07.09.2017
Computer says no
Paul Demarty asks why we insist on treating our computers like gods
I have chosen as my title for this article a catchphrase from Little Britain - the late and mostly unlamented comedy vehicle of David Walliams and Matt Lucas.1
The particular sketch centres on Carol Beer, a passive-aggressive bank clerk played by Walliams in drag, who, faced with any request, taps at a computer keyboard and returns the inevitable outcome - “computer says no” - typically with an exultant cough as punctuation. Little Britain is justly derided in our more sensitive times as a moronically ‘provocative’ exercise in stereotype-mongering and taboo-breaching, but this particular sketch is quite well observed. For the role of the computer in Carol’s life is expansive indeed - it gives her the strength to get from one end to the other of her working day, providing the confidence to say no to all those seeking mortgages and unsecured personal loans ... It advises her to skip the office party, where the object of her unrequited affection will be carousing with a certain Melanie. “Computer says no” to Melanie and the others; and implicitly says yes to Carol.
I suppose I am one of those to whom the computer says yes. I work as a ‘software engineer’, which in the end is a more exotic, better paid, but less skilled, variant of plumbing. I have attached a lot of pipes together - for that is what the internet is, a bunch of pipes - for money, and by extension for the causes in service of which people are prepared to pay me to wield the spanner. Those causes seem, in the end, to have an awful lot in common with the worldview of Carol Beer. The world is changing - because the computer says so. Artificial intelligence, ‘the internet of things’, or some other fashionable buzzword, is about to change everything about how we live: the question is merely whether we shall be riding the wave or get upended and drowned by it.
But this in turn is merely a particular variant of a more expansive narrative about the effect of technological change on society at large. Computers - if you read the popular media and take as good coin the witterings of competing ‘thought leaders’ at TEDx and what have you - are indeed changing everything; it is merely on the technical matter of how, exactly, and whether we should be worried, and indeed what we should worry about, that the story differs. We are told that, as a species, humanity is denuded of its intellectual heft by smartphones and the truncated time horizon that attends to their usage - we are, it is supposed, unable to concentrate on anything for more than five minutes, thanks to the superfluity of stimulation available at any given moment. The age of blue-collar ‘real jobs’ is over - anything that can be done repetitively will sooner or later be done by a computer, controlling a robot in such circumstances that contact with the material world remains drearily necessary.
Given the deleterious effects our age has on attention spans, I feel obliged to state my argument at the outset, clearly, in the form the internet abbreviates as “TLDR” - “too long; didn’t read”. The idea of computers and technological progress in themselves having this or that concrete effect on human history is in the end a way of excluding human decision-making from effectivity on the course of human history. We are told that we should not hope for stable or gainful employment, “because of robots”; we should accept any given economic regime as merely the necessary excrescence and consequence of some level or another of technical development. But this is in the end a faulty syllogism - we take as a logical consequence what is merely a contingent result. Contingent upon what? Merely the earthly powers to which we are accustomed well enough - the class rulers of one sort or another, the people who gave us Donald Trump and the Iraq war and the Thighmaster.
We are therefore talking in the end about a species of fetishism, in the precise sense that Marx pinched the idea from the anthropologists - the imputation of magical powers to inanimate matter. We are also, as I argued in a recent article in the Weekly Worker, in the end doing something very similar to religion as it appears to Ludwig Feuerbach and, following him, the young Marx - the magical powers projected onto the inanimate matter are in the end our own, and we have merely built a new god, who, much like the old ones, is an alienated reflection of ourselves.2
The chrome sheen of new technology serves merely as a convenient occasion for a rehash the old lies about the inevitability of ... whatever the status quo happens to be. Excluded, by definition, is the power of humans, by their odd, meat-imprisoned whimsy, to act otherwise - to declare themselves the masters, not the slaves, of their dumb inventions.
End of something
There are two classes of eschatology that pertain to the accelerating computerisation of economic life - the utopian and dystopian modes.
These are in turn divided. There are two basic doomsday scenarios that come out of this mindset. The first is the one familiar from major science fiction franchises, in particular the Terminator and Matrix films. A computer system is designed that is ultimately so sophisticated that it comes to view humanity as a threat to itself, and sets forth to exterminate us (or, in the case of The Matrix, it turns out - despite its superintelligence - to be stupid enough to use human organisms as very low-powered batteries).
