In the Clutches of Big Data

Big data fuels the algorithmic concentration of wealth. It happened first in music and finance, but is spreading to every other theatre of human activity. The algorithms don’t create sure bets, but they do gradually force the larger society to take on the risks associated with profits that benefit only the few. This in turn induces austerity. Since austerity is coupled with a sharing economy (because certain kinds of sharing provides the data that run the scheme), everyone but the tiny minority on top of the computing clouds experiences a gradual loss of security.

This, in my view, is the primary negative consequence that has occurred thus far through network technology. To observe that is not to dismiss another problem which has gained much more attention, because it is sensational. A side effect of the rise of the algorithmic surveillance economy is the compelled leakage of all that data into the computers of national intelligence services. We know much more about this than we would have because of Edward Snowden’s revelations.

Curbing government surveillance is essential to the future of democracy, but activists need to keep in mind that in the big picture what is going on at the moment is a gradual weakening of governments in favour of the businesses that gather the data in first place, through the mechanisms of wealth disparity and austerity. That is only true for democracies, of course; non-democratic regimes take control of their own clouds, as we see, for instance, in China. I do sometimes wonder if we’ve outsourced our democracies to the tech companies simply in order to not have to face it all. We deflect our own power and responsibility.

The algorithms don’t create sure bets, but they do gradually force the larger society to take on the risks associated with profits that benefit only the few.

Here I feel compelled to foresee a potential misunderstanding. I am not ‘anti-corporate’. I like big corporations, and big tech corporations in particular. My friends and I sold a startup to Google and I currently have a research post in Microsoft’s labs. We must not put each other through purity tests, as if we were cloud algorithms classifying one another for targeted ads. The various institutions that people invent need not annihilate each other, but can balance each other. We can learn to be ‘loyal opposition’ within all the institutions we might support or at least tolerate, whether government, business, religion, or anything else. We don’t always need to destroy in order to create. We can and ought to live with a tangle of allegiances. That is how to avoid the clan/hive switch.

The conflict between the Left and Right has been so acute for so long that we don’t even have an honest vocabulary to describe the honest mathematics of the bell curve. We can’t speak of a ‘middle class’ because the term has become so fraught.

And yet that impossible-to-articulate middle is the heart of moderation where we must seek peace. As boring as it might seem to be at first, moderation is actually both the most fascinating and promising path forward. We are constantly presented with contrasts between old and new, and we are asked to choose. Should we support old-fashioned taxis and their old-fashioned benefits for drivers or new types of services like Uber that offer digital efficiencies? These choices are false choices! The only ethical option is to demand a synthesis of the best of pre-digital and digital designs.

One of the problems is that technologists are often trapped in old supernatural fantasies that prevent us from being honest about our own work. Once upon a time, scientists imagined coming up with the magic formulas to make machines come alive and become self-sufficient. After that, artificial intelligence algorithms would write the books, mine the fuels, manufacture the gadgets, care for the sick and drive the trucks. That would lead to a crisis of unemployment, perhaps, but society would adjust, perhaps with a turn towards socialism or a basic income model.

 After that, artificial intelligence algorithms would write the books, mine the fuels, manufacture the gadgets, care for the sick and drive the trucks.

But the plan never worked out. Instead, what looks like automation is actually driven by big data. The biggest computers in the world gather data from what real people – like authors – do, acting as the most comprehensive spying services in history, and that data is rehashed to run the machines.

It turns out that ‘automation’ still needs huge numbers of people! And yet the fantasy of a machine-centric future requires that those real people be rendered anonymous and forgotten. It is a trend that reduces the meaning of authorship, but as a matter of course will also shrink the economy as a whole, while enriching those who own the biggest spying computers.

In order to create the appearance of automatic language translations, for instance, the works of real translators must be scanned by the millions every single day (because of references to current events and the like.) This is a typical arrangement. It’s usually the case that an appearance of automation is actually hiding the disenfranchisement of the people behind the curtain who do the work, which in turn contributes to austerity, which in turn rules out the possibility of socialism or basic income as a way to compensate for all the theatrically simulated unemployment.

The whole cycle is a cosmic scale example of smart people behaving stupidly. ‘Disrupt’ might be the most common word in digital business and culture. We pretend it’s hard to differentiate ‘creative destruction’ – a most popular trope in modern business literature – from mere destruction. It really isn’t that hard.

Just look to see if people are losing security and benefits even though what they do is still needed. Buggy whips are obsolete, but the kinds of services being made more efficient by digital services lately are usually just being reformatted, not rejected. Whenever someone introduces a cloud service to make some aspect of life easier, like access to music, rides, dates, loans, or anything else, it also now expected that innocent people will suffer, even if that is not strictly, technically necessary. People will be cut off from social protections.

We humans are geniuses at confusing ourselves by using computers.

