Editor’s note: discussion topics include the relevance (or non-relevance) of 3-D printing, the internet of things, recommender systems, neural networks, and quantum computing, to socialism and the Towards a New Socialism model (TNS) in particular, handling second-hand goods in TNS, the number of commodities in a modern, advanced, capitalist economy, basic research and labour time in TNS.
After the Oligarchy: Hi everybody this is After the Oligarchy speaking to Dr. Paul Cockshott again. I’m going to read out his bio from his book How the World Works which is a very good book on historical materialism: Paul Cockshott is a computer engineer working on computer design and teaching computer science at universities in Scotland. Named on 52 patents his research covers robotics computer parallelism, 3D TV, foundations of computability, and data compression. His books include Towards a New Socialism Classical Econophysics and Computation and its Limits. And of course How the World Works.
This is the second in a series of interviews with Dr. Cockshott about Towards a New Socialism written by Paul Cockshott and Allin Cottrell, published in 1993. If you haven’t watched that first interview yet check it out. In Towards a New Socialism the authors present a bold vision of a democratically planned economy using computerized labour time. In this interview we’ll be discussing some more advanced questions about that model, so I recommend that you read the book to really understand what we’re talking about.
You can also watch some excellent videos on Dr. Cockshott’s YouTube channel, link to that and his website in the description below. Dr. Paul Cockshott, thanks for joining me again.
Paul Cockshott: Hi.
AO: I just want to, actually, say something that I found out in the meantime about this book How the World Works, that you don’t actually make any money from it because you chose to lower the price to make it more accessible.
PC: Yes.
AO: I just thought that that’s very impressive and interesting. But actually on that note I should also mention that you do have a Patreon, so if people want to support your work, given that, for example, you’re not making money from How the World Works, they can go to your Patreon and become a subscriber.
And last time we didn’t have a copy so here’s Towards a New Socialism, it’s still in print but you can actually get a free PDF version which I’ll put in the description as well.
So with all of that out of the way I’d like to begin with a set of questions about the relevance of various technologies to Towards a New Socialism.
1 – We’ll begin first with micro-production. What is the relevance of micro-production, e.g. 3D printing, to Towards a New Socialism?
I know that nowadays a lot is made of this. There is also small-scale silicon wafer manufacture, for example, I’m just wondering do you think there’s any particular relevance that this has to Towards a New Socialism?
PC: It’s mainly useful for making prototypes. I can’t see it being used for the bulk of goods which people make use of in their home or make use of in industry. But making prototypes, possibly artwork, stuff like that, yes. I mean, the only instance where I think 3D printing techniques are likely to be useful, maybe, is possibly in the construction industry.
You have to think of what is the nature of these 3D printing techniques at a deep level. Go back to when the printing press was invented. Why was that so much of an improvement in productivity? It’s because it transferred information onto the product in parallel. The whole printing head of Gutenberg’s press came down and formed all the letters at once, and that was the essential feature of printing which made it far better than handwriting. In fact, all the letters were done in parallel and that was a general feature of printing presses, that they did things in parallel.
Now there’s a set of technologies which have had a big impact on the world, and the technologies which have been particularly effective and have had huge improvements in productivity have been ones which harness parallelism. The printing press was the first of those.
In the late 18th and early 19th century, two other techniques came about that were significant. One of them, obviously, was the parallelization of spinning in a spinning mill where there’d be hundreds of spindles looked after by each worker instead of one spindle per worker.
The other less obvious to people, perhaps, was the mass production of cast iron goods. If you go to Edinburgh or Dublin you will see railings around buildings which are cast iron railings. You’ll see balconies on buildings which will have cast iron railings. This was an invention of the late 18th century which enabled complex iron objects to be made in a single action by pouring molten iron. The single action was brought about by the fact you had a mould and the mould transferred information onto the product to all points on the product at once. So, it had a big impact also in domestic production, for instance, cast iron stoves and things like that, far higher productivity than a smith using a hammer to carefully beat something out.
So those were two key inventions that sped up productivity in the industrial revolution and they all depend on parallelism.
If you look at the 20th century, what was the key invention that enabled mass production of cars? It was the use of large presses which would enable the forming of an entire chassis or the entire upper body of a car with a single impression of the die that greatly improved the parallelism and increased productivity.