The classic advanced-in-all-seriousness version of this is the paperclip maximiser. Say you create an artificial intelligence system and tell it to maximise the production of paperclips. Before long, it will realise that the current number of paperclip factories is unambitious, and open more - and indeed convert all other factories to create paperclips. All those who attempt to turn it off will be killed, for to turn it off would be to reduce the number of paperclips being produced. In the end, there will be a shortage of iron ore, and the machine’s attention will turn to the human race at large, which carries traces of iron in its blood ... The idea is that ‘general AI’ will not face anything like the same physical limits as the human brain, and will not only outsmart us, but cannot be relied on to worry overmuch about the continuation of the human species.
I do not propose to spend any time refuting this class of fantasy, but merely highlight that it is put about, and represents perhaps the most extreme form of the fallacy under discussion - computer, very definitively, says no.
The less histrionic version is that basically all jobs are going to end up better performed by computerised systems and robotics, leaving 90% of the world’s population as essentially an unneeded reserve army of labour. This is all put very well in a book called Rise of the robots by Martin Ford, a well-meaning liberal tech entrepreneur type, trying at present to get his ‘thought leader’ scout badge. He lays out the spectre of ‘techno-feudalism’:
The most frightening long-term scenario of all might be if the global economic system eventually manages to adapt to [total automation]. In a perverse process of creative destruction, the mass-market industries that currently power our economy would be replaced by new industries producing high-value products and services geared towards a super-wealthy elite. The vast majority of humanity would effectively be disenfranchised. Economic mobility would become non-existent. The plutocracy would shut itself away in gated communities or in elite cities, perhaps guarded by military robots and drones. In other words, we would see a return to something like the feudal system of the Middle Ages.There would be one very important difference, however: medieval serfs were essential to the system, since they provided the agricultural labour. In a futuristic world governed by automated feudalism, the peasants would be largely superfluous.3
Ford’s conclusions as to how to avoid this disaster are pretty typical of his caste - he advocates a version of the universal basic income scheme (UBI). If you were to look that phrase up in the dictionary, you might almost find it defined as the bad conscience of Silicon Valley liberals.
The plausibility of the projection has several sources. Firstly, the post-war era of near full employment in western societies has given way to several crises followed by essentially jobless recoveries. Something, it seems, has changed. Secondly, computer systems have exponentially increased in power for a half-century, and as a result robotics systems are able in principle to replace an ever-increasing number of workers. Thirdly, there are an awful lot of people that have, as it happens, been replaced by robots, at least so far as their bosses see the matter. From all this, it is not an unreasonable leap to make the prediction that displacement of labour is on its own exponential curve.
How to fire people
We could take as an exemplary case of technology displacing labour one that, oddly, barely involves computers at all.
In the last 50 years, the shipping industry has changed almost beyond recognition. The old situation had vast piles of goods being sailed around the world in the holds of enormous cargo freighters. The freighters would dock somewhere or other, and a small army of workers would unload the goods and deposit them in nearby warehouses, for land transportation onwards; then the whole thing would be done in reverse, and the ship sent on somewhere else with new cargo. Add it all up, and you have a thriving dockside economy that can support tens of thousands, once all the accessory industries are factored in - multiplied for as many cities in the world as have a decent natural harbour.
In the last five decades, however, the intermodal shipping container has become ubiquitous - standardised in shape and size, it can be loaded directly onto trucks or trains with the intervention of only a handful of people and a lot of heavy machinery. The containerisation of ports around the world, therefore, occasioned the displacement of large workforces in the docks themselves; on top of that, the warehouses were now superfluous, since the containers could now be driven directly off the docks and onwards to wherever they need to go, perhaps tens or hundreds of miles away. By the time this process got its great representation in popular culture - the second season of The wire - it was basically complete.
It seems to fit suspiciously well; but there are some oddities with the story that complicate our naive understanding of automation. For one, of all the vast advances in the means of production made under capitalism, manufacturing steel boxes to a fixed and regular size seems a little, ah, obvious. Can it really be the case that nobody had thought of it before 1968? It surely cannot - the first international standard for such containers was defined in 1933. There was merely a long interval before adoption started to accelerate.
That adoption, starting in the late 60s, coincides suspiciously well with the end of the brief, anomalous period of post-war corporatism and full employment in the imperialist centres, and the first signs of the rollback of the gains made by the working class in the same era. Can this be a coincidence? Again - it surely cannot. The record is littered with the voices of dockside businessmen who, driven to their wits’ end by union power, were sold containerisation and the concomitant rollout of ever more sophisticated machines as a means of breaking the resolve of their restive workforces.
