Jobs, machines, robots and civilizational collapse

My good friend Pedro Linares has been pointing lately a surprising number of good articles in the web to me, from this little piece in the New Yorker: Better kind of Happiness (it alleviates my guilt from reading that quintessentially elitist outfit that my friends also indulge in such habit), which prefigured many of the ideas I later reflected in my latest post on how to live (Live Kantianly!, or engage in long-term projects that are hard and keep you engaged, ‘cuz Kant already told you so, you moron!) to this other piece from Acemoglu and Restrepo Job race: machines vs. humans which he posted in his always commendable blog. The latter, even though full of logical flaws in reasoning that reveal the sad state of what passes today for advanced social science, got me thinking about the stated race between automating technologies (job killing) and the supposed creation of cognitively more complex tasks that it enables (job creating), and allowed me to reach a number of conclusions that I’ll share in this post, as they clarify the likely form that our civilization’s final demise may take in the next decade.

First, as I mentioned in my recent review of Robert Gordon’s magisterial The Rise and Fall of American Growth (3rd Industrial Revolution? fuggedaboudid!), in this area surprisingly the techno-optimists are driven to the gloomier predictions, as they (wrongly) believe the (inexistent) acceleration of innovations will speed up the pace of automation and drive countless (and ever growing) hordes of workers (starting with the less skilled ones, but soon affecting the ones engaged in more cognitively demanding jobs) out of work. On the contrary, according to Gordon there is not much innovation going on (as measured by the ability of new technology to produce increases in Total Factor Productivity, or TFP, which has been notoriously flat since a comparatively anemic spurt in the 90’s of last century), which means that the much-touted and much-ballyhooed automation that would condemn most of us and our descendants to technological unemployment is simply not happening (here you can find none other than Robert Gordon himself explaining it in Vox: The Master Speaks!).

Everything else being equal, the lack of technological progress (in Acemoglu and Restrepo’s model, the diminishing pressure of automation, even in an environment of “cheap capital” which would make substituting for human labor attractive) should ease the pressure on labor and thus contribute to a decrease in the rate of unemployment, an upward trend in salaries and a growing percentage of labor’s reward as a percentage of GDP (as opposed to capital, which would see a diminishing reward). As that doesn’t reflect at all what we see, we have to conclude that either the model is crap (my first option) or that everything else is not equal (also true). I should note that the model mentions the existence of automatic stabilizers that should prevent it from letting one of the competing tendencies get too much ahead the other one. According to the authors, in a situation of crazy technical progress, with automation getting seriously out of whack the downward pressure on salaries would make it more attractive for employers to hire humans to do cognitively more demanding tasks, thus creating new openings for positions that did not exist previously that in turn would increase the number of jobs (countervailing the downward pressure on salaries, as the new jobs would be better remunerated than the lost ones).

There is, of course, a more parsimonious explanation of why the progress of technology since the Industrial Revolution has not left all of us idle (Leontieff’s paradox, more about it later): as more people have lost their jobs salaries have indeed go down and capital’s bargaining position has improved, thus moving the supply curve for labor to the left (people “at the margin” were willing to work for less), thus intersecting the demand curve at a higher level of employment, even if such employment was worse paid for a total higher number of hours (on aggregate, although it may go either way for the average number of hours worked by each individual worker). As I showed in my post about fiscal policy, even one that dwarfed the content of a Keynesian wettest dreams would not do much to change that and get more people back to work (lots of supply and demand curves) but remember how I laid out in that post the way things have played out historically: the reason technology has not consistently depressed the total employment figure is because population has consistently added additional consumers to the market, thus increasing the aggregate demand more than what the aggregate supply could meet just by the technological improvements. Thus industry (and services) had to resort to adding more of just about any factor (more machinery, but also more workers) to meet it. But population has more or less stopped growing, except for a few remnants of African villages that have no resources whatsoever to become a source of additional consumers, and are busily trying to migrate to more promising (economic) climates, while the potential receiving locations are similarly busy trying to avoid them doing so.

So I think we are on the last ropes of Leontieff’s paradox. We have seen people finding work in the face of labor-reducing inventions because such inventions were not enough by themselves to supply the appetites (continuously whetted by a very successful dominant reason, y’all know that) of an ever growing demand. Now you may say that even if the number of people demanding additional thingies has stabilized, nothing prevents each one of them from wanting even more, so the aggregate demand could in theory keep on growing forever, even in the absence of population growth. That would be indeed the final goal of a sane economic policy: see how we can stimulate a perpetual improvement of people’s standard of living without a perpetual increase in their numbers, so they produce ever more for each individual without an ever increasing environmental footprint (for example, by providing each other ever more fulfilling services, which have typically a lower consumption of non-renewable resources  than thingies, as things have a tendency of requiring a whole dirty industry behind them to be manufactured and distributed). The “not increase the environmental footprint” is not an additional and extravagant condition that I’ve conjured out of thin air, but a “contour condition” that has to be taken into account if we don’t want to run very rapidly into nature’s limits and trigger all sort of perverse negative feedback effects that send the whole system crashing down (something we may be already doing, btw). Unfortunately, such goal is intrinsically unattainable.

