happy reading and let me know what you think about it
I think of Occam’s razor in the modern context as a preference for parsimony. Leave out elements of your explanation whose inclusion does not increase explanatory power.
I think about Occam's Razor as a useful tiebreaker. It only really applies when I have two (or more) equally plausible explanations for a phenomenon; it's not a go-to measure but instead a final (potentially) swaying data point. It mainly works because if you need 20 things to be true for Y explanation but 2000 things to be true for X, it's exponentially more likely that Y is true simply due to more ways things can go wrong for X.
"Deutsch suggests that a good explanation is:
Deep - it aligns with more fundamental explanations of how the world works.
Broad - it can explain a range of phenomena of the same class.
Hard to vary - changing a small part of the explanation would change its predictions."
1. There's little mention of prediction or falsifiability. That's a rather extraordinary omission, compared to everyone else's philosophy of science. (The two are of course closely related, because failure to predict is the gold standard of falsification)
2. Almost everyone wants Depth .. ultimately correspondence to.reality itself. But it can't be measured directly .. that's possibly the major problem in scientific realism. And the reason philosophy of science tends to emphasise simplicity, in particular, is as a proxy for Depth/Correspondence. (Predictiveness is a virtue in its own right).
3. The substitution of Hard to Vary for Simple is another novelty of Deutsch's approach. However, it is not an improvement.
Scientific epistemology has a distinction between realism and instrumentalism. According to realism, a theory tells you what kind of entities do and do not exist. According to instrumentalism, a theory is restricted to predicting observations. If a theory is empirically adequate, if it makes only correct predictions within its domain, that's good enough for instrumentalists. But the realist is faced with the problem that multiple theories can make good predictions, yet imply different ontologies, and one ontology can be ultimately correct, so some criterion beyond empirical adequacy is needed.
Realism is "deep", instrumentalism is "shallow". Deutsch is on the realist side, because of his concern with Depth, lack of interest in prediction, and disdain for induction.
Now, in most forms of scientific epistemology, simplicity is key to obtaining realism over and above empirical adequacy. You can have a set of theories which all make good predictions, and the same predictions, but have different ontological interpretations, so they cannot all be correct representations of the reality.
There are multiple simplicity criteria, but not multiple truths. So you need the right simplicity criterion.Minimally, it needs to be able to single out a "best" theory from the candidates.
A theory is easy to vary if part of it isn't doing anything: that's a part that can be removed, or substituted. Removing redundant parts makes theories simpler, so picking hard to vary theories is picking simpler theories, all other things being equal. So the HTV criterion is a simplicity criterion. Bur the HTV criterion, while it removes N theories that have redundant parts, leaves M theories that don't -- and M is likely to be greater than one. So it doesn't hone in right on the truth -- admittedly a high bar! -- and is also less selective than most rivals.
Another problem is that Kludges, patches and epicycles, are not redundant!
A theory can be baroquely complex, yet consist entirely of elements that are doing useful work -- each epicycle in a Ptolemaic theory adds some accuracy.
Thanks for taking the time to really engage with this - I’ve learnt a lot from your comment. See my reply to the note you restacked
In an era where nothing is what it seems, the use of Occam's razor to determine what the supposed truth and reality is, brings about a dangerous misconception, and those who use Occams razor in this manner will most definitely end up cutting themselves repeatedly.
Well put!
Really good stuff, Zan.
I absolutely love what you said here —-
“An explanation should never be chosen just because it's simpler than the alternative. It should be chosen because it's better than the alternative.
Some things are just hard to explain, and Occam's razor might point us away from the more complicated, but better explanations of the world.”
—- that was wonderful.
And I think the example you gave of the planets and the planetarium, illuminated your point really well.
Thanks Michael! The planets vs planetarium analogy is one I've expanded from Deutsch's book - but the macroeconomics was all me ;)
"Better" needs to be broken down into measurable properties and actionable steps, otherwise you might as well say "believe in true theories".
Agree - but would you not say that those specifics are domain dependent and don’t detract from the use of the word “better” in general?
To push back just a bit - “better” is not the same as “true” as you suggest
If everything is domain specific, nothing is better in general.
Ah - that’s not what I meant. I’m saying that you can declare one explanation as being better than another, but the specific definition of “better” depends on the 2 explanations.
That shouldn’t stop you from being able to make a generalised statement
The definition depends on both, but only points to one being the worse?
Interesting stuff, always appreciate some David Deutsch. And thank you for the mention!
Thanks Martin! It was the first time I came across such a unique analogy and thought I must share it with my readers. Looking forward to reading more of your writing
A quote wrongly attributed to Einstein: "If you can't explain it simply, you don't understand it well enough," leads us to a well-intended, but most often misapplied concept. If we simplify enough, we will understand the problem.
Yet micro-solutions, ignoring the systems, don’t solve macro-problems.
Furthermore, a quote that Einstein did say, in response to a request for a simple explanation of his Nobel Prize winning physics solution, captures the challenge of systems thinking very well:
“If it could be summarized in a sentence, it wouldn't be worth the prize.”
What you hit on is the core challenge with Systems Thinking. Namely the difficulty in oversimplification when problems are truly wicked.
https://www.polymathicbeing.com/p/systems-thinking
Thanks Michael! You're right - reductive thinking is often idolised as "clear" thinking but they're not strictly the same thing. I'll be sure to read your essay and let you know what I think
Great piece Zan! Love David Deutsch’s work and the book Adaptive Markets is free on Audible so I look forward to reading that also.
Thanks Dylan! Yes - I highly recommend adaptive markets. I don't agree with everything in the book, but he lays out some compelling ideas for sure. Happy reading and let me know what you think about it
I think of Occam’s razor in the modern context as a preference for parsimony. Leave out elements of your explanation whose inclusion does not increase explanatory power.
Good stuff, Zan. Just to add:
I think about Occam's Razor as a useful tiebreaker. It only really applies when I have two (or more) equally plausible explanations for a phenomenon; it's not a go-to measure but instead a final (potentially) swaying data point. It mainly works because if you need 20 things to be true for Y explanation but 2000 things to be true for X, it's exponentially more likely that Y is true simply due to more ways things can go wrong for X.