dotsam

Relatedly, David Deutsch's "Simple refutation of the ‘Bayesian’ philosophy of science"

By ‘Bayesian’ philosophy of science I mean the position that (1) the objective of science is, or should be, to increase our ‘credence’ for true theories, and that (2) the credences held by a rational thinker obey the probability calculus. However, if T is an explanatory theory (e.g. ‘the sun is powered by nuclear fusion’), then its negation ~T (‘the sun is not powered by nuclear fusion’) is not an explanation at all. Therefore, suppose (implausibly, for the sake of argument) that one could quantify ‘the property that science strives to maximise’. If T had an amount q of that, then ~T would have none at all, not 1-q as the probability calculus would require if q were a probability.

Also, the conjunction (T₁ & T₂) of two mutually inconsistent explanatory theories T₁ and T₂ (such as quantum theory and relativity) is provably false, and therefore has zero probability. Yet it embodies some understanding of the world and is definitely better than nothing.

Furthermore if we expect, with Popper, that all our best theories of fundamental physics are going to be superseded eventually, and we therefore believe their negations, it is still those false theories, not their true negations, that constitute all our deepest knowledge of physics.

What science really seeks to ‘maximise’ (or rather, create) is explanatory power.

https://www.daviddeutsch.org.uk/2014/08/simple-refutation-of...

You don't keep searching until a point of "statistical confirmation". This implies you have arrived at an infallible truth. Instead you look for ways you could be wrong, and try to correct any errors you find.

For instance, if you guess 'all swans are white', you don't ever get "statistical confirmation" that your guess was right. When you eventually see a black swan, you find out you were wrong. Then it's time to come up with a new theory.

It is trivial to prove otherwise - AlphaZero move 37. After 4,000 years of gameplay (yes, it is that old!) we still didn't get this level of insight in its strategy.

Are you saying that AlphaZero contains knowledge that we can't understand, even in principle? It is somehow beyond science, beyond all explanation?

Why is search a better concept that creativity?

Search suggests a fixed set of options, whereas what is crucial is creating new ones.

This article doesn't do a good job of getting at the main points of Valiant's book Educability, in my view. You can see some of them in e.g. this talk he gave here: https://www.youtube.com/watch?v=W4fIoLGjFtM

He makes various arguments in the book that I disagree with, two of which I've put below. On the whole I think it is directionally correct though, and worth reading.

The first quibble I have is about humanity's most characteristic trait. In the book, he writes: "The mark of humanity is that a single individual can acquire the knowledge created by so many other individuals. It is this ability to absorb theories at scale, rather than the ability to contribute to their creation, that I identify as humanity's most characteristic trait".

I don't think that ability to acquire knowledge from other people is our most most characteristic trait. Creativity is. Learning is a form of knowledge-creation, and it is a creative process. We don't passively "absorb" theories when we learn from someone else. Instead, we actively look for and attempt to resolve problems between our existing ideas and the new ideas to create something new.

Another thing I disagree with is when he touches on AGI. He makes the argument that "we should not be fearful of a technological singularity that would make us powerless against AI systems". This is because it will "asymptote, at least qualitatively, to the human capability of educability and no more".

This is reminiscent of David Deutsch's argument that people are universal explainers, and AGI will also be a universal explainer; there is nothing beyond such universality, so they will not fundamentally be different from us (at least, there is nothing that they could do that we couldn't in principle understand ourselves).

I think this is true, but it misses something. It doesn't address the point that there is a meaningful difference between a person thinking at 1x speed (biological human speed) and a person thinking at e.g. 100000x speed (AGI running on fast hardware). You can be outsmarted by something that wants to outsmart you, even if you both possess fully universal educability/creativity, if it can generate orders of magnitude more ideas than you can per unit time. Whether we should be fearful or not about this is unclear, but I do think it is an important consideration.

His overall message though, is good and worth pondering: "Educability implies that humans, whatever our genetic differences at birth, have a unique capability to transcend these differences through the knowledge, skills, and culture we acquire after birth. We are born equal because any differences we have are subject to enormous subsequent changes through individual life experience, education, and effort. This capacity for change, growth, and improvement is the great equalizer. It is possible for billions of people to continuously diverge in skills, beliefs, and knowledge, all becoming self-evidently different from each other. This characteristic of our humanity, which accounts for our civilization, also makes us equal."

Here are some things that I think about often from that book.