If artists enjoyed copyright, that will be lost in the new system. If workers were in a union, they will no longer be. If drivers had special licenses and contracts, they no longer will. If citizens enjoyed privacy, then they must adjust to the new order. The familiar expectation that one must incinerate old rights, like privacy, or security through the labour movement, in order to introduce new technological efficiencies, is bizarre. Techie idealists often focus on how the old protections were imperfect, unfair, and corrupt – all of which was often so – but we rarely admit to ourselves how the new situation offers spectacularly inferior protections and astoundingly greater levels of unfairness.

If you are a technology creator, please consider this: If you need to rely on dignity destruction as a crutch in order to demonstrate a new efficiency through digital networking, it only means you’re not good at the technology. You are cheating. Really efficient technological designs should improve both service and dignity for people at the same time. We humans are geniuses at confusing ourselves by using computers. The most important example is the way computation can make statistics seem to be an adequate description of reality. This might sound like an obscure technical problem, but it is actually at the core of our era’s economic and social challenges.

The algorithm might have been set in place by a government intelligence operation, a bank, a criminal gang, a Silicon Valley company, who knows? It is ever-cheaper to do it and everyone who can, does. That algorithm will probably be correct for at least a little while. This is true simply because statistics is a valid branch of mathematics. But beyond that, the particular reality we find ourselves in is friendly to statistics. This is a subtle aspect of our reality.

Our world, at least at the level in which humans function, has an airy, spacious quality. The nature of our environment is that most things have enough room to continue on in what they were just doing. For contrast, Newton’s laws (i.e. a thing in motion will continue) do not apply in a common tile puzzle, because every move is so constrained and tricky in such a puzzle. But despite the apparent airiness of everyday events, our world is still fundamentally like a tile puzzle.

It is a world of structure, governed by conservation and exclusion principles. What that means is simple: my finger will probably keep on moving as it was, but not forever, because it will reach the limit of how far my arm can extend, or it will run into a wall or some other obstacle. This is the peculiar, flavourful nature of our world: commonplace statistical predictability, but only for limited stretches of time, and we can’t predict those limits universally. So cloud-based statistics often work at first, but then fail. We think we can use computers to see into the future, but then suddenly our schemes fail. (Good scientists who work with theory, beyond statistics, understand this problem and also model the wall that interrupts the progress of your finger. That level of effort is rarely expended in cloud business, however, since billions are still made without it.) This is the universal and seductive pattern of intellectual failure in our times.

The algorithm might have been set in place by a government intelligence operation, a bank, a criminal gang, a Silicon Valley company, who knows? It is ever-cheaper to do it and everyone who can, does.

Why are we so easily seduced? It is hard to describe how intense the seductive quality is to someone who hasn’t experienced it. If you’re a financier running cloud statistics algorithms, it feels at first like you have the magic touch of King Midas. You just sit back and your fortune accumulates. But then something happens. You might run out of people to offer stupid loans to, or your competitors start using similar algorithms, or something.

Some structural limit interrupts your amazing run of perfect luck, and you are always shocked, shocked, shocked, even if it has happened before, because the seductive power of those early phases is irresistible. (A baseball team where I live in California was celebrated in the book and movie Moneyball for using statistics to become winners, and yet now they are losing. This is utterly typical.)

The users will typically feel like they are getting tremendous bargains. Free music! They seem to be unable to draw a connection to their own lessening prospects. Instead they are grateful. If you tell them, through the design of algorithms, how to date, or how to present themselves to their families, they will comply. Whoever runs one of these operations, which I call Siren Servers, can set the norms for society, such as privacy. It is like being king.

That is the raw economic snapshot that characterises so many aspects of our society in recent times. It was the story of music early on. Soon it will be the story of manufacturing (because of 3D printers and factory automation), health care (because of robotic nurses), and every other segment of the economy. And of course it has overtaken the very idea of elections in the United States, where computational gerrymandering and targeted advertising have made elections into contests between big computers instead of contests between candidates. (Please don’t let that happen in Europe.) It works over and over and yet it also fails over and over in another sense.

Automated trading crashes spectacularly, and then starts up again. Recorded music crashes, but then the same rulebook is applied to books. Billons are accumulated around the biggest computers with each cycle. The selfish illusion of infallibility appears over and over again – the serial trickster of our era – and makes our smartest and kindest technical minds become part of the problem instead of part of the solution. We make billions just before we slam into the wall. If this pattern is inevitable, then politics don’t matter much.

Politics, in that case, could at most delay a predetermined unravelling. But what if politics can actually matter? In that case, it is sad that current digital politics is so often self-defeating. The mainstream of digital politics, which is still perceived as young and ‘radical’, continues to plough forward with a set of ideas about openness from over three decades ago, even though the particular formulation has clearly backfired.

