Capitalism is a paperclip maximizer

Capital is not what most people conceive of it to be. Capital is commonly (and correctly) viewed as a "store of wealth", or as "the crystallisation of labour-time" - but capital comes in different flavours, and when you deal with capital in the form of goods which provide an incremental benefit to the user, what you have is indeed capital, but when you deal with money... it's up to you whether it's capital or not.

Often, people confuse money with capital. Money is an exchange mechanism which facilitates the transfer of value between parties through a common unit of exchange. Money can be viewed as a subset of capital (i.e. as a sedentary store of value), or it can be viewed as the destruction of the benefits of labour, as it converts an active instrument (i.e. the work I am doing, or the goods I have produced) into a sedentary instrument which provides no incremental benefit, unless it is spent.

So - how this relates to the article - capitalism is a system which currently optimises for money. When you take a step back and consider that by converting labour and value into money, you are effectively freezing the benefit of that labour away, accruing money as an end of its own ceases to make sense, as much of this value ends up lying in the ledgers of corporates, governments, or wealthy individuals, doing nothing for anybody other than perhaps dealing with some of the more fragile elements of human psychology (security).

We have let a means become the end, and have forgotten that the Purpose of all of this is to promote human happiness - I mean, what other possible Purpose can there be?

It's a faulty analogy. In fact, a paperclip maximizer goes against every single principle of capitalism out there.

The paperclip maximizer is bad exactly because it isn't based on capitalism; it will maximize paperclips regardless of whether someone values them or not! At _any cost_, it will maximize paperclips.

Markets, on the other hand, have the goal of providing products/services that people are willing to pay for. It adapts. The more people value it, the more capital it will attract, and the more resources will be allocated to it.

The article irrationally assumes the best effort maximizer has no competition and long life, which we have not seen in endless lower powered implementations, and there seems to be no trend so unless it starts going exponential...

There might be a barrier here, like speed of light or various mathematical limits that formalize the old saying of too many cooks spoil the broth. The required complexity of a uber-maximizer to handle the complex universe means its unprovably complicated and unreliable. Or you'd need a bigger computer than the universe to answer the halting question for a universe controlling control loop, or even something much smaller like a mere multinational corporation. Or another way to put it is a predatory competitor maximizer would inevitably be better at predation than its victim that focuses more on some arbitrary goal than on avoid becoming a meal, so in a hyper optimized playing field all you'd get is ever better predators eating each other and eating weaker non-predation focused orgs, or basically the modern business/finance system.

Wait, what?

Capitalism maximizes "giving people what they want". This is much closer to the benevolent AI models than to paperclip maxizers.

Economic systems are optimizers (each with slightly different objective functions).

One of the first things we learn about optimizers is that they are rarely globally suited to solving problems, and that the best technique is to identify and apply the locally optimal solution over the range of input.

TLDR: We need to admit that there is not a globally applicable economic system, and learn to apply different models where they are most efficient/least disastrous.

Capital doesn't maximize capital; it maximizes productivity. In absence of controls, capital naturally flows to productivity in order to be maximized by the holders of capital; but only because productivity creates wealth, not the capital itself.

People fantasize too much about capitalism. The definition of capitalism is: capital and income are expressed under the same monetary unit (like mass and energy could be):

income = capital x interest rate

The former alternatives was to put capital in a separate "namespace". In Middle Ages, only nobles could have land and land was the only capital that could exist ("usury" was forbidden, you know). Then came mercantilism, that defined wealth as a quantity of gold that one had; and so on.

Capitalism will certainly pass, but hopefully not going back to restricting access to capital. I imagined something like a "complex" interest rate, with the imaginary part representing natural resources, while the real part is the human effort that is the only think currently measured by monetary unit. But bimetalism failed everywhere it has been adopted, I think that the imaginary part of a price would become dominant since natural resources are becoming the limit of economic output.

What we really want is a happiness maximizer but no one knows how to make one of those which leverages the innate characteristics of people. Capitalism seems like a pretty good system to pragmatically approximate this happiness maximizer.

Yes, there are negative externalities to capitalism including possibly maximizing short term happiness at the expense of long term. Is there something better?