To find out why we need to delve deeper in an aspect I touched only in passing in my post about the limits of fiscal policy: the subtle but consequential difference between the industrial (and agricultural and extractive) sectors and the service sector, a difference that will show why Acemoglu and Restrepo’s model (along with most techno-optimists set of assumptions) is an elegant amount of sheer bunk. Such difference is much obscured by the fact that both products (for which it may be more clear to use the old Marxist term “commodities”) and services are exchanged in the market, so both categories are offered by the best price they can fetch, and end up being paid for in the same currency (money), and thus it is easy to think that both  have an equivalent “value” that is in the end determined by the amount of labor that goes in its production, may be somewhat adjusted by the time that the person doing the labor had to devote to acquire the required skills to perform it. For this to be true Jamie Dimon (estimated salary 27 million $ in 2016, or 10,800 $/hour assuming he worked a quite taxing 2,500 hours that year) should have devoted about 1,000 hours of learning per hour, as to offset the difference with the 12 to 15 $/hour his lower salaried employees will perceive in Chase (and after a “generous” raise, as in 2015 they were making a paltry 10 $/hour), or, to explain a single year difference, he would need to have devoted 1,000 years more than his janitors of arduous study to prepare himself for the extraordinary skill of leading a major financial institution that, let’s not forget, was essentially bankrupt five years ago and had to be bailed out with taxpayers money (that unfortunately didn’t came with any clause regarding the acceptable compensation levels of its top executives). Well, that’s my circuitous way of saying that the labor theory of value is baloney (something I already did here: Value and Wages), but that doesn’t take us an inch closer to solving the problem at hand.

So back to the difference between offering services in the market and offering material products (commodities), the fact that both are given to whoever can pay the highest price for them obscures, as I said, a fundamental difference: commodities can be stored and loose value (understood as the price they can command, I don’t want to go all metaphysical here) very slowly if at all, while every hour a service provider is not providing the service is an hour lost forever, that he will never be able to recover. That makes the services market inherently more buyer dominated, and puts an additional pressure on the providers to find someone willing to pay for their time, as they can’t afford the luxury to keep the product of their labor sitting idle in a warehouse waiting for a better deal. I am aware that a famed neurosurgeon or a sought-after life coach have throngs of people willing to pay a premium for their services, but let’s be honest here: such figures constitute so tiny a percentage of the labor force that we can entirely ignore them in our analysis (although they are very useful to justify the status quo and convey the message that for everyone willing to put the time and devote the necessary effort there are great rewards to be reaped from the current system in any walk of life). We can confidently state that for the vast (90% or more) of people offering services in the market, such market is tough, they can not claim to have any substantial advantage over the other providers, there is fierce competition and they have little influence over the price they can command.

You may say such is life, the market, although a stern master, is still the best social mechanism for optimally assigning finite resources to the most socially desired ends and for distributing them, and that applies as much to services as to goods. Doesn’t really matter if those resources are small mud huts, Versailles-like 10,000 square feet mansions, little mopeds, scrumptious Ferraris, a 6$ haircut or a 1,000 $ massage in a luxurious spa. The best way to ensure everybody receives as close as possible what she wants, given means distributed by as close as possible to just dessert, is by the magical action of the free market. Market schmarket, I say, and anybody that believes that load of bunk has been either brainwashed or born without a wholly functioning brain. Back to my argument, I would say the services sector is morally tainted, and it not only doesn’t conduct to proper human flourishing, but seriously stunts the moral development of people trapped in it. And I say trapped with full understanding that the investment banker and the interior designer that earn gazillions fool themselves thinking they are making a killing and are most fortunate to have chosen such profitable line of activity to devote their life to. They are equally wrong, and should see themselves as not that different of the poor schmuck flipping burgers in a nondescript joint in the middle of nowhere.

Again, when most economists discuss the service economy, they tend to focus on its glitziest aspect: business consultants, web designers, designers of successful apps that rake in millions… although the reality is somewhat grimmer: poorly paid housemaids, janitors, waiters, repairmen from foreign countries without legal status, nail polishers, retail clerks, call center attendants on a Taylorist schedule, programmers on perpetual crunch time whose salary barely allows them to make ends meet, bank tellers, designers in advertising agencies working 20 hours a day for creative directors they will never replace… what they all have in common is that, regardless of how glamorous the title they are given sounds, their contribution is pretty easy to replace, so they are paid little, they can be fired on a whim and they have little prospect of professional advancement. You may counter that they should build a “differential set of skills” that made them stand out from the potential competition, and I would laugh you off the room. Build when? Using what resources? By definition, the majority of the population (in a services economy as in an agrarian, or an industrial one) can not have above average skills, or capabilities.