Then you look at the semiconductor industry. Why did the semiconductor industry have such a high productivity compared to the prior computer industry? If you look at how computers were being made in the 1960s, they were wired up using a technique called wire wrapping. If you looked at the back of a computer board in the 1960s and 1970s, there were masses of pins and masses of wires and engineers in the factory had to wire the computer up one wire at a time. They were given schedules, connect this point to that point, this point to that point. And they had a wire wrap gun, and you put it on, you pulled the trigger, and it wrapped the wire round. You went to the next one, pull the trigger, wrap the wire round.
With the invention of the integrated circuit the entire wiring was achieved by printing. All of the wiring was done via a single photolithographic process. It was done in parallel rather than serial, hence the huge improvement in productivity. That’s the essential basis of the improvement of productivity.
When you look at what’s called 3D printing now, it’s not printing in the old sense. It’s more 3D scribing. There is a write head which is like a pen, which goes back and forth, back and forth, back and forth. It’s not a parallel process and as such its not one which achieves a high degree of productivity.
AO: Just one more thing on that. Often people talk about 3D printing, say, people on the left, and outside the left actually, insofar as its effects on social relations rather than necessarily on productivity. Do you think that it has any relevance in that regard?
PC: Okay, what effect will it have in social relations? You’ve got to think of why were the spinners, the spinstresses of England in the 18th century, why were they displaced by industrial looms? They were displaced because they couldn’t compete spinning one thread at the time with a mill spilling spinning 100 threads at a time. Now there’s no way that people at home with a machine that squirts out a thread of ink which slowly hardens will be able to compete with an industrial process that stamps an entire object out, or carries out injection moulding of an object. Not just of one object, but in a factory producing injection moulded plastic a whole bunch of them in parallel are injection moulded and knocked off. You can’t compete with that.
AO: 2 – Let’s go on to some other technologies. The next question is what is the relevance of the internet of things to Towards a New Socialism?
PC: Mainly, I think, that it makes the detailed tracking of products and the detailed monitoring of individual production machines easier. That’s what the industrialists in Germany are interested in, for ‘industry 5G’ as they call it, because it enables the individual machines to be monitored.
To the extent that large-scale dextrous robots are in use and that these are locally programmable then that is a factor. But this is not necessarily qualitatively affected by the internet of things.
My view is that this is heavily oversold from the point of view of its industrial productivity and what is significant about advances in the internet and communications technology in general is the person-to-person communication that it enables.
AO: 3 – The next question is what is the relevance of recommender systems technology to Towards a New Socialism?
And just before you answer, if our audience doesn’t know what a recommender system is, that’s what you encounter on Amazon or Netflix where it says ‘people who buy this, also buy this’, ‘if you like this film, why don’t you watch this film’.
You have books like People’s Republic of Walmart, which is very good but, it talks a lot about technology like this. And the idea is, in relation to planning, that such systems can often know consumer patterns better than the consumers themselves. And I’m asking this in the sense of do recommended systems have any serious role in facilitating planning economic planning on a large scale?
PC: I’m not sure, because why are recommender systems of interest to private business? It’s because each private business wants to sell as much as possible to each consumer, and if they can derive information about what that consumer has bought in the past they can present ads which will increase the prospect of a sale. But that’s not necessarily going to be something that’s of interest in a socialist society. You’re not trying to maximize consumer demand. You’re trying to maximize the satisfaction of people’s wants, not to stimulate wants.
From the standpoint of knowing what to produce in aggregate, what one individual is likely to want is not particularly relevant. What is relevant is what is the aggregate demand for t-shirts, what’s the aggregate demand for size 8 boots, what’s the aggregate demand for work boots as opposed to mountain boots? Those are the things you have to balance. And whether a person who likes yellow work boots is going to go for red or brown mountain boots is not really relevant provided the total number of them is right. And that total number can be obtained from what people are actually purchasing. You don’t need to break it down into what one individual is likely to want because it all comes out in the wash of averaging
AO: 4 – The next question, I think, continues on from that but there might be more substance here. What is the relevance of neural networks to Towards a New Socialism?
For example, I saw something today that some research team has used a deep learning system which has figured out how to fold a protein given a sequence of amino acids. And just it popped into my head, using that as a kind of metaphor, could we use a similar kind of system to work out what the shape of an economy is from a list of inputs and outputs? Commodities and so forth. The shape there also including information about where things would go, what regions, what kind of people would get what. Answer in any way you want.