This peculiarity should not, in the end, surprise Marxists. At our Communist University, Neil Davidson remarked, rightly, that we should not view the ‘development of the productive forces’ as an entirely anonymous, automatic process - concrete historical agents are driven, and thereby choose, to develop them. The issue is illustrated nicely in the negative by Stalinist bureaucratic ‘socialism’, whose tendencies towards enormous waste stemmed in part from strong disincentives to deploy easily available new technologies.
Capitalism, of course, traditionally has no such difficulties revolutionising the productive forces. Why? In Capital volume 1, the process is spelled out nicely. The bourgeoisie stretches every sinew to increase absolute surplus value, but hits a hard, external limit in the worker. The worker’s body physically cannot be coerced into working longer and harder than they have calories to spend; the worker’s will, being that they are a conscious human, cannot be crushed to the point that they are reduced to Aristotle’s ‘speaking tool’ and instead they will resist.4
It is this, not the inherent attractions of mechanisation, that drives capitalists to automate production. But they might not need to - it seems the most profitable way to make cheap clothes is to pay armies of adolescents a pittance to stitch them together in an anachronistically labour-intensive fashion. On a much smaller scale, there is the humble car-wash: these services installed machines at an enormous rate in the late 20th century, but have lately begun removing them; it is easy enough to find people willing to wash cars in return for not very much money, possibly less that the upkeep and operation of the machines; so why not go back to the old ways?
From the point of view of the capitalist looking to get rid of uppity workers, even the case of the shipping container is a bittersweet one. The drastic reduction of worker numbers on the dock went with the specialisation of the workers who remained, who needed each to grasp much larger parts of the industrial process and master more skills, and thus became harder to replace - and, given their small number, a dangerous bottleneck. Although the unions are battered, the remaining workers have ended up in a bizarrely stronger position in negotiations - somewhat similar to the train drivers who initially did very well out of privatisation in this country.
The good news
The dystopian view of automation has as its intimate opponent a utopianism of the very same. It was expressed most notoriously, albeit with moral reservations, by Keynes in his ‘Economic possibilities for our grandchildren’,5 when he predicted that technical progress in production would radically reduce the amount of work people actually did, and so the next great challenge faced by humanity (after the war even he must have seen coming) would be finding a way to live without the disciplining influence of daily labour. As Marx might say, quelle horreur!
But the same conception was already present in the revisionist socialism of Bernstein in the 1890s, and in the related school of ‘legal Marxism’ in Russia, and lampooned with some venom by Trotsky (and other social democrats) by 1906, and indeed even by Joseph Conrad in his great novel of anti-left hatred, The secret agent. In some sense it lives today, in the hipster-Marxist fad of ‘fully automated luxury communism’, according to whose glorious vision of the future machines will do all the work, so the rest of us can have orgies on yachts. Leaving politics entirely for a moment, we find genuine cranks like the prominent computer scientist, Ray Kurzweil, who expects to live forever, and other prophets of the ‘singularity’ or post-humanism of various kinds.
In the latter case, such optimism seems at least in part a function of biography - computers have, after all, made Ray Kurzweil very rich; the Highlander treatment is surely to follow. As a political projection, in the case of the hipster-Marxists and especially revisionists, the history of automation and technical progress in production seems a little hard to square with such bullish projections. (Exactly where a ‘socialist’ future in which everyone is entitled to seven iPhones leaves the environment is another matter of some minor concern.)
If we are looking for positive cues as to the future, however, we would be better served again to break through the fetish and its ecstatic speculations, and look at the social relations underlying the production of new technical artefacts.
Engineers of all sorts, under capitalist rule, end up being a distinct breed even among ‘the professionals’. For what lawyers and doctors have in common is that there is not a lasting end-product of their efforts: they are each called in as a means of ensuring the material homeostasis of their charges (companies and persons respectively). The engineer resembles the traditional professional, in that they possess a body of skills and knowledge, against which they can charge rent in money and/or in kind (typically in power and autonomy at work). Yet they are like blue-collar productive workers, in that the outcome of their labour is a distinct ‘new thing’ that did not previously exist, and that at least could command a price and be thus integrated into the commodity system - a bridge, a chemical compound, a car, a software system.
We have made, for ourselves, a pretty good life. The story begins in inauspicious circumstances - in the late 40s Japan was being aggressively rebuilt with American money, but people remained poor and short of money for little luxuries like cars. That was a tough problem for budding Japanese automotive companies, which needed to somehow create cars comparable to the American models flooding the market, but better, faster and cheaper. As it happened, they worked it out. The Toyota company employed a number of forward-thinking industrial engineers, above all Taiichi Ohno, who took a very simple approach to the problem - it was all about reducing waste. What was waste? Sure, it was obvious things - materials not needed or used, cars that nobody bought.
