Friday, November 21, 2014

The second machine age (Brinjolfsson & McAffee) - II

In my latest post on the subject (which you can find here: The second machine age I if you are too lazy to scroll down about 20 cm...) I denounced what I felt was a very incomplete (albeit common) understanding of what AI was about, and its potential impact on how we manufacture and distribute stuff. I told then that I would deal with the economic aspects of the book in a separate post, and this is going to be it.

To give some credit where credit is due, the book managed to exceed my (admittedly very modest) expectations regarding economic analysis, as in the central chapter it recognizes the three main trends that are shaping the fate of the workforce in all advanced countries: diminishing growth (in year over year increase of GNP), stagnant (for the 90%) or even diminishing (for the bottom 20%) income for salaried workers and growing inequality (as both the income and the wealth of the top 10% -mainly concentrated in the top 1%, and even more so in the 0,1%, has indeed growed at an ever increasing pace). Many techno-utopians and naive optimists simply ignore those trends, either by minimizing their importante, outright denying their existence, or balancing them with the increase in output and standard of living in the emerging economies.

Too bad the book then forgets almost entirely about them, and in the final chapters go back to its authors unfounded cheerfulness, stating without proof that, because technology has given us driverless cars (that are not yet fully operational, and may never be: Google self-driving car may never actually happen), voice recognition software that doesn't completely suck and maps that suggest the fastest path under conditions of heavy traffic, it will magically solve all of society's current problems. Well, it very well may not, because a) as I mentioned in the first post, technology may not be advancing that fast at all (and may even be regressing, a fascinating possibility that merits a post of its own) and specially b) those problems may not be solvable by technical advances alone, as they do not have to do with producing more stuff, but with how the stuff that is indeed produced gets distributed. Many people (like these bozos The zeitgeist movement and these The Venus Project) maintain that with the current level of technological advances we should be able to get rid of war, hunger, poverty and disease, and transition from a society where scarcity rules (hence the famous definition of Economy by Lionel Robbins as the sciente of allocation of scarce resources) to a "resources based economy" where everybody has plenty of everything, and doesn't need to unneccessarily toil to survive (the Utopia of a society where work is optional, people can devote their time to creative, artistic and intellectual endeavors and nobody has to obey the orders of anybody else is a constant, from More to Marx). According to such naive view (that, interestingly enough, coming from evident nerds and technophiles, show a blatant misunderstanding of how technology really works, at least "dirty", "stuffy" technology, not Sw development and IT systems, which is probably where most of this ingenues come from) the only thing preventing us from achieving that golden age is the greed of a tiny clique of plutocrats, the heads of multinational corporations that have all of the world's politicians in their pockets, who selfishly deny all of humanity the unlimited boons of our age...

But I'm not going to loose any time pointing to the obvious stupidity of those visions, which are not shared (at first look) by the authors of the book under review. I only wanted to note that the naivete is the same in both cases: Sw allows us to do wonderful things (normally self-contained, as Sw does not generate energy, does not grow crops, does not give us shelter, does not take us phisycally anywhere, and only tangentially, and with the use of vast amounts of capital and raw materials, builds things), hence developments in Sw enabled by that most wondrous of laws (Moore's) will neccessarily put an end to all of humanities woes. Indeed, the authors note one glaring hole in their theory: if technological advance is the mechanism that explains economic growth, how is it possible that we have (thanks almost inevitably to Moore's law) ever fastest technological advance AND for the last two decades, ever slowing economic growth (measured in terms of GDP)???

