Does digital socialism have a future?

No, not a good future, according to Jesús Fernández-Villaverde:

Can artificial intelligence, in particular, machine learning algorithms, replace the idea of simple rules, such as first possession and voluntary exchange in free markets, as a foundation for public policy? This paper argues that the preponderance of the evidence sides with the interpretation that while artificial intelligence will help public policy along with several important aspects, simple rules will remain the fundamental guideline for the design of institutions and legal environments. “Digital socialism” might be a hipster thing to talk about in Williamsburg or Shoreditch, but is as much of a chimera as “analog socialism.”

The paper is an excellent response to a growing set of claims, I would add further material on the work of Michael Polanyi and the importance of inarticulable knowledge.

Comments

I had been unaware that some people were backing the idea of "digital socialism" (or really I think it's "digital central planning" that this paper is critiquing) but yeah AI and ML improve our existing toolset but do not change the fundamental nature of the complexities and unknowns in an economy.

Google, with probably the best "big data" analysts in the world thought they'd come up with a superior method of forecasting influenza epidemics. Until it turned out they hadn't. (Google "google flu failure" for some articles.)

These caveats about the limits of AI and ML apply to capitalism too though. Self-driving cars are not going to be taking over our streets, unless and until we fundamentally overhaul the laws and enforcement of the laws for human drivers and pedestrians -- or restrict the self-driving vehicles to controlled environments such as freeways. I haven't seen an AI model that can predict weather better than the existing models, although maybe there's one out there. Etc.

"I've got the Facebook pneumonia and the Google doodle flu!”

In the novel "Red Plenty" (which includes a lot of actual history) there's much about technocratic central planning in the immediate post-Stalin years, based on believable mathematics (Kantorovich and others).

Quick take:

> the importance of inarticulable knowledge.

Isn't that what AI creates? From wikipedia:

Demis Hassabis, the co-founder and CEO of DeepMind, said that AlphaGo Zero was so powerful because it was "no longer constrained by the limits of human knowledge".
- https://en.wikipedia.org/wiki/AlphaGo_Zero

This is "inarticulable knowledge," is it not? At least from our perspectives.

I very much agree! As a 'digital socialist.'

"Digital Socialism" is predicated on the notion of simple rules. I would argue that we could use this as a rubric. Any system that is centrally planned (and inefficient) will be replaced a digital social system, if that system simplifies the service.

It is exactly because of their simplicity that those systems will out compete centralized systems on the market. However-- it is not because of AI.

But because of the DLT solution the byzantine generals problem that allows for synced distributed/ decentralized computing. So you hit the nail on the head, AI will not enable digital social systems, but DLT will.

Just like double entry accounting, DLT is the game changer.

Digital Business knows this well we are grappling with how to formulate new institutions, in these grey areas, especially when you have a weapon such AGI, at stake.

You can look at the Economic of Uber as example. Each ride is subsided by investor notion of value of the entire system.

https://www.ted.com/talks/clay_shirky_institutions_vs_collaboration

so your argument is that blockchain will enable socialism to defeat the free market, k

"Socialism" is some form of redistribution away from private ownership, and it is specifically centrally enforced redistribution.

Or to put is another way, if there were a voluntary or grassroots form of socialism, what would it look like? People voluntarily giving away some of their property in order to support social goods like helping the poor. This already exists, and we don't call it socialism, we call it charity.

The idea that "socialism" can be achieved through blockchain, or other decentralized mechanisms, makes no sense.

If you disagree, please provide a counterexample of a "socialist" outcome that could be achieved through DLT.

Yet large digital corporations don’t use block chains for their own governance at all: They have been around for over a decade now, and all they have revolutionized is digital blackmail and trade in illegal goods.

Between the heavily understated security risks, the unfixable meat/digital divide, and the impossibility of always accurate, efficient, civics encompassing human language that could be encoded digitally, all the miracle blockchain use cases disappear.

I used to go the Brooklyn Bowl and watch the concerts while I bowled. It wasn't just that I could do two things at once, it was that each was separate, and never the twain shall meet.