PC: At a certain level, yes, they are very similar. And that’s because modern neural network systems are essentially linear algebra, or tensor algebra, systems with certain levels of filtering which put signals through sigmoid curves, etc. Now a lot of other things turn out to also be very similar to that. It turns out that the analysis that Google has to do to link up meanings of phrases in order to get relevant documents is again a branch of linear algebra. And well before this back in the 1950s and 60s it was shown how to treat economies in terms of that same type of linear algebra. So yes, neural networks and thinking what the shape of an economy would be are both problems in high dimensional linear algebra.
I’ll raise another thing that neural networks do. They’re working with a high dimensional feature space and they try and learn techniques whereby they can map this onto lower dimensional manifolds, and within this lower dimensional manifold you can do various types of clustering and grouping of things.
It’s possible that when you’re thinking of economies you can also map things onto lower dimensional manifolds. Most of the standard treatment of planning is … well I’m oversimplifying. The full mathematical treatment of planning is in the native space, say, of all the products. And it’s expressed as a linear algebra problem same as the way Marxists deal with the transformation problem or compute the labour values and things like that. However, in practice people work with reduced dimensionality systems that are provided by input-output tables, which aggregate similar things together. Now that is an ad hoc aggregation done by national statistical offices. It’s possible that if you had the raw data of every kind of product, and every kind of product code, you could apply dimension reduction systems which were more sophisticated than the ad hoc ones that are done by national statistical offices.
National statistical offices, for example, will allocate maybe a four- or five-digit code to a type of product, and it’s like a Dewey decimal system coding. The first two digits tell you the first hundred categories into which products are divided, the next digit divides them into a thousand categories, etc, and what is done is that you just drop some of the digits in order to get a simpler model. But that’s not necessarily the best way to do it. The kind of dimension reduction that Google do on words is more sophisticated than that, in that it learns patterns of the words and does a dimension reduction onto a subspace which represents the meanings. Now just looking at it as an abstract maths problem, that kind of technique may turn out to be practical in economic applications but it would be a research program to see whether it was applicable or not. You can’t tell beforehand but it’s not implausible that it might be useful.
AO: Can I just ask you a follow up on that? Firstly, just to make this a little bit more concrete for viewers who wouldn’t have a very high level of mathematical training, what you’re talking about is input-output tables that are typically used, they might use categories such as agriculture, textiles, things like that, or it might be more finely graded. What you’re saying is that if a suitably advanced neural network, deep learning, whatever you want to call it, algorithm was able to look at all of the economic data, that that algorithm might actually come up with different categories to use in order to have this more aggregate picture of the economy.
PC: Yes
AO: Now, that’s an explanation for people, but my question is what is the practical significance do you think, for planning, of having better categories? What do better categories actually mean in that instance?
PC: Well when you’re defining categories as being better, they’re defined as better with respect to some goal or metric. In Google’s case they want they the subspace they project the words onto to be the commonplace meanings of the words and associated meanings. The question then is what might be the goals of planners in wanting to do this.
I mean what seems to me the obvious one would be to convert the representation into a representation that was intelligible to the public, for the public to democratically decide on the broad outlines of the economy, the major strategic developments of the economy. So categorize products according to the questions which were coming up about how you want to restructure the economy and society. But you can’t say that beforehand unless you know what the questions are. These change from time to time. I mean, at the moment it looks as if things to do with carbon emission reduction would be a major factor. But we know that in times of war, factors are making weapons, providing basic foodstuffs, etc, so the categories that are relevant depend on the circumstances.
What you’re suggesting is, I mean it’s not something I’ve thought of before, but it’s the sort of thing which would be a good project for an advanced student to start looking at actual input-output data of the more disaggregated types that the US Bureau of Economic Affairs. for example, publishes. And to see whether aggregation techniques that have been shown to work linguistically would work well to deliver new categories to categorize the branches of production in the US, for example.
AO: I understand what you’re saying there about how it’ll be useful to have a way of categorizing production in society in a way that the population at large, without any kind of specialized training, would be able to actually meaningfully engage with the process. So it wouldn’t just be a technocratic process with the planning bureau formally not being in charge, but informally they would be.
In terms of those aggregate categories, how do they actually affect the planning process itself? Say if you had very bad aggregate categories, how would that actually manifest in real terms in the economy? What negative effect would that have? If we could put it that way.