This is a tricky topic and deserves a careful explanation. The ‘sharing economy’ offers only the real time benefits of informal economies that were previously only found in the developing world, particularly in slums. Now we’ve imported them into the developed world, and young people love them, because the emotion of sharing is so lovely. But people can’t stay young forever. Sometimes people get sick, or need to care for children, partners, or parents. We can’t ‘sing for our supper’ for every meal.

Because of this reality, the sharing economy has to be understood ultimately as a deceptive ritual of death denial. Biological realism is the core reason formal economies came into being in the first place. If we undermine both union protections, through the sharing economy, and trap governments in long term patterns of austerity and debt crisis, through that same economy, who will take care of the needy?

Sometimes I wonder if younger people in the developed world, facing the inevitable onslaught of aging demographics, are subconsciously using the shift to digital technology as way to avoid being crushed by obligations to an excess of elders. Most parts of the developed world are facing this type of inverted demographic cataclysm in the coming decades. Maybe it’s proper for young people to seek shelter, but if so, the problem is that they too will become old and needy someday, for that is the human condition.

Free language translation services actually depend on scanning the work of millions of real human translators every day. Why not pay those real people?

Within the tiny elite of billionaires who run the cloud computers, there is a loud, confident belief that technology will make them immortal. Google has funded a large organisation to ‘solve death’, for instance. There are many other examples. I know many of the principal figures in the antideath, or post-human movement, which sits at the core of Silicon Valley culture, and I view most of them as living in a dream world divorced from rational science. (There are also some fine scientists who simply accept the funding; funding for science these days often comes from oddly-motivated sources, so I cannot fault them.)

The arithmetic is clear. If immortality technology, or at least dramatic life extension technology, starts to work, it would either have to be restricted to the tiniest elite, or else we would have to stop adding children to the world and enter into an infinitely stale gerontocracy. I point this out only to reinforce that when it comes to digital technology, what seems radical – what at first seems to be creative destruction – is often actually hyper-conservative and infinitely stale and boring once it has a chance to play out.

Another popular formulation would have our brains ‘uploaded’ into virtual reality so that we could live forever in software form. This despite the fact that we don’t know how brains work. We don’t yet know how ideas are represented in neurons. We allocate billions of dollars on simulating brains even though we don’t really know the basic principles as yet. We are treating hopes and beliefs as if they were established science. We are treating computers as religious objects.

The pattern we see today is not the only possible pattern and is not inevitable.

We need to consider whether fantasies of machine grace are worth maintaining. In resisting the fantasies of artificial intelligence, we can see a new formulation of an old idea that has taken many forms in the past: ‘Humanism’. The new humanism is a belief in people, as before, but specifically in the form of a rejection of artificial intelligence. This doesn’t mean rejecting any particular algorithm or robotic mechanism.

Every single purported artificially intelligent algorithm can be equally well understood as a non-autonomous function that people can use as a tool. The rejection is not based on the irrelevant argument usually put forward about what computers can do or not do, but instead on how people are always needed to perceive the computer in order for it to be real. Yes, an algorithm with cloud big data gathered from millions, millions of people, can perform a task. You can see the shallowness of computers on a practical level, because of the dependency on a hidden crowd of anonymous people, or a deeper epistemological one: without people, computers are just space heaters making patterns. One need not specify whether a divine element is present in a person or not, nor precisely whether certain ‘edge cases’ like bonobos should be considered human beings. Nor must one make absolute judgments about the ultimate nature of people or computers.

One must, however, treat computers as less-than-human. To talk about specific ways out of our stupid digital economics pattern is to enter into a difficult argument. I have mostly explored and advocated one approach, which is to revive the original concept for digital media architecture, dating back to Ted Nelson’s work in the 1960s.

Ted suggested a universal micropayment scheme for digital contributions from people. Once again, this was not a radinor a corporation should be a person! The new humanism asserts that it is ok to believe that people are special, in the sense that people are something more than machines or algorithms. This proposition can lead to crude mocking arguments in tech circles, and really there’s no absolute way to prove it’s correct. We believe in ourselves and each other only on faith. It is a more pragmatic faith than the traditional belief in God. It leads to a fairer and more sustainable economy, and better, more accountable technology designs, for instance. (Believing in people is compatible with any belief or lack of belief in God.)

To some techies, a belief in the specialness of people can sound sentimental or religious, and they hate that. But without believing in human specialness, how can a humanistic society be sought? May I suggest that technologists at least try to pretend to believe in human specialness to see how it feels?

Death and loss are inevitable, whatever my digital supremacist friends with their immortality laboratories think, even as they proclaim their love for creative destruction. However much we are pierced with suffering over it, in the end death and loss are boring because they are inevitable. It is the miracles we build, the friendships, the families, the meaning, that are astonishing, interesting, blazingly amazing.