Digital socialism is like the promised rockets to Mars: tech promising to do good while making a mess of everything. Some people will believe anything tech billionaires tell them. Nonsense.

So you are saying digital socialists are just like digital capitalists? Both want to control the masses don't they?

So comunism is not that great?! But isn't Sanders your Messiah?

Who in his right mind would trust Bernie to run anything?

Most democrats until now, it seems. Also, according to experts, Sanders is beating Trump: https://www.vox.com/platform/amp/2020/1/31/21113780/bernie-sanders-socialism-electability-primaries?__twitter_impression=true

Only Mike can stop Sanders because Mike can run on his record. Mike will get it done!

Vox? Don’t make me laugh!

"replace the idea of simple rules, such as first possession and voluntary exchange in free markets, as a foundation for public policy"

Clearly ruling out the US as an example. Reinforced by Trump denying rights to those in first possession of wasteland and backing people with guns taking land from those who were in first possession, and then away from those in second possession by use of violence against America's first nations.

The thing about "digital" is the "stuff" is available in infinite supply: the number of 1s and 0s is infinite, and very little work is required to produce huge strings of 0s and 1s, basically a clock and some radioactive material.

You don't respect the guy if you put a - between the 2 last names, so why bother with the article?

AI : GIGO

'Socialist AI' is only as good as its data inputs -- and we KNOW it is impossible for a national central planning & control/command economy to calculate relative values and efficiently allocate resources.

Socialists lack the critical dynamic market pricing system to discover & communicate relative values of economic resources.

I didn't see anything major that I didn't get from my teacher Hubert Dreyfus in 1975. See his book...What Computers Can't Do.

"Dreyfus argued that human intelligence and expertise depend primarily on unconscious processes rather than conscious symbolic manipulation, and that these unconscious skills can never be fully captured in formal rules. His critique was based on the insights of modern continental philosophers such as Merleau-Ponty and Heidegger, and was directed at the first wave of AI research which used high level formal symbols to represent reality and tried to reduce intelligence to symbol manipulation.

"When Dreyfus' ideas were first introduced in the mid-1960s, they were met with ridicule and outright hostility."

He had a Yeats quote on his wall....

'Man can embody truth but he cannot know it'

Painfully cringe-y abstract.

A innocent superhuman intelligence asked to devise the optimal means of producing and distributing the goods and services to satisfy human wants within the constraints of the natural world, would probably come up with something like the state capacity libertarianism we all know and love.

An AI programmed to guide socialism would have to be supplied with barriers to direct it away from the clean, simple solutions, and down the the Byzantine byways of Marxian illogic. You would have to deliberately create an evil AI.

Apparently, the Chileans tried something like this long ago. It doesn't seem to have accomplished much.
https://en.wikipedia.org/wiki/Project_Cybersyn

"inarticulable knowledge": there must be a better adjective than that. Any suggestions? I don't think "unspeakable" quite works.

Ineffable? Not finitely describable? Incapable of being formalized? 'unformalizable' works, but is equally awkward...

"The peculiar character of the problem of a rational economic order is determined precisely by the fact that the knowledge of the circumstances of which we must make use never exists in concentrated or integrated form but solely as the dis- persed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess. The economic problem of society is thus not merely a problem of how to allocate “given” resources –if “given” is taken to mean given to a single mind which deliberately solves the problem set by these “data.” It is rather a problem of how to secure the best use of resources known to any of the members of society, for ends whose relative importance only these individuals know. Or, to put it briefly, it is a problem of the utilization of knowledge which is not given to anyone in its totality.
F.A. Hayek, The Use of Knowledge in Society (1945)"

Hayek's view I view as a form of Pragmatism, akin to Frank Knight's views...