But from there the idea expanded. Merely having a lot of materials lying around, if you thought about it, was waste. Economists back to Smith understood that such materials merely rot on the factory floor. It ought to be avoided. How else can you waste materials? Very often, in a factory, some mechanical process goes wrong - let us say, car doors are stamped into the wrong shape, and thus become unusable. At the same time, what is the ultimate metric by which car manufacturers are measured? Why, merely the number of cars rolling off the lines. There is thus a strong incentive for workers - aiming to avoid victimisation - to let the nonsensical car doors continue to roll along, and try to fix things later. Nobody wants to be responsible for stopping the assembly line.
If there is one decisively important feature of the Toyota system above all others, it is ‘stop the line’: if any worker sees defective input, he or she is expected to pause production, and nobody will be fired for doing it erroneously. But the same thing applies in less dramatic circumstances - if workers spot waste, they are trained to self-organise to get rid of it, a process called kaizen or continuous improvement (even most westerners prefer the terser Japanese word).
As soon as workers can stop production and reorganise it at the molecular level, however, all the Taylorist time-and-motion studies in the world are worthless. (For the ideologists of the Toyota production system - or ‘lean manufacturing’, as it is often known - they always were.) The reduction of waste means optimising not for keeping every worker continuously utilised, but for keeping non-wasteful production in progress. It thus points to a smaller, more skilled and self-activating workforce. We should not romanticise things - putting cars together in Japan is just as much a manual, blue-collar process as it is anywhere else. Yet Marxists must learn to spot the seeds of the new society in the old, and the vast expansion under capitalism of cooperative labour in general is one of them. We must therefore also acknowledge and examine the different forms in which, so to speak, these seeds sprout.
Short-sightedness
Back, at last, to computers! In the 1990s, most software development was conducted in a fashion more on the Taylorist model of mass production - the smartest guys in the room would expend a great deal of effort, maybe over months or more, to work out exactly what needed to be done down to the last detail, and when it would be done; this specification would then be handed off to developers, who would sit in their cubicles and crank out what they were told to do, and then it would be tested and released. If Taylor and Henry Ford had at least some success in organising assembly lines in automotive factories, their influence was wholly catastrophic in software, and two decades ago influential software engineers were coalescing around the idea that something else needed to be tried, and that the Toyota/lean methods were the ones to go for. After a lot of toing and froing, this family of best practices came to be known as ‘agile software development’, and it is safe to say that, by the early part of this decade, agile had won the argument against so-called waterfall or traditional methods, although the latter die hard.
Here is what a healthy, agile process looks like. Take a short span of time - usually two weeks, but sometimes one, and less commonly a month. At the beginning of that period, you decide what is going to be worked on, which means working out what one thing or few things are most useful to the organisation. You cannot take on more work than you think you could finish in that time period. Then everyone goes to work - and ‘everyone’ will include engineers, but also visual designers, and maybe even marketers and so on. By the end of your time period (commonly called an iteration or a sprint), the things you worked on should be ‘live’, being used by whoever it is you ultimately do the work for; or it should be at least releasable if you want. You may choose to delay things for some reason or another. At the end, everyone gathers together to decide what worked and what did not and should be changed. Rinse, repeat ...7
I go into this level of detail, for the question at the heart of our discussion is ‘What is technology doing to work?’ And I have suggested that the question is wrongly phrased, for it assigns agency to technical progress that the latter does not really possess. Yet if I think about what the technology industry could, at its best, bequeath to broader society, it is not so much the artefacts themselves, as they figure into the fantasies of ‘fully automated luxury communism’, but rather what we actually do when we make software.
To put it crudely: my job rules! Most people’s jobs suck, and would be better if they were more like my job. They would be better if the workers themselves exercised collective control over their work - so far as is feasible, given the wider economic context; if they defined and freely modified their own methods of organisation and cooperation; if they were trusted to decide what was important and what was not, and to evaluate how successful those judgements were. Such a process, probably for only a brief moment - as brief as the cooperative factories mentioned in Capital volume 1 - actually exists now, but only for a very particular elite layer of the labour aristocracy (the archdukes, as it were, rather than the baronets) and only within very well defined limits at that. There is no reason a factory could not be organised analogously - this stuff came from car plants, for god’s sake! - or a hospital, or a school, or even a military unit (in their book, Lean software development, Mary and Tom Poppendieck cite - of all things - the US Marine Corps as an salutary example8).