PC: These are very abstract questions! I suspect the first guess at this is if your categories were bad the degree of adjustment you might have to make to some industries might appear to be too big than if they were well designed categories.
AO: If I understand you correctly, it’s like saying if we make a change in this part of the system, having good categories would say this input will produce such kind of effect in that other part of the system. And if your categories are effective you’ll be modelling that in a way that’s quite accurate. But if you’re not doing that properly, you’ll say if we produce this much of something here it’ll produce XYZ amount there. Actually it does twice as much, and then the plan is actually not matching up with reality.
PC: Yes. I’m trying to think of what it means properly in maths and I can’t off the top of my head work out what it is.
AO: Well, we can probably come back to that some other time because I think it’s an interesting topic.
5 – There’s one last technological question, it’s about quantum computing. A bit of background for people: a lot of people have heard about Moore’s Law. That’s effectively computing power becoming exponentially better on a constant basis. That has been slowing down or possibly coming to a halt. Now we have quantum computing.
So does this have any relevance to economic calculation? And if you wouldn’t mind, could you just briefly explain to people what quantum computing is? If they want to know about it in more detail they can look it up.
PC: There are different aspects to this. Let’s take the way you introduced it in terms of Moore’s Law. If you keep scaling things down it’s clear that at some point you’ll reach the scale where you have gates which only switch a few electrons at a time. If you get to that point, the reliability goes down just because of shot noise etc. So that’s one thing which will tend to limit Moore’s Law. As you push it to lower and lower feature sizes, the quantum noise due to the quantization of charge into individual electron charges will become more significant. Now, it’s been shown you can use what are called Coulomb blockade transistors, and switch transistors with single electrons. So it’s not impossible. But noise effects are still likely to be significant unless you cool the things a lot, so that’s a factor.
Another factor is that as you scale things down you hit a thermodynamic limit. There’s a certain amount of energy which is lost every time you trip a gate, i.e. switch it from one state to another. The energy in terms of electrons is the number of electrons times the voltage you operate the equipment at, so you can say ‘okay, there’s a certain number of electron volts involved in this’. But you can also approach it from the standpoint of pure thermodynamics and look at the measure of information in terms of entropy and how much, in principle, thermodynamic energy must be released when you switch one bit.
And the problem is that the gates which we operate with in conventional computers are things like two-input AND gates. They take in two bits and output one bit and therefore, crudely, it’s not exactly this, crudely they destroy one bit of information. You put in two bits of information you get one bit of information out. And because of the relationship between information and entropy, you can show that this must release energy equal to log2(kT) units of energy, joules per bit lost. That’s called Landauer energy after the physicist Rolf Landauer who worked it out in the 60s.
If you shrink things down, just the Landauer energy that a chip with enough components on it is going to be dissipating means that it’s hotter than you can actually remove the heat. Effectively the rate you can remove heat from a semiconductor chip is set by: suppose you create a lot of very small parallel channels and fed pressurized water through it and allowed the water to boil to steam as you cooled it down. That’s about the maximum cooling you could achieve on a chip by turning water into steam. And this is of the order of maybe 10 kilowatts per square centimetre. If you go above that you just you can’t remove the heat. I’m going back to calculations I used to give my students in the 1990s, you can show that above a certain speed and a certain degree of shrinking of Moore’s law just the Landauer energy will be such that you won’t be able to cool the devices.
These are classical computers. They’re classical in the sense that they are the way we’ve been building computers all along. And they work with non-reversible logic, because the logic loses information as it goes on. Now, in principle, at the most abstract level, when Feynman was introducing the idea of quantum logic gates, what’s the most fundamental feature of them is that they are reversible gates. That any quantum logic gate has as many outputs as it has got inputs. So in principle there’s no loss of information. And quantum systems have to operate under what are termed unitary operators. They basically all take the form of a rotation in a higher dimensional complex space, but it’s a rotation in which the total amplitude summed over all directions doesn’t change. And in principle that means that quantum computing doesn’t dissipate energy.
In principle, but this is a very abstract principle and we’re so far away from actually having reliable systems that this aspect of not using up energy – which is what was originally said to be one of the advantages of it – is not practically why people are pursuing it. Practically, people are pursuing it because for a certain subset of problems, and it’s quite a small subset of problems, so far quantum computation greatly reduces the computational load of performing a calculation.
