1) From a contradiction, anything follows. It is insanity all the way down.

2) Post-Modernism is the philosophy which undergirds all the contradictions and insanity.

3) NRx is primarily concerned with undoing this insanity in *all* its forms.

4) Therefore, NRx is, at its base, nothing more than the rejection of post-modernism.

5) That NRx seems much larger than this is only due to the extent that post-modernism has infiltrated every single facet of our society.

Chris B, will we be seeing your synthesis, ” It’s all fitting together really naturally. Spontaneous order, price signals, common law, cybernetics, chaos theory, complex systems etc.” made into a post at New Internation Outlook?

Democracy replaces thought with a newly-castrated voice and a body shared by all. Left-singularity is its destination.

Similar thing happening here but with somewhat different areas.

Keep in mind these are my irresponsible speculations/extrapolations, not authoritative statements. I’ll refer to the papers in the order of appearance in my previous post.

[1] Shows some of the subtleties in the interaction between mutation and fitness. Too much instability and you get a bunch of harmful ones, but too little and you can’t get any good innovative ones. Mutation in specific circumstances helps but in others it has diminishing returns. This sets up ground to think deeper about this relationship of optimality and variation in the context of biology.

[2] Tries to give a plausible theory of how sexual reproduction can improve fitness. If your father has excellent genes but your mother has awful ones you could end up with many awful genes out of luck, so how come sex ends up being adaptive? This theory says it’s basically because it helps select not genes which are good (but only in the presence of particular genes), but genes which are more likely to make good combinations with other genes. In other words, sex makes selection go from trying to get one good set of genes to trying to get a set of genes which if mutated would also be a good set of genes, so gene sequences which are in a neighborhood of other good sequences.

“Why the reduction in variance has this effect can be understood intuitively by analogy to finance. To diversify a portfolio (13) among various investments and thus average its growth rate over time, an investor must keep rebalancing the portfolio at regular intervals (14). Otherwise, the portfolio will soon be biased toward the investments that had the best individual returns so far. The investments here correspond to the genotypes that share an allele, the portfolio corresponds to that allele, and by analogy, it is the persistent rehomogenization by sex of frequencies of genotypes carrying each allele that shifts the focus of natural selection over the generations from the individual performance of genotypes to the average performance of alleles.”

It also contains a deeply reactionary insight.

“The mechanism enabling the selection for mixability operates on the multigenerational time scale. It has been out of the purview of previous theory, which often focused on single-generational changes and the equilibria that could be calculated from them. This mechanism shows that, while at any one generation natural selection operates on genotypic fitnesses, over the generations and in the presence of sex, it is particularly efficient not in increasing population mean fitness but in increasing the ability of alleles to perform well across different combinations. Thus, in this mechanism, natural selection and sex operate interdependently and need to be understood in the context of each other.”

A little background on [3]. Geoffrey Hinton is a researcher in machine learning, specially in neural networks and deep learning (wiki “artificial neural network” if you’re not sure what they are, ask me about them if you need help). He was originally an experimental psychologist and has a lot of interest on and knowledge of the brain. Also, a curious hereditary trivia (reactionaries tend to like this sort of thing); he’s a descendant of George Boole.I think he’s at Google now. So perhaps inspired by [2] his team discovers “dropout” which is a very useful technique.

Imagine you have to train a classroom full of kids to accomplish a particular task in the aggregate (that is, one mark for the whole class, a joint project). They get tested at the end of every lecture and there’s also a final exam which uses problems they’ve never seen in the coursework but aren’t more difficult and still the same general task. This can go bad in a few ways. First, the kids might just be stupid, but assume this isn’t the case. There’s another, much more subtle way things can go wrong. The students can do very well on coursework but a LOT worse in the final exam.

How come? Well, what can happen is a case of overfitting (there is such a thing as overthinking a problem). The students learn not the patterns in the data you gave them, but just the data itself. Think of it like a plastic glove that fits so snugly that as soon as you move your hand it rips apart. Models have to be flexible and generalize, not just fit a particular set of data well. You observe the class and get an idea. You’ll randomly pick half of the class at the start of each lecture and the rest leaves. Why would this improve things? Because of co-adaptation. Think of free-riders or cliques which only know how to work with each other. This way you force the kids to focus on the data, not each other, and to make independent interpretations, conclusions and to learn how to work well with anyone, not just their friends.

Breaking co-adaptations to increase adaptivity is the important lesson here.

I’m going to skip [4] for now and go back to it a bit later.

In [5] and [6] Bengio articulates several hypotheses relating machine learning and in particular neural networks to human brains and behavior, mainly related to difficulties inherent in some of the learning processes that might be going on in the brain.

From [5],

“[…] and (5), language and the recombination and optimization of mental concepts provide an efficient evolutionary recombination operator, and this gives rise to rapid search in the space of communicable ideas that help humans build up better high-level internal representations of their world.” – Remember [2]?

“These hypotheses put together imply that human culture and the evolution of ideas have been crucial to counter an optimization difficulty: this optimization difficulty would otherwise make it very difficult for human brains to capture high-level knowledge of the world. The theory is grounded in experimental observations of the difficulties of training deep artificial neural networks. Plausible consequences of this theory for the efficiency of cultural evolutions are sketched.”