And that’s why the powerful, the clever, the wealthy and the well educated  find the whole arrangement so lovely and convenient. When all those features tend to coincide, and when such coincidence becomes, through assortative mating and high heritability mostly transferred from one generation to the next, you have all the ingredients for a class-based society of low intergenerational mobility where the elite becomes more and more detached from the concerns of the average “guy in the street”, where inequality grows and human’s natural selfishness, unbridled, allow public benefits to whither. Oooops, just the society we live in! And a society where only “services” grow and supposedly provide opportunity for all is a society where the slightest modicum of justice has become nearly impossible to achieve, as such services are the more or less fancy names given to the starkest servitude, to the centuries old exploitation of man (or woman) by man.

Think about it. I worked the first fifteen years of my career in a professional services firm, and the thing that never stopped to amaze me was precisely how servile, how debased even the most powerful partners (most already filthy rich, but you know what they say: you are never rich enough) were towards the clients that had in their hands their further professional advancement. In such firms, unsurprisingly, the best career is offered to those that can sell, because selling is the whole point of the enterprise, regardless of what the “mission” and the “vision” state about delivering value to the clients and putting their interest first and building a better world. Technical competence, being an inspiring leader towards your subordinates or being an efficient business operator were all nice qualities to have, but being able to convince a client (which usually required obscene amounts of brown-nosing and sycophancy) to hire our services was absolutely paramount, and dwarfed the importance of any other skill when it came to promotion. And what is the clue to such convincing? In the service world, doesn’t matter how much you want to beat around the bush, it is all about making your potential client believe you will serve him better. Strategic partnership, win-win situation, maximize the value of the relationship and not the transaction and unlock hidden value levers my foot, the one who wins the business in a competitive situation is the one who shows he will be a better server (more docile, more slavish, more pliant, more obeying, more unprepossessing; and less confrontational, less arrogant, less unruly, less independent-minded, less free-spirited). Not what management books teach, huh? I have this shocking piece of news for you, kids: management books lie (blatantly in most cases).

So service is mostly about debasing oneself to satisfy the ego of those hiring you (which many times means kicking the balls of their underlings so they don’t have to do it themselves) while (in the upper echelons of the profession) lying to yourself thinking that you are in control, you do it for your own good, you can choose when to quit, you are the one really taking the decisions and having a measurable impact in the companies for which you work… if it sounds like the rationalizations prostitutes use with themselves to justify why they are servicing the next John it is not a coincidence. But what is of more importance for our current argument is that in such branch of activity it is difficult both to reach real meaning and to grow productivity. Stroking your customers’ ego is awfully difficult to automate, as having a robot tell you how clever you are and how most of your subordinates are morons doesn’t give the same kick as having a Harvard graduate tell you the same thing. Seen from the other side of the table, having our best and brightest young minds spend most of their careers telling old white men in positions of power how they are so exactly right that their companies should do little but what they have already decided is not likely to produce breathtaking technological breakthroughs.

And there you have it in a nutshell: the growth of the service economy makes it trebly unlikely we will ever get out of the current stagnation, because:

a)       What people is paying for is not to have a physical need satisfied, but a “spiritual” need (feeling superior to someone) that requires another human being, and can not be fulfilled by a machine, it doesn’t matter how intelligent it is

b)      The kind of careers such economy promotes are antithetical to innovation and scientific development, as they channel the most ambitious, most disciplined and most gifted overall in activities that can only be described as “self-important brown-nosing”

c)       True scientific development is left mostly to second-rate intelligences (with some notable, highly laudable exceptions), and further hampered by a deluge of unnecessary regulations (and a layer of self-inflicted insanity in the form of the academic publication industry and its peer review creativity-stifling mechanism)

So you can sleep soundly, no robots are coming to take your jobs (as Gordon notes in the interview linked above, no robots are being developed for the tasks that occupy the vast majority of the population that still holds a job, from pushing wheelchairs along hospital corridors to replacing perishable goods in supermarkets’ shelves), and not much effort is being devoted to such development, because many of those jobs exist to allow even the people lowest on the social scale to boss over someone even lower, independently of cost, effort or economic rationality. We will see the number of people working drop more and more (although not necessarily find them in the unemployment roll, as they may just give up even trying to find some occupation, as the part of this interview with Erik Hurst dealing with time allocation of youngsters without a college degree highlights: "What are young men doing"), some robots in traditional manufacturing eating a bit of additional jobs at the margins, may be applications of AI changing the cognitive content of a bunch of others (but with no real substitution effect, for reasons I would explain in another post, other than the fact that the bosses of the substitutable employees want them to feel important, not because their automatable contribution is that important).

What will continue happening is that a higher percentage of the population will work providing “services”, and deriving less and less satisfaction from it, as the stark “needlessness” of their activities becomes more and more apparent. Until, of course, they revolt, bring down civilization as we know it, either because they overthrow the ruling elite that has been benefitting from their servility -sorry, from their services, or because they are crushed by the robot army that such elite has developed in the meantime…