* Problems are inevitable, but problems are soluble

* All evils are the result of insufficient knowledge

* Knowledge being the result of trial and error, and there being no such thing as certain knowledge.

* The idea of 'wealth' being the set of all physical transformations you can bring about.

* The thing that distinguishes people from non-general intelligences is the ability to create an endless stream of explanatory knowledge; that is, to have unbounded creativity.

* People are universal explainers; anything that can be understood, we can understand

* If something is permitted by the laws of physics, then the only thing that can prevent it from being technologically possible is not knowing how

* How commitment to knowledge growth entails a commitment to particular moral values (tolerance, openness to being wrong, valuing the truth); objective morality

I love how the ideas support each other and have such tremendous reach (morality, politics, epistemolgy, computing). And it is written very cleanly and lucidly, which perhaps makes it easy to read quickly and miss how dense and deep the ideas actually are.

Does this come down to the fact that you don't like the topic of his book? As you have said elsewhere that you're not going to criticize someone for wanting to make money.

What makes you think his promotion of the book was particulary motivated by making money rather than any other reason an author might promote their book?

The Beginning of Infinity by David Deutsch. Has an exceptionally high density of good ideas.

The first time I read it, I didn't love it and only engaged with it superficially. But gradually I began thinking about it more and came back to it, and I read it with more attention. After re-reading it several times I think it is one of the deepest and most important books I have ever read. It has changed how I see the world.

Me too. I’ve been impressed with some essays I’ve listened to via Open AI TTS. Much better than the librivox ones I’ve occasionally suffered through, and it’s only going to get better.

David Deutsch opposes this view in his 'It from Qubit': https://www.daviddeutsch.org.uk/wp-content/ItFromQubit.pdf

[the view Deutsch opposes] entails giving up on explanation in science. It is in the very nature of computational universality that if we and our world were composed of software, we should have no means of understanding the real physics – the physics underlying the hardware of the Great Simulator itself. Of course, no one can prove that we are not software. Like all conspiracy theories, this one is untestable. But if we are to adopt the methodology of believing such theories, we may as well save ourselves the trouble of all that algebra and all those experiments, and go back to explaining the world in terms of the sex lives of Greek gods.

...

these approaches fail because they attempt to reverse the direction of the explanations that the real connections between physics and computation provide. They seem plausible only because they rely on a common misconception about the status of computation within mathematics. The misconception is that the set of computable functions (or the set of quantum-computational tasks) has some a priori privileged status within mathematics. But it does not. The only thing that privileges that set of operations is that it is instantiated in the computationally universal laws of physics. It is only through our knowledge of physics that we know of the distinction between computable and non-computable (see Deutsch, Ekert and Luppaccini 2000), or between simple and complex.

There's a related video here: https://www.youtube.com/watch?v=UohR3OXzXA8

And more on what he calls the Mathematicians’ Misconception: https://www.daviddeutsch.org.uk/wp-content/uploads/2018/03/M...

Incidentally, Wheeler was his boss at one stage.

I recently read Deep Utopia and I found it frustratingly disjointed and inelegant. I kept expecting a big reveal, or a drawing together of the various threads into an unexpected insight, which never came.

I agree with the review by Steve Jurvetson, linked at the start of Hanson's review.

My biggest frustration with the book is that he takes over 500 pages to convey what could be more clearly said in well under 50...

there is no high-level organization to the book...

more of a survey of all possible answers versus the much more difficult task of making specific predictions....

Sad news. I aspire to be as intellectually acute in old age as Dennett was. His recent autobiography was engaging, although somewhat too indulgent at times. I admire how he created a life and a world-view that worked so well for him.

You can change the LLama 3 system prompt to coax the right answer. Shame it doesn't work out of the box though.

Tegmark’s institution is the Future of Life Institute, this is the Oxford Future of Humanity Institute.

Tegmark’s institute is well-funded, apparently largely due to a big crypto donation from Vitalik Buterin. https://www.politico.com/news/2024/03/25/a-665m-crypto-war-c...

A tool for searching transcripts from a selection of videos featuring David Deutsch.

The transcripts were AI-generated using whisper, and may contain some errors.

Search for keywords using the search bar, and then click on a result to see the video and its transcript.

You can click the text of the transcript to play the video at the relevant time.