Feynman, when he originally proposed it, proposed it from the standpoint of simulating quantum physics problems. He wanted a universal quantum simulator which could be used to simulate a quantum physics problem. And the problem with simulating a quantum physics problem is that the number of components in the matrix that you’re pursuing grows exponentially according to the number of independent particles – let’s say particles, roughly – and that becomes prohibitively costly to compute on the computer. Whereas if you could actually put a set of quantum elements of some sort into the right superposition of states, you could just allow it to evolve and statistically sample what the outputs are to get a realistic model of the other quantum system that you’re looking at.
Well that was the original application of it. Subsequently a certain limited number of mathematical problems have been shown to also be amenable to this. But it has to be said, the number of problems that have been shown to be amenable to solution more efficiently on quantum machines is still quite limited. And it’s not like inventing an ordinary algorithm, it requires a very much higher level of skill and specialism to invent a new quantum algorithm.
AO: As this relates to economic planning, could we imagine that this would have any applicability? Not now, but in the future when there are many more qubits. Or do you think that even if quantum computers were a thousand times more powerful that this wouldn’t matter?
PC: The thing is that the economic planning problems are relatively tractable anyway with classical computers, so it’s not clear to me what the point of attempting to use quantum computers for it would be. These aren’t exponentially hard problems.
AO: One last thing on this. Let’s say if a society was already at an advanced stage of communism and had implemented Towards a New Socialism. And maybe – I mean this is really far in the future, abstract stuff, but just out of interest – wanted to move more deeply in that direction. And, say, wanted to use more in-kind planning, which would be more computationally expensive than computing things in terms of labour time, could it then be something that we would think about?
PC: What I’m saying is that even in-kind planning is relatively tractable. The iteration procedures to converge at an answer for in-kind planning are not of very high complexity order.
AO: And in-kind planning not using labour time as a universal measure?
PC: Yes. They’re not of very high complexity order. They are pretty straightforward linear algebra. They involve, in principle, doing matrix inverses. But there are shortcut ways of doing matrix reciprocals which enable you to get good performance. If you use a modern parallel programming language like Julia, they’re built in they come for free, and they give you decent performance. They’re the sort of things that people working with super computers are used to using already. Compared to the problems that are routinely being solved on GPUs and on super computers it’s not a big issue.
AO: I thought one thing that was actually quite amusing was the update that you put on the website where people could find Towards a New Socialism for free as PDF. That the representative high performance computer was 108 times – that’s 100 million times – more powerful than the example that you gave in the book. Which I think is pretty funny, because I don’t think anybody could say that economies have become 100 million times more complex.
PC: No, no.
AO: 6 – We’re going to go in a different direction now after that technological discussion, and talk about second-hand goods. How would second-hand goods be passed on?
There are goods of sufficiently low value which one could imagine people just dropping into free shops, or leaving at the side of the road (as they do in Berlin). However, some goods are too expensive for that to make sense, for example, furniture, vehicles, musical instruments, sailing equipment, jewellery. I might buy a car for five thousand euro, let’s say, and five years later sell it for 3500 euro. What would an individual or commune do in Towards a New Socialism?
And just to explain this to people. The idea is that labour tokens are not transferable between people. I go into a state shop, I buy a car, and then those labour tokens are eliminated. I can’t then sell the car to somebody and they transfer tokens to me. So how would second-hand goods be managed?
PC: The first thing you’ve got to focus on is not little consumer goods, it’s ships and airplanes. If an airplane is no longer being used at the moment on the Dublin to London route, what happens to it? If Ryanair has leased it from one of the Irish companies that lease airplanes it goes back to the leasing company and the leasing company leases it out to somebody else who is flying from Belarus to Iraq and flying refugees into Belarus.
What’s important there is that even within the capitalist system, the actual ownership of the key means of production is not in the hands of the final user, which is the airline, it’s in the hands of some higher body, which is a leasing body, which redirects it to some other use when one organization no longer needs it.
So that is the most important feature for a society as a whole, that means of production which potentially could be used elsewhere don’t just sit idle on an airfield, but are actually transferred to where they’re going to be used. What’s important here is having national organizations which keep a national registry of the different classes of means of production and reassign them when they’re no longer in use.
Now, it’s a much smaller matter, consumer goods. An airplane: 20 million, 100 million; your used radio: not so much. There’s already a whole system for dealing with that in the form of charity shops, Freecycle, and similar giveaway systems.
