This quote, "it is easier to imagine the end of the world than to imagine the end of capitalism," has been attributed to the cultural analyst, Fredric Jameson.
It bears deep reflection.
Capitalism is the first mode of production which separated the mass of people from the means of securing their own livelihood. Previous modes of production saw ordinary people as hunters, gatherers, pastoralists or farmers.
Given the economic precariousness of the proletariat, some social organising principle is needed to coordinate actions to secure the reproduction of collective life. This is called an economy.
Historically there have been only two forms of economic organisation consistent with the existence of a proletariat (a class of resourceless labourers): the privately-owned, decentralised means of production constitutive of capitalism; and bureaucratically-owned means of production in a centrally-planned 'socialism'.
We know the track record of both models. In particular we know that a statist centrally-planned economy does not work - it lacks incentives to grow and optimise, and is captured by bureaucratic elites.
There are two further models: the naive Leninist-Trotskyist model of a worker-controlled (soviet) state and economy .. and a fully-automated economy.
The Leninist-Trotskyist model was only ever theorised in the sketchiest form and founders under modern political, industrial and financial complexity .. and human nature.
The fully-automated economy is a limit point of the natural tendency of capitalism to automate (the law of increasing organic composition of capital). However in that limit the last workers are displaced and therefore surplus value can no longer be created. Profits tend to zero as competition equalises prices to costs.
So no more capitalist economy, but that does not mean no economy. A collection of fully-automated companies (perhaps with non-waged human involvement in setting objectives - and certainly in consumption) is rather like a collection of villages each of which has a monopoly of some kind of natural resources (water, fruits, meat, ...) which are essentially not scarce. Such an economy engages in exchange, because human needs straddle the portfolios of the various producers, and it's similar to petty-commodity production, albeit without the involvement of human labour.
The working class may have vanished as a social category (no wage labour) but humans certainly haven't. Provided the machines have not been so badly designed as to eliminate humanity, people experience this world as super-aristocrats: their every feasible, practical need met.
"The future is already here — it's just not very evenly distributed," said William Gibson. And if we look at the super-rich today we see a preview of behaviours which could be open to all in a society of totally-automated abundance.
As expected, the behaviour-set is just what would be expected given the innate nature of the human animal - all drives are exhibited from the most prosocial to the most antisocial.
On a sufficient timescale (a century or two) human nature can itself be engineered and hopefully there will be major speciation events. We will find it easier and more robust to adapt to Mars than to make Mars adapt to current-us, for example.
Capitalism will abolish itself not through socialism but through self-elimination. And the human race will diversify into every possible ecological niche in this awaiting universe.
Unless we're stupid about it.
By pushing this super-automated future ahead by one or two centuries, it obscures any mechanism by which this might be achieved, or whether it could be achieved faster. Perhaps economic historians will look back from that time and say: "what a waste of the last two centuries, with a bit of forethought they could have gotten here over a hundred and fifty years ago."
ReplyDeleteThe main problem over the current decades with high-automation economy has been the "winner takes all" phenomenon, and its corollaries - not just for individuals but for States (and the High Street!) too.
First we had Microsoft's dominance of the PC (and Congress once thought that was bad!); later we have Social Media dominance (an area which seems completely out of control); and then the so-called "Gig Economy"; and now the various "High Street" problems. Then there is "Surveillance Capitalism". Coming are AI based Judgement systems trained on limited, biased (and un-paid-for) User/Citizen data.
How those Economic Historians of two Centuries hence would wish we could have avoided the next two hundred years of problems...
The bottleneck is - ultimately - AGI. The same thing which is pushing the Google car back into the indefinite future.
DeleteSo are we hoping that AGI wont be a reified commodity which could become the sole property of some future B. Gates III?
DeleteWe are also assuming that AGI can - someday - exist.
A lot of people could die, for sure. No guarantees. We are an existence proof of GI. It's a good question whether total automation can be achieved without artificial oppression.
DeleteBut how do we know that the AGI wont become the sole property of some future B. Gates III?
DeleteWe are also assuming that AGI - somehow - does actually exist.
The claim that "we are an existence proof of GI" could be deconstructed.
DeleteAs a GI we are strongly fallible in several directions. This fallibility is not what people probably mean by AGI.
The view that AGI is a computer hardware implementation of GI assumes:
(a) that a Turing Machine style model is possible - not yet proven but assumed; and
(b) if possible (in some small scale "house Butler" sense) the "computer" aspect implies a level of data processing in the economic context that humans cannot do.
For example the claim that "Centralised Economic management" might be possible with a computer AGI is asking the AGI to do *more* than humans have demonstrated they can do.
So no existence proof of an "Economically superior AGI" (able to show Central Planners how the job is done) is available.
P.S. I am reading I.McEwan's "Machines Like Us" in an alternative history in which Sir Alan Turing (in 1968!) solved the P vs NP problem generating an AGI which formed the basis of the (butler-style) machines of a later age.
I don't think a centrally-planned economy is possible in principle (Ashby's 'Law of Requisite Variety', or desirable (incompatible with individual creativity) or genetically possible (no genetic common interest).
Deletehttps://en.wikipedia.org/wiki/Variety_(cybernetics)#Law_of_requisite_variety
In that case I don't get your first reply - "AGI is the bottleneck". I had assumed that you meant that a "fully automated economy" meant roughly that economic activities and (almost necessarily) associated decisions would be under computer, more specifically, AGI control. Then this would take centuries because the AGI was no-where near ready.
DeleteBut if a "Law of Requisite Variety" is to kick in preventing the very existence of any form of automated Central Control, then there is some limitation to automation after all.
So the AGI "bottleneck" is only part of the story...
I meant the replacement of human workers by AGI 'androids'. Would they be oppressed in a psychological sense? Perhaps we could engineer them not to care.
DeleteWe could achieve the same rsults - more easily? - with genetic engineering: primate servitors. There have been SF stories. We might need to tweak ourselves a bit so as not to care about their situation - dial down the empathy.
Truly there are no moral laws in this universe, only the laws of physics!
How the "workers" in android, human or genetic human form are treated depends in part on the control mechanism.
DeleteReviewing the Ashby Law I don't see that as a "no go theorem", just constraints that must be understood and respected in the design process of the Control system. In fact the current world of Machine Learning has Amazon predict how to pre-stock the warehouses (and perhaps even send full lorries on their way) without having received all the orders yet. This is based on the predictive aspects of (statistical) machine learning of population (in that city, time of year, weather...?) which already covers a lot of human economic activity.
Extrapolating from here one might wonder whether a small socialist state (Cuba?, N.Korea?, Vietnam (if still socialist) (0r Singapore)..) might decide to replace their current Central Planners with some improved Amazon-type ML in say 10-15 years.
I am not sure at this stage whether the Laws of Physics play any role, but the Laws of Machine Learning and AGI (yet to be discovered) might play a role in what is possible, and in what eventually happens.
Centrally-planned economies put a veto network on innovation. Little of what Steve Jobs accomplished would have occurred in a centrally-planned economy.
DeleteIf an AI is allocating investment, how could it anticipate the creativity of individuals as regards innovation? It's not human.
There is more to an economy than optimal allocation.
If the authorities in small socialists countries want some advice on how to benefit from Amazon style ML allocation, and yet avoid the Steve Jobs/innovation bottleneck, for a small fee (a larger one if China asks!) something might be available...
Delete