The right principle is to respect all the principles, take them fully into account, and then use good judgment as to how far to follow one or another in the case in hand. All principles are false, because all are true—in a sense and to a degree; hence, none is true in a sense and to a degree which would deny to others a similarly qualified truth. There is always a principle, plausible and even sound within limits, to justify any possible course of action and, of course, the opposite one. The truly right course is a matter of the best compromise or the best or “least worst” combination of good and evil. As in cookery, and in economic theory, it calls for enough and not too much, far enough and not too far, in any direction. Moreover, the ingredients of policy are always imponderable, hence there can be no principle, no formula, for the best compromise. That laws must be stated in sentences partly accounts for the familiar “principle,” “the law is an ass.” And if people don’t have good judgment, or won’t use it, it is “just too bad,” for themselves and for others over whom they have power. ■
—On the History and Method of Economics"

This Pragmatist view seems as relevant today as ever.

The Lucas Critique has always seemed to me a version of Giddens ' Double Hermeneutic...

"New Rules of Sociological Method: A Positive Critique of Interpretative Sociologies" by Anthony Giddens

"The themes of this study are that social theory must incorporate a treatment of action as rationalized conduct ordered reflexively by human agents, and must grasp the significance of language as the practical medium whereby this is made possible. The implications of these notions are profound, and the book is confined to tracing through only some of them. Anyone who recognizes that self-reflection, as mediated linguistically, is integral to the characterization of human social conduct must acknowledge that such holds also for his or her own activities as a social ‘analyst’, ‘researcher’, etc. I think it correct to say, moreover, that theories produced in the social sciences are not just ‘meaning frames’ in their own right, but also constitute moral interventions in the social life whose conditions of existence they seek to clarify."

"To assess the validity of these ideas it is necessary to go over some of the ground covered in New Rules about the concept of the double hermeneutic – in respect not just of the meaning of ‘double’ but also of that of ‘hermeneutic’. The idea of the double hermeneutic is partly a logical and partly an empirical one. All social science is irretrievably hermeneutic in the sense that to be able to describe ‘what someone is doing’ in any given context means knowing what the agent or agents themselves know, and apply, in the constitution of their activities. It is being able (in principle) to ‘go on’ – mutual knowledge shared by participants and social-scientific observers. The hermeneutic element involved here does not have a parallel in natural science, which does not deal with knowledgeable agents in such a way – even in the case of most animal behaviour. This is the logical side of the double hermeneutic. Lay actors are concept-bearing beings, whose concepts enter constitutively into what they do; the concepts of social science cannot be kept insulated from their potential appropriation and incorporation within everyday action."

Large government institutions can open their own cash management account. Their cash flow and value flow better match, an easy 15% productivity improvement in programs. Something as simple as an automatic liquidity machine for FX transers likely adds 3% productivity improvement in banking.

From the blurb, it seems the author might even neglect the productivity improvement from the telegraph.

of course. what we will have instead is mass and regular redistribution aka UBI.

Let's get the humans out of humanity.

If the AI is really smart enough to understand each person’s utility function and maximise overall utility accordingly then I suppose it could work. But in that circumstances there probably wouldn’t need to be any work done, since pretty much everything could be automated. I wouldn’t call this socialism, more a successful AI society.

File Under "The Academic Emperor Has No Clothes" -- for does not a form of "digital socialism" already exist? It's called Google Search, which is available to all for free ...

I don't disagree with the main conclusion. However, being in the AI business, I think AI can plausibly help with some of the "small data" problems in economics in exactly the same way that it is/promised to help out in physics/chemistry/biology, where the use of AI/Machine Learning as a "function approximator" is allowing researchers to run experiments "in silico" for a many-orders-of-magnitude reduction in cost, time. I see no obvious reason why the same shouldn't hold true for economics. Perhaps "AI"-driven models of firms and consumers would allow for "in silico" experimentation with economic policy?

Then, there is the "dirty secret" of our economy, all these outwardly-respectable capitalist firms the _internally_ are hugely reliant on Planners: AI (machine learning) is already (to the surprise of absolutely no Computer Scientist) outperforming, for example, supply chain planners; our market economy has vast hinterlands where markets do not exist, and in these places there are many planners susceptible to disruption by mechanical optimisation.

Neuralink can solve all those issues

Comments for this post are closed