In [6] he outlines a task that is actually quite simple but which nearly all algorithms fail spectacularly at. He then shows that if you have decent enough priors the task can be learned perfectly and stipulates that this might be the origin of forms of supervision, guidance, curricula, etc. Think of riddles, puzzles and other kinds of moments where you were struggling to solve something then went “D’oh! How did I not see THAT!? It’s so simple!”, hints in our environment or institutions help us solve many such problems without even being aware of them or their solutions.

In [4] hbd chick looks at what happened to european tribes and what impact the catholic church might have had on them. Tribes, you know, those highly co-adapted forms of living like in [3] where you can rely on your clan or family? Maybe you can see where I’m going with this since I’ve tried to set up the proper priors before [6]. You have the catholic church, a more structured and rigorous culture [5] imposing forms of selection that break co-adaptation [3] on the population and giving them hints [6] like “hey, maybe don’t fuck your cousin?” for forms of organization which are more versatile in the long run like in [2].

Now, Bengio doesn’t ask some questions which should come quickly to a reactionary (his first name is Yoshua and he’s canadian, so…) like “is there a co-evolution of psychology and culture? “, “if culture is a matter of massively parallel searches and solution retention, recombination and selection, doesn’t this mean some cultures really are better than others?”, “what about informal institutions and structures? do they unwittingly pass on solutions too?”, “if you get rid of them, what happens?”, “are different sets of solutions (that is, different cultures) compatible?”, and so on.

So in short, these papers help explain my perspective that mental, cultural aspects of human civilization are soft reflections of evolutionary processes that exist to provide greater adaptivity, organization and complexity to them, and that many of the processes and institutions in human society can effectively be considered sets of interacting virtual superorganisms, like supra-rational software operating on more local human hardware.

The optional paper comes up with an interesting way to define adaptive behavior. It’s mainly lead to me thinking of intelligence more in terms of a peculiar property of certain types of systems rather than an interior characteristic of intelligible informational processes, but there are probably a lot less technical papers out there by now that make clearer definitions like this, so it’s not too important aside from the conceptual interest.

Hopefully I’ve managed to make a decent summary without mangling the ideas too much and that this at least has been a somewhat stimulating post. In retrospect perhaps it’s best to start with the two papers from Bengio, specially Evolving Culture vs Local Minima because it is much more readable, cogent, less technical and overall a much more pleasant read, then going on to read the others with a better idea of what I have in mind when posting some of this stuff.

My view on these sorts of matters is fundamentally mathematical, in practice computational and universally evolutionary. But like I said, this is mostly my own irresponsible mind wandering, so don’t go with just my word on anything.

For a minimal set of references to serve as an introduction to this stuff I guess it could be these 6 papers above plus

An Introduction ot Statistical Learning – Hastie et al – free @ http://www-bcf.usc.edu/~gareth/ISL/

Evolutionary Game Theory – Weibull

Social and Economic Networks – Matthew O. Jackson

http://www.scholarpedia.org/article/Ensemble_learning

http://www.eecs.wsu.edu/~holder/courses/CptS570/fall07/papers/Dietterich00.pdf

This way you get the stats, the ML, the game theory, and generally the most informative stuff as far as this perspective is concerned (there are still plenty of others, most of which I’m trying to get some kind of grasp on) and enough references to continue looking if you’re so inclined.

Oh, also there’s another paper which shows pretty well why I have an interest in mathematical sociology. Some of this stuff sounds like it comes straight from the mouths of reactionaries.

The Economics of Social Stratification in Premodern Societies – http://www.tandfonline.com/doi/abs/10.1080/0022250X.2012.724488 , it’s filled with historical goodies and other stuff – : “We present a microeconomic model of social stratification, which includes an endogenous fertility component. In the model, egalitarian and stratified societies coexist. The latter are divided into 2 hereditary classes: a warrior elite and a productive class. The model entails that the extra cost warriors must incur to train and equip their children for war determines the relative sizes of both classes and the degree of economic inequality. Higher costs of warrior children imply a greater economic advantage for warriors and a smaller ratio of warriors to producers. These results are consistent with the historical evidence. Finally, we explore conditions under which the social contributions of the warrior elite could discourage a revolution.”

It ends with : “The spread of social stratification, despite its social injustices, may have been an endogenous evolutionary response to intersocietal competition.”

Fantastic list of links. I love Turchin’s work, he’s quality.

@Chris B

“It’s all fitting together really naturally. Spontaneous order, price signals, common law, cybernetics, chaos theory, complex systems etc.” Intriguing. Will you be flesh out this convergence at New International Outlook?

‘Nick Land may be the quintessential machinic philosopher; but he’s still a person. And I think he misses his old friends.’

Expect missionaries.

I’m busy digesting lots of other stuff I’ve been reading in the past and now so that I can formulate a number of articles. It’s all fitting together really naturally. Spontaneous order, price signals, common law, cybernetics, chaos theory, complex systems etc.