Framing this in terms of the paperclip maximizer made this idea appear more interesting than it turned out to be.

War can only destroy capital, perhaps the most breathtaking destruction of capital in history was WWII. So War would be antithetical to this hypothetical capital maximizer and it would be it's among its top priorities to avoid it. Similarly, the market system based on human rationality is far too volatile for our capital maximizer, it would look to replace that ASAP.

Thinking about this some more, it would seem that if we really have a Capital Maximizer then the system's process toward its objective (creating capital) is highly inefficient. If your goal is to create capital in the world you need better infrastructure, after all what capital is more useful than that which can be easily used to create more capital? Long term, if you're a Capital Maximizer trapped on this planet, you only have a small finite amount of resources than can be converted to capital, you would want to spread out to other planets very soon. These are obvious things we aren't doing very well.

The root problem here is not economics, it is basic human nature. Humans have forever been seeking to increase their power base at the expense of others, and been corruptible by greed. This is not new, and it is not limited to capitalism. It's basic human nature, and it's one of our key weaknesses.

Capitalism depends on price theory, which predicts that making too many paperclips will cause them to be in glut and the price will drop. So a paperclip-maximizing AI might oughta be built to take that into account.

Corporations can be usefully described as 'profit-maximizers', but capitalism itself doesn't have a coherent behavior - you can't describe it as any kind of AI.

I disagree with much of what was written, but I find it hard to address what seems, to me, more like a long analogy than anything else.

"Capitalism" doesn't maximize "capital", in any sense. The word "capital" generally refers either: to some asset that might be used in production, such as a piece of machinery; or structures that might render services over a time, like buildings; or to some sum which might be used in buying either of these, through financial services.

Now, "capital" wouldn't be maximized by "capitalist" firms. Why should they? Firms produce additional capital units so long as they can be sold above production cost, but they won't produce indefinitely. They are concerned with their profits, not with the amount of product. They aren't even concerned with the products' value, so long as it is above production costs. Hence, there's no incentive to inflate the capital good prices, which would be necessary to maximize the amount of money "capital".

Also, an it really be said that capitalism is an adaptive intelligence system? That, I'm not knowledgeable enough to tell. On the other hand, human interaction is unstructured enough that it seems unlikely that there's any goal to such supposed intelligence, or that it has capital-maximization as it's goal. Rather, I see a collection of different goals, each corresponding to different agents, and some of which might even be contradictory. The noisy neighbor, the ambitious monopolist, or the NIMBY all seek to maximize their private benefit, but often beyond social benefits. Hence I find it hard, whenever markets fail through the presence of either externalities, imperfect competition, or adverse incentives, that it ought nonetheless be the work of some powerful demiurge.

Yet, in so far as it puts into words an anxiety that so many feel, I can't say much else beyond that I don't believe it. The author certainly proceeds by analogy when he likens "capitalism" to an artificial intelligence whose goals themselves mean nothing to humanity at large; and mentions alienating "monuments" to capital, as if to liken it to a 20th century pharaoh, or to cruel tyrant. But since so many feel this is really the case, I have nothing else to say (other than that it's a very vivid picture). I disagree, for instance, that the majority of goods are worthless, as they ought to have some value in order to be sold. Even Marx reckoned this much, and thinking otherwise seems like relying too much in the stupidity of others (which should never be trivially admitted). That would be the case if capital were built beyond all possible utility (say, like a bridge between London and NY). And I don't think it is necessary to make analogies to speak of wealth and income distribution issues, which are real problems in themselves, and which the majority of people are already acquainted with. Rather, I think that such analogies make problems much larger than they ought to be, by giving some monstrous character. Would be it be easier to tackle inequality by conceiving it as a result of an incredibly power artificial intelligence? I don't think so (even if the system is rotten, we'd be waiting on a constructive proof of one that isn't).

Nailed it.

Oh good. The Americans are rediscovering Das Kapital.

Forgive the snark, but when I first heard about paperclip maximizers, I thought, "Oh, you mean like capitalism." That these two things have a likeness is the reemergence of an idea that had been buried in order to make capitalism look good.