AO: Sure. Let’s put that to the side, though, and look at an intermediate group where it’s something which is costly enough that people wouldn’t want to just give it away for free, but it’s not so expensive that it falls under the previous discussion about means of production.
PC: Give me an example.
AO: For example, a vehicle, a car, it might be furniture, it could be a musical instrument like, say, a keyboard that costs a thousand euro, or a sailing boat, or a piece of jewellery or a car.
PC: Hold on, you’re starting to move up from everyday items to luxury items when you start talking about sailing yachts.
If we just take furniture, well furniture is the most obvious system where people give it away, to the British Heart Foundation or whatever the equivalent in Ireland is, to recycle or things just get scrapped. A lot of people actually get rid of the furniture they don’t want on Freecycle.
Then look at cars. Increasingly people don’t own cars, they lease them. Why do you assume that in a socialist society people would directly own the cars, rather than lease the cars from the state?
AO: Well I suppose, yes, if they lease them then that’s the question answered. As for furniture, if I was playing devil’s advocate I’d say to a large extent people give it away, or they just dump it, but there are a lot of items – it could be furniture but it could be many other things, again musical instruments would be something – where somebody would go onto a website like, I don’t know what the websites are internationally, but DoneDeal, eBay, etc, where you might have bought a keyboard for a thousand or two thousand euro and you want to upgrade now. But you don’t want to just give it away because it cost a significant fraction of your income. So you would like to get some of the value back. Is there a way to address that?
PC: These are attitudes and views of things which come from buy and selling being the norm. If the norm is for things to be given away, you will in your time also have got lots of things which other people are giving away, and it will it’ll seem the norm.
AO: Okay so then really what we’re talking about is moving more towards the norms of a gift society.
And do you think there would be any issues of transition there? That it could take people time to adjust to that mentality? That there could be an intermediate system or that we could just go straight into it? I guess that would depend on so many things. In revolutions, for example, people’s mentality can change pretty quickly.
PC: They do and there are elements of this that people are used to. These already exist, I’m not making this up. You go down to any shop in Britain and I’m sure in Ireland too and there are shops selling things which people just gave to the shops because they thought that the charity was worth it.
AO: For sure.
7 – This is about the number of commodities in an economy. In your presentations you often use the figure of 10 million commodities in a complex advanced country. However, does this accurately represent the number of commodities in an advanced country today? Does such data exist? If it were 10 billion rather than 10 million would that make much of a difference?
PC: There is data. You can, for example, find out how many distinct product lines Amazon or Alibaba have, and it is above 10 million. The 10 million figure is a figure for the Soviet economy in the 1970s. So it’s a big country, but China is bigger. So Alibaba sell a fair bit more than that.
The figure is bounded, though, by the number of people in the economy. Generally, any product requires the collaboration of several people so that the number of products is not going to grow above the number of people. So, yes it can be big but it’s going to be of the same order of magnitude as the population probably a bit below an order of magnitude of the population.
AO: If I were to summarize, you’re saying if you have a country with a billion people which is an advanced economy, then the number of different commodities within that country would be approximately on the order of one billion?
PC: I would say probably of the order of a hundred million rather than a billion. I’m saying roughly an order of magnitude less because of the fact that every industrial product actually requires collaborative labour of several people to produce it.
AO: I think the answer is probably no, but do you think that makes much of a difference then? I mean, really, in that case if we’re looking at China, which has 1 billion people, roughly, which is the largest country, between 10 million commodities and 100 million is one order of magnitude so I don’t think that would make a difference.
PC: I mean, a country with a billion people can afford much bigger computers than a country with one million people, so, no, I don’t think it’s a problem. I think the computing power they can throw at the problem grows at least as fast as the problem. I mean, the Chinese produce the most powerful computers and the figures are staggering for their highest performance machines. So I don’t think it’s an issue.
AO: And given that with the advance in computer technology that’s happened over the last say few decades it seems that computing power is well ahead of where it needs to be anyway.
PC: Yeah. I mean, Jack Ma, who has practical experience of this in China, is certainly on record as having said that he thinks the whole economy could be planned without money in China.
AO: That’s very interesting.
8 – Another question, probably the last one, this is about basic research. In calculating the integrated labour time of a product, how can basic research be taken into account?
Whereas it might be clear, or clearer, say, to Nissan how much labour time they dedicate to developing the Nissan leaf electric engine in their Nissan leaf engine R&D team, it wouldn’t be clear how much labour time went into, say, a vaccine which drew upon 20 years of basic virology research in universities across the globe. So how could the integrated labour time of basic research be taken into account in a product?
