Caused Pearls

Recommended highly it is, young Skywalker, to read Judea Pearl’s The Book of Why? — the link there is to a useful book review — before absorbing the textbook “Causality, Models, Reasoning and Inference” .

I linked to the book review since it is a good warning to not take Pearl’s polemics too much to heart. But I want to go beyond that critique by Aronow and Sävje. Aronow and Sävje fail to critique Pearl’s scientism and physicalism.

My quick review: nerds who try to capture the infinite are doomed … to receiving too many government scorepoints they did not deserve, and all the consequences thereof.

One thing the reviewers, Aronow and Sävje., get right is close to my beef, but they couch it in humbler terms, they note Pearl himself is not claiming his “casual revolution” has solved the problem of causal discovery.

J. Pearl (pp. 79–80):

“… in contrast, the focus of Wright’s research, as well as this book, is representing plausible causal knowledge in some mathematical language, combining it with empirical data and answering causal queries that are of practical value. Wright understood from the very beginning that causal discovery was much more difficult and perhaps impossible.”

This is good.

It means even Pearl recognizes there is a possibly insoluble problem, which in my mind reveals a faint trace of The infinite. Aronow and Sävje rightfully do not go there, because their audience is the fellow geek, who probably subscribes to materialism.

What’s a little annoying to me is that Pearl does not seem to even entertain the (extremely strong) possibility that the impossibility of solving causal discovery might be the exact obstacle to Strong AI. We may never achieve construction of Strong AI systems, except my having babies.

The issue here for Physicalism, is something I write about a lot, or whenever I have a new angle on it, which is the issue of the 4D Block Universe. In my personal favourite conception of fundamental physics, we have to postulate the 4D Block Universe. Which means there is no such thing as physical causality. All causality is exogenous. Like government scorepoints are to the economy. (Banks, commercial banks, create bank scorepoints, not government scorepoints. Someone please inform the Keen Pilled Post-Keynesians!)

Further Education

I saved you 10 seconds of typing into a search bar:

• Intuitivee MetriX| Causal Inference
• Brady Neal | Causal Inference
• McElreath | Science Before Statistics: Causal Inference

I have not watched all these, so they’re not recommendations. They’re for people who only have spare time to do continuing education at lunchtime while the boss is not looking.

Capturing the Infinite

My overarching issue is that Pearl proclaims too loudly to have captured a decent notion of Causality, so decent he thinks, that Strong AI can be based upon his concept. All the machines lack is “understanding” of “Cause”.

I think he has the Cart before the Horse.

You cannot have understanding of causality before you have understanding. No one knows how we achieve understanding. So the idea programming notions about cause & effect using arrow diagrams into a computer is going to accomplish this is … barking mad.

This provides Bijou’s Law of AI:

For every new idea in cognitive science there will be someone claiming, “Aha! That is all we need for Strong AI!”

Such n3rd intelligence is indeed artificial. We’ve already achieved AI. Every single Tech CEO.

My deep philosophical warning is that in attempting to capture causality you are attempting to capture the Infinite. Not just $\aleph_0$, but The Absolute.

Should I continue going through Pearl’s howlers about Strong AI claims?

No, that’d be almost puerile. I’ve had my say. You can be warned, or not be warned. Invest in the AI tech stocks and see what happens. You might even get rich like Garrett Lisi and be able to fund your own theoretical research Institute in Hawaii. But you will not achieve human intelligence. Why not? Well, I’ve written too often about this, about how mental qualia are non-physical, and I have nothing new to say. Most people either just agree or disagree, few are truly agnostic, and I’m only interested in talking with the agnostics. And even then, only agnostics with no bias, so those who are flexible in mind. This is not to stifle debate in my little circle (what circle?) but to avoid depression. Yes, I’ve hospitalized my philosophy.😔

Macroeconomics Causation

Pearl’s Book of Why? is a good read if you skip all the Strong AI paragraphs. I was heartened by an economics writer, Nick Huntington-Klein willing to employ Pearl’s tools in economics,

• “Pearl before economists: the book of why and empirical economics” .

The guys at AppliedMMT USA were aiming to get into such modelling, so if you are an insider there and have some progress to report do let me know. These days I do not routinely track MMT whitepapers, but there is still a gap I think in causal modelling in MMT scholastic studies. The issue being you cannot take the MMT principles as the basis, since what MMT is claiming is that most of the causality is in the human beings making policy decisions. So you need to model their mental models! FN-1 They are not using MMT knowledge! So the MMT Base Case is not in today’s real world a causal model. It is a model of what might have been, and what could be. Something many of my critics do not seem to understand.

I have some very quixotic quests. Money is best defined as an IOU of the issuer. ꘝꕯꕷꕷ off with all your other definitions, they’re not helpful. These quests are not things I will ever back down from, because the apparent absurdity is the whole point. Once the Paradigm Shift is resolved they will not seem absurd, and may even seem quaintly tautological. When paradigms shift it can be that catastrophic (René Thom meaning). People will all understanding the meaning of “government debt” and will think of it as a credit. To justify it we will need people working to produce things. Hopefully with a gazillion low energy throughout helper robots (let’s try and avoid the nano-junk garbage of non-biologically degradeable nanobots, Mr Kurzweil).

I do not care if Huntington-Klein is a dumb-dumb neoclassical or New Keynesian. If he is prepared to accept causal inference tools as valid, then we’ve got a slim chance, but importantly non-zero, of “proving” to people like him (beyond all reasonable doubt) that not only is MMT correct, but the MMT Money Story explains, causally, the source of all unemployment. This is a project I have on-going, unfunded.

Could be sped up if I had a wealthy benefactor. But I like doing things at my own slow impoverished pace, it is intellectually richer this way, so I won’t complain if you do not throw heaps of government scorepoints my way, or into my KoFi account . I’ll be spending them on espresso and sandwiches, not on producing MMT words, the latter will be collateral damage.

Footnotes: