Software vs Capital

John Ohno on 2018-05-31

I have periodically said “software has nothing to do with business” as a criticism of tech journalism that focuses on firms and corporate politics over tech itself, or that considers all software to be products or potential products, or that considers success to be defined primarily in terms of money. I get a lot of pushback, including from intelligent tech journalists who I feel should understand my criticisms, so I’d like to expand on my quip.

When I say ‘luckily software has nothing to do with business’, I do not mean that business is irrelevant to software in practice. What I mean is that software is uniquely positioned, compared to all other engineering, to be free of economies of scale: it is cheap to develop and has near zero cost of reproduction. Business has the capacity to be irrelevant to software without any radical changes to how business is run or software is developed in practice.

In other words, software development can be done outside a capitalist context to a much greater extent than material forms of tinkering. Coding is not an expensive hobby: only the free-time component applies. Even a fairly weak form of UBI could give a big boost to free software, because most available software (still) is not developed by businesses or for business purposes but as a hobby by hobbyists who expect nothing in exchange.

I am not claiming that software engineering for the money is somehow illegitimate. I consider the entire ‘work to live’ arrangement to be illegitimate. Software engineering, like writing, benefits more from the abolition of that arrangement because of the combination of low resource costs & a lot of people willing to do it for fun.

The purchase of a physical computer for the sake of developing software is much like the purchase of a typewriter for the purpose of writing a novel: it is a large one-time cost, after which only small incidental costs are incurred (paper and ink, or power and internet), and it is a cost that most people have already paid. You can use the same computer for a decade or two. Other technical hobbies (like knitting or electronics) have much larger ongoing material costs — and even near-ubiquitous situations like car ownership are much more onerous in this respect.

Amusingly, the very attributes that make it possible for something to escape from capitalist domination also make it particularly desirable to Capital. The elevation of sign or Spectacle over reality is like Capital engaging its death drive, since value is wholly imaginary.

The logic of markets depends upon the (strictly incorrect but historically mostly good-enough) notion that there’s some natural value to things, inherent to them and stable, which can be determined by looking at what people are willing to pay. When use value is the primary factor, sure. When labor value is the primary factor, sure. When the primary factor is speculative (what other people are predicted to be willing to pay), cost becomes totally unstable. This instability is mostly unpredictable. Of course, anybody who bets on instability and wins is going to attribute that to skill or insight, and since losing money is harder than making it once you’re rich, they can remain unchallenged in this attribution. But, there is no major insight factor — merely survivorship bias. The winners win bigger and the losers lose more as we become untethered from the world and get stuck in never-never land, where the tinkerbell effect is the only law.

In other words, speculation is the end of those elements of markets that have ever been generally beneficial, but since they are overconfidence-generating machines, Capital is pulled into their gravity. I think we’re past the event horizon here already. IP is a lot like speculation: the value of social constructions is usually dominated by social constructions, as production & marginal costs approach zero.

“Software is eating the world” is maybe better phrased as “The Spectacle is eating the End of History”.

Why hasn’t the essential spookiness of software eaten capital yet? I think a big part of this is an accident of history: the ascent of the web as the highest-profile hypertext system at the same time the ban on commercial use of the internet was lifted. The use of host-oriented addressing in URLs, while non-problematic in smaller networks where the number of viewers has a low upper limit, produces problems when the number of viewers becomes large. Making even a static website scale to extremely large numbers of requests requires expensive hardware, because all the requests must be tricked into thinking they are connecting to a single machine with a single address, even in situations where a single machine of arbitrary power couldn’t possibly serve all of them. This opens up a convenient opportunity for rent-seeking: anyone with sufficient capital can provide the expensive hardware necessary for working around the web’s broken scaling, and so we get hosting services, data center services like AWS, and cache services like CloudFlare.

Even free/open source software, nominally isolated from the constraints of capital, typically makes use of host-oriented URL schemes and the host-oriented assumptions that come with them. Projects get hosted by github, or pay for domain names and get hosted in data centers. In this way, these projects shore up the finances of the merchants of scale, slumlords of the internet.

I expect that if named-data networking systems like IPFS, sneakernet-capable social network systems like Secure Scuttlebutt, & alternatives to for-profit ISPs like mesh networks & long-range wifi become popular among the numerati, we’ll see an acceleration of Capital’s slow sinking into masturbatory irrelevance in the domain of software & computer technologies. Even if you don’t have personal reasons to support these kinds of technologies (such as a concern for privacy and autonomy), I recommend supporting them on the grounds of siezing the means of computation.

You have nothing to lose but your blockchains.