Brinjolfsson and McAffee propose two explanations, but both of them fail: first, they state that it takes some time for a GPT (General Purpose Technology) to elicit the associated changes in processes and behaviors required for the society to be able to reap its rewards, citing the example of the gap between the invention and ubiquitous availability of electricity and its translation to improved output in the industries that adopted it. Second, they contend that GDP is not a comprehensive enough measure of our well being (duh!), as it does not capture the additional richness that new (mostly electronic) gadgets bring to our life (all the music and movies and photos you can store and stream and enjoy almost for free now). The first argument fails because comparing IT with electricity and thinking it is almost a magical wand that creates boundless increases in the productivity of every industry it touches is, quoting Sheldon Cooper, "baloney with a side dish of malarkey". If you read enough academic papers (written by professors paid directly or indirectly, in terms of access and information, by the same companies that profit from the installation of those IT systems) you may end up believing such a thing, but I'm gonna go over one foot here and call it for the load of BS it is. I spent 15 years of my professional life designing and implementing those very same IT systems everybody raves so much about, and not in a single instance did I see any productivity gain at all, and not a single business benefit other than the maintenance of the grossly overpaid consultants that did the implementation (and that in the initial years of the industry, departed afterwards , leaving the hapless users to deal with the monstrosities we were paid to put in place... of late it is the same consultants, or their underlings, who do the subsequent quite unpleasant dealing, which at least has some poetic justice). I've done most of them: ERP's, CRM's, DW's, GL, AP, AR, B2C portals, B2B portals, B2E portals, you name it, and I would gladly debate with anybody about the real productivity gain derived from the tiny sliver of those that were actually implemented, and not discarded after years of fruitless struggle.

The second argument also fails, for reasons that would take us a bit far from economics into social commentary and, finally, ethics (what are the ingredients of a life well lived, and what part do material goods play on it). Let us say that I find any contribution to our well being brought about by the Internet (because it all boils down to that) dubious at best. Take music: the authors argue that we have better and more varied music available at our convenience, without having to go to a shop to purchase it, and it being (almost) always on (or at least one click away). Sounds great. Now go to any teenager's den and have a look at what he is hearing (in any corner of the world): the same rubbish, from the same limited menu of choices. Uncultivated teenagers, you may say, are not representative of the wider society. OK, go to any salaried worker house, any average Joe, and see how much music they are hearing, and how varied it is. Or how many books they are reading (different from 50 Shades of Grey and any crap from Dan Brown). Or how many films they are watching. Ooops, reality TV and having a couple jobs to make ends meet don't let them much time to hear sophisticated music, read profound books or see intellectually challenging movies... or, outside of Harvard professors may be not many people is enjoying those non-GDP registered advantages of the IT revolution.

Finally, Brinjolfsson & McAffee contend that the decrease in the percentage of people participating in the workplace we see the world over (more marked, again, in the most advanced economies) is a temporary thing, and that the productivity gains enabled by the wider and wider use of more and more intelligent machines will in the end make all of us richer, specially those who learn to collaborate with those machines, and use them to their advantage (in their words, those who "race with the machines, instead of against them"). Again, claptrap. I was also of the opinion that considering the total amount of work a fixed quantity to be distributed between an ever increasing population was a fallacy, and that you could always identify new needs to be satisfied, and hence more work to be done, and that ingenuity and inventiveness are the real limiting factors to economic growth nowadays (that's exactly the authors' position), but I've evolved my position. The fact is, most ideas require increasing amounts of capital to be monetized, and most people do not have the cognitive capabilities to harness that capital. There is indeed a limited amount of well-paid work to be done, and more and more it requires pretty exclusive (uncommon, and difficult to expand) skills. Most of the work that well-heeled people is willing to pay for to those with a more limited skill set is menial, and what they (we) are willing to pay for it is a pittance. So, contrary to what B&MA say, Tomorrow's (or Today's) world is not going to be chock full of opportunities for web designers, fashion consultants, marketing gurus, music composers and successful screenwriters, but moderately  abundant in unfilled positions for boot polishers, maidservants, nurses, burger flippers and the like, paid barely above the subsistence salary to do repetitive tasks, with almost zero chances of professional advancement.

And the solution does not lie in more technology, or hoping that Moore's law will deliver us from dystopia ensuring effortlessly a renewed economic growth that lift all boats. The solution is, as always has been, political. The decisions to be made have to do with how we distribute the goods we produce, with who gets what, with who DESERVES what. And there is not (and there will probably never be; as we do not have a hint of a clue of how to algorithmize it) a procedure to decide the optimal distribution that we can unload to a machine, as "artificially intelligent" as we deem it to be. We will need to discuss, to empathize, to bargain, to painfully renounce to some things we cherish to gain others we consider more worthy... the same old same old we have been doing for millennia. Just thinking we have come to a point in our development as a species where we can forgo that tiresome discussing is just delusional, as delusional as the unfounded optimism that permeates the book we have just discussed.

No comments:

Post a Comment