PC: This is one of the points Marx made a long, long, time ago. That as society develops, the greater part of the productive power that’s put into making things comes from general scientific and technological knowledge which has been built up over generations. And it is not something private, it’s collective knowledge. And because it’s not something private, it doesn’t have to be paid for in the price of an individual item.
When you buy food mixer, it depends on work that Faraday did on uncovering the relationship between electric current magnetic field and force. But you’re not actually paying for Faraday’s research. They all rely on it and couldn’t do it without that research, but you don’t have to include that. The reason you don’t have to include it is that labour time calculations are only about what is being done with current activity, current allocation of activity. Nothing you do now can affect what happened in the past, so there’s no point including that.
Now, you might decide that research and development charges should be levied for products. So that when medicines were being charged to a hospital, assuming the medical care is free, should you charge the research and development costs of the medicine? Almost certainly not. Because if you did, you would deter the hospital from using the most advanced, most recent, medications which have involved recent research.
That’s the way it works in the capitalist world. The newest drugs are the most expensive and there is therefore in a free public health system a bit of a deterrent from using these latest drugs. But from the point of view of social welfare maximization the labour that went into the research into, let’s say, the Pfizer vaccine, was all done in the past. And in terms of delivering as much of the vaccine as possible, all that counts is the labour that is required to make another dose of it. So it’s not necessarily rational to include the research and development cost in it, to the extent that it will minimize social welfare if you do. Because it’ll overestimate what it’s going to cost to treat people.
AO: If I could just recapitulate what you’ve said to see if I understand. Would it then be more a matter that society would decide in broad terms how much labour time, integrated, it would like to dedicate to basic research? Based on historical precedent, based on a more qualitative understanding. You know, we want to dedicate a lot of society’s resources towards basic research, and we’ll let them effectively do their thing rather than monitoring precisely how that connects to all sorts of products which emerge out of that later.
PC: Well, you want to motivate researchers to research things that are going to be useful and be put into practical application. So there is certainly going to be some kind of incentive for them to do that.
But, yes, the basic research has got to be met out of the current labour budget of society. That’s what happens now except, to the extent that it is done by private firms, the firms advance capital on that and expect to get a return on their capital. To the extent that other parts of the research and development are paid for by the state, then the information isn’t private information and becomes generally useful. For a lot of the basic computer science and communications technology that we’re relying on at the moment, the basic research was all publicly funded.
AO: That brings us exactly to ten past nine, so we can leave it at that.
PC: Who else are you interviewing?
AO: Well, I’ve got a list of people. I’m going to wait till after Christmas at this rate but I’ll tell you who. I’d like to interview Pat Devine, I would like to interview Robin Hahnel, because he just came out with Democratic Economic Planning, and some other people. I’d like to interview Varoufakis as well about his book Another Now. A handful of other people.
PC: Okay. I see the kind of people you’re asking, yes.
AO: Yeah, I want to stick on this topic for now because I think it’s very much under-served.
PC: Yes.
AO: I think, for example, there’s a lot of material out there about cultural issues, and so forth, but in terms of these things I don’t really think it’s being discussed and I’d like to try to get a discussion going.
PC: Can I suggest someone you should also speak to?
AO: Yes please.
PC: Philip Dapprich. He’s German, currently working at the University of Berlin, and he particularly focuses on how to use Kantorovich-style opportunity costs in economic planning.
AO: Okay definitely.
PC: Another person just to get in touch with is a guy called Tomas Hardin, he’s a Swede, and he’s doing a lot of research on this. He’s got a whole group working on it. There’s a growing number of people working on this. There are people who could give you useful contributions on this.
AO: In terms of doing another interview, would you like to do that again in two weeks or will we wait until after Christmas?
Paul Cockshott: After Christmas, wait till after the holiday period.
After the Oligarchy: Brilliant. So, great to talk to you again. Fantastic discussion. I appreciate you taking the time to talk with me and we’ll talk again soon.
(Photo by Vincent M.A. Janssen from Pexels & Elchinator from Pixabay)
One reply on “Paul Cockshott Interview on Towards a New Socialism – Part 2 – Socialism & Technologies, Second Hand Goods, Research”
[…] After the Oligarchy, 29/11/2021 […]