Here is a video of a discussion between Richard Dawkins and Denis Noble (who wrote the linked article, and who was Dawkins' doctoral examiner): https://www.youtube.com/watch?v=uLC0akD1WOE

I highly recommend the audiobook of Paradise Lost read by Anton Lesser.

I had a Book of Kells colouring book when I was a child, I really loved it. Such beautiful designs.

I have played around with the OCR on my mac, and have been very impressed. It has been consistently better than tesseract for my purposes.

However, when creating a PDF from images using Preview and exporting using ‘Embed Text’ option to OCR, I have noticed the text is worse than if you OCR the exact same images using the shortcut above or using a script. Presumably Preview is using the Vision framework’s less accurate fast path when preparing the PDF. https://developer.apple.com/documentation/vision/recognizing...

Certain interpretations (e.g., the "Many Worlds" interpretation of QM) are more conducive to this possibility than others.

Generally, "free will" is simply stated as "the freedom to do otherwise".

In the Many Worlds interpretation, as I understand it, everything that can happen, does happen (as it would in a single universe of infinite extent). So no matter what you choose to do in one universe, the multiverse-wide outcome is predetermined: copies of you always make all choices not forbidden by the laws of physics.

I think this just pushes the problem of determinism out to the multiverse as a whole.

Then you have n free and conscious agents! And yes, if their simulated environments were identical, then the copies would all make identical decisions (all the copies would still have the same conscious experiences of free will that you and I have).

Whilst this feels unsatisfactory, it is the only way I can see of talking about agency without invoking decision-making processes that disobey the laws of physics, or rely on randomness. And unless a supernatural version of free will appeals, I think this is the best kind of free will available: governed by rules, but nevertheless fundamentally unpredictable. The only other option would be to rely on complete randomness as a basis for decisions, which seems worse.

How do you predict what the full simulation will do, before you run it for the first time? If it is computationally irreducible, then you can't predict the outcome - the only way to know what will happen is to run the computation and see what happens. And this computation, this full simulation, is a person with free agency. They are just running on a non-biological computational substrate.

Stephen Wolfram's discussion of free will in his book A New Kind of Science is the most relevant:

https://www.wolframscience.com/nks/p750--the-phenomenon-of-f...

Daniel Dennett's book Consciousness Explained comes at the problem from a similar angle.

I find the idea of computational irreducibility useful for rescuing a type of free will in a fully determined world. Computationally irreducible programs are perfectly deterministic, yet you can't predict the output before you run them, and nor are there any shortcuts to get there before running them for the first time. The program running in our brains for 'free will' can be computationally irreducible in this way. No-one can say with perfect accuracy what we will choose to do in advance, not even us.

For me, this kind of free will is enough, though it is rather deflationary when compared to some kind of godlike non-causal free will. I think that if we search for non-causal free will, we lose any hope of explanation or understanding: presumably if it's non-causal, then anything can happen, for no reason.

Scientific knowledge, like all knowledge, is fallible. All we can hope to do is correct errors and generate better explanations.

There is no "ultimate truth" or airtight reliable outlook at which we can stop and say we know everything there is to know -- that would be a dogmatic assertion, and it is false.

They are baguettes. They’re eating at the Copper Kettle and these look like their ‘breakfast baguettes’

What objective values are you judging it by?

You made a relative value judgement in your comment:

At the time, given the lack of offer and the standards they had, sure, the books were great.

From today's perspective, it's boring, it's not insightful and the style is excruciatingly dated

From your perspective. From my perspective, Madame Bovary and Sentimental Education are exceptional. Fresh, engaging and psychologically insightful. I would recommend anyone who enjoys reading to try Flaubert.

Stating there is no free will does mean we can predict everything you'll do

You can only predict it after you've run the simulation once, and after you have let the simulated person do what they want to for the first time. During the first run, free will is unpredictable.

Only after you have run it can you say "if I run the simulation again with the exact same parameters, I know what they will do next".

And if you find a way to create a faster version of the simulated person algorithm, then the problem just devolves to the faster version. How can you predict what the fast-free-will person will do before you let it do what it wants to do? And so on.

How could you know what a perfect simulated version of me would do in advance, without first running the code? Even if you found a more efficient algorithm for getting the exact same end states faster, at some point you will reach a computationally irreducible limit, where you can’t predict what the improved simulated agent will do without first actually running the code. When you run the code it will freely decide to do one thing over another, and no one can know what this will be in advance, even in principle.