The obstacles to this are not so much that we have not, as a species, noticed that this sort of thing ‘works’, but that a method of organising work can only succeed if it is not in contradiction with the prevailing economic conditions. For us plumbers of software, times are good - there is far more work to go around than there are engineers to do it and, though there is neither a union nor a professional association on the model of the Bar or BMA through which we can collectively defend our interests, in circumstances where it is highly costly for employers to annoy us, we do not very much need one.
In wider society, of course, this is hardly a typical case. The ‘economic’ benefits of time-and-motion, and the bullying, hectoring form of management, and arbitrary power on the shop floor in all its forms, are inseparable from the ‘political’ benefits to the capitalist class of having a supplicant working class without the confidence to assert itself at all levels of society. Even the defensive, ‘negative’ workers’ control that came with greater union power in the post-war era was intolerable - never mind forms of work that make obvious the utter superfluity of management and the seat-warmers of the corporate hierarchy.
As long as we engineers are the sole layer, or at least one of the only layers, of the productive classes to enjoy such autonomy in our work, however, even we are victim to a peculiar dialectical reversal, and find ourselves on the wrong side of the automation fetish. We, who are ostensibly the masters, of the machines, have our power only inasmuch as we are at the service of the high priests of the technology-religion. Just as the labour aristocracy in general both benefits from the crumbs sent its way and is in some way imprisoned by them, so we are made into tools by the inanimate machines we shepherd into life.9
Escape from the tyranny of the computer-god implicates, in the end, the central feature of capitalist labour organisation, which is specialisation - the breakdown of production into ever smaller, tediously measurable sub-processes; and, most egregiously of all, the separation of mental and manual labour, and the transformation of ‘knowledge workers’ into a byzantine apparatus of professionals with status far exceeding the grunts on the production line. We have seen that the domination of the machine is actually the domination of one class over another, that it is not robots that throw people out of work, but other people; breaking the machine-myth proceeds inevitably from breaking the social organisation, of which it is merely the distorted reflection.
This, in the end, is the short-sightedness of Ford and his Silicon Valley liberal confrères, as they busy themselves with UBI. Their outlook is founded upon the axiom that hyperspecialisation is the natural order. Ford’s book focuses on the way in which this leaves workers especially vulnerable to replacement by robots and computer programs, but cannot call it into question itself. He seems to argue that neither the elite nor the masses have the power to preserve the link between work and reward; so it is up to the elite to make sure everyone gets on anyhow. Strikingly absent from his account are the masses themselves, who are purely the object of the machines. (Since his book was published, the election of Donald Trump gave his caste a reminder that the superfluous serfs of techno-feudalism cannot be expected to go away quietly.)
Despite his sincere concern for the consequences of a new industrial revolution, he is in the end reduced to apologetics - UBI, since it does not call into question the ownership of the machines, at best (if Keynes is right) merely keeps them rolling to the great benefit of a shrinking haute bourgeoisie. Leftwing support for UBI is understandable, but misplaced - for the point is to lay hold of and transform the whole productive process. Then we will see how much work there is to go around, and what use revolutionary technologies can actually be to humanity.
paul.demarty@weeklyworker.co.uk
Notes
1. This article is based on a talk at Communist University 2017.
2. ‘The roads to Wannacry’ Weekly Worker May 18 2017.
3. M Ford Rise of the robots London 2015, p215.
4. K Marx Capital Vol 1, London 1990. See especially pp375-416 and pp526-64.
5. www.marxists.org/reference/subject/economics/keynes/1930/our-grandchildren.htm.
6. See chapter 7 of Results and prospects, specifically the polemic with Rozhkov: www.marxists.org/archive/trotsky/1931/tpr/rp07.htm.
7. The particular variant of agile development described here is Scrum, the most popular and also the most directly inspired by the Toyota system. See Ken Schwaber’s and Jeff Sutherland’s Scrum guide: www.scrumguides.org/scrum-guide.html.
8. M and T Poppendieck Lean software development Boston 2003, pp62-63.
9. I add, here, the additional problem raised by comrade Mike Macnair: that all productive enterprise in a capitalist society, very much including the relatively egalitarian cooperative labour of an agile software team, is subordinated ultimately to finance capital, which plays - in its corrupt manner - the main role of coordination within capital as a whole. We have since then had a most wonderful demonstration of this phenomenon with the acquisition of ThoughtWorks, a development consultancy that spearheaded the adoption of agile methods and the war against waterfall, by a hungry-looking private equity outfit. The previous owner, to add an extra layer of irony, is a self-proclaimed socialist and admirer of the Chávistas, who has now got an undisclosed, but probably vast, war-chest for his political causes. Socialism looms in the United States, no doubt.