Predicting the Future (and Other Abilities We Don't Have): Part 1

Humans are good at lots of things.  This series of two posts is about a number of abilities that are not among those things.  Part 1 discusses the fashionable tendency to make guesses about the future course of society and then heavily imply (without usually stating outright) that these guesses constitute accurate predictions.  Part 2 discusses the trouble we tend to have in viewing ourselves in a historical context.  This makes it easy for us to believe that we happen to live in revolutionary times, or that none of the old rules apply.  These types of charlatanism can have far reaching consequences, as it turns out.

Part 1:  Extrapolation is an Art, not a Science

One of the ways to get people to pay attention to your predictions is by preaching the good news: eternal life.  Ray Kurzweil has made a number of predictions about the bright future that technological growth will bring us, with this being by far the most notorious.  Although his version of immortality, uploading ourselves onto computers, differs somewhat from the standard Christian view, one can't help but notice the religious flavor of this prediction.

Kurzweil's other predictions for this century include, yes, flying cars, but also reverse-engineering the human brain, nearly perfect simulations of reality (for our digital selves to live in), and, crucially, an AI that is more intelligent in every way than all humans combined.  He has freed himself from any responsibility to explain how these things will be accomplished.  Nobody has the slightest idea how to do the interesting ones.

I will defer the actual technical explanation of why these are truly goofy predictions to authors who have basically handled it: Steven Pinker, Douglas Hofstadter, PZ Myers, and many others have noted how technology and scientific discovery don't progress in the way Kurzweil has claimed.  Instead, I want to draw attention to the fact that these attempts to predict the future are actually a very human tendency.

In the early 1960's, progress in programming machines to do certain tasks (like proving theorems), gave researchers supreme confidence that essentially human-like A.I. would be a solved problem within 20 years.  They should have said that at this rate, it would be done.  What actually happened was that computers became more and more sophisticated but left AI behind: the problems they were doing were just much harder than they had anticipated.  Even relatively simple tasks like constructing grammatical sentences proved to be far out of their grasp.  Now, the most successful language tools largely involve throwing our technical understanding of language out the window and using probabilistic models.

Economies are not spared from erroneous predictions about the future.  Kurzweil and others jumped on the tech-boom bandwagon, claiming in 1998 that the boom would last through 2009, bringing untold wealth to all.  Maybe they should have been reading Hyman Minsky instead of Marvin Minsky.

Enough about the good news.

The other way to get people to pay attention to your predictions is by telling them the bad news: social breakdown the end of the world.  Overpopulation is an issue along these lines that receives attention that is disproportionate to the seriousness of the claims made by its scare tacticians.  Among these claims is the belief that we are imminently reaching the carrying capacity of the Earth, at which point starvation, crowding, and wars over scarce resources will tear human civilization to pieces.

This hypothesis relies on progress not happening, the opposite of the singularitarians' reliance.  But the very same question can be asked of both: how do you know?  This is where it becomes apparent that extrapolating patterns is an art for the Kurzweils of the world.  If you extrapolate one variable, you get intelligent machines.  If you extrapolate another, you get the end of the world.  But if you extrapolate yet another, say the total fertility rate (TFR), it doesn't look so scary.  Defined as the average expected number of children born per woman, the world's TFR has been steadily declining in the post-war period, from almost five in 1950 to almost 2.5 today.  As it approaches two, the world population approaches equilibrium.

Phony overpopulation scares are common in the history of anglophone countries, from Thomas Malthus to American anxiety over the "yellow peril" around the turn of the century (see Jack London's The Unparalleled Invasion for a rosy portrait of the future).  Wealthy people and business interests are often the biggest proponents of the theory that population growth is the largest problem facing the world.  Conveniently, it's one of the only major global issues that isn't their responsibility.  In reality, the only reliable way to lower growth rates is to facilitate the economic growth of poor countries to the point where people there have a decent standard of living.

The danger in blowing the perils of overpopulation out of proportion is that it leads people to prioritize population control above reproductive rights and, more generally, morality.  If it really is that serious, then we have carte blanche to do whatever is necessary.  The bottom line is that, despite the very real possibility of overpopulation becoming an issue, there is no reason to think it is serious or imminent enough to change what our goals would be, were it not imminent.  Our immediate task is still figuring out how to get communities to lift themselves out of poverty while handling climate change and other real crises.

We have a predisposition to weigh the likelihood of possible futures, either good or bad, based on how exciting or terrifying they are instead of how probable they are.  Anyone interested in solving problems should be aware of this bias.

Part 2 covers a second, related bias that people have: the impression that the times we currently live in offer wider and more revolutionary possibilities than existed in the past.  This impression, created by the fact that we live now, not in the past, is the source of huge blunders and the general abandonment of reason.

My Solution to the Fermi Paradox

I think that the most likely solution to the Fermi Paradox is that, while life is exceedingly common in the universe, intelligent life is incredibly rare. In the four and a half billion year lifespan of this planet, life has existed on it from almost the very first moments. However, not until the Cambrian explosion about 530 million years ago did complex multicellular life exist. That means that for nearly four billion years, and starting from nine billion years after the birth of the universe, Earth contained nothing but single-celled organisms and colonies of such. And it is not until about 2 million years ago, 0.04% of the history of life on Earth, that the first technological intelligence emerged. And even that was simply monkeys hitting rocks against each other in a clever way! Anything more complicated than a stone arrowhead was invented in the last ten thousand years!

And let's look at how unlikely it is that humans (or any other hominid) ever even achieved technology. We had access to wheat, barley, and rice, easily domesticable plants that produced high yields and had good nutritional content. We had access to large animals with very particular internal dominance hierarchies which were not so recently evolved alongside us to attack us on sight, but not so distantly separate from us to be completely unwary at the sight of hunters with spears. (Here by "we" I am referring to any subset of humans who had such access — Jared Diamond goes into great detail as to who in fact had access to what.) We had access to workable stone, copious woodlands, and various ore deposits. We had an abundance of fresh water on a planet with an atmosphere suitable for lighting fires. Our luck in the development of our culture and eventual civilization was astounding. We were never subject to an extinction-level impact or eruption (although it seems that we came damn close).

I think that such luck is not only astounding, but is in fact astronomical. While life seems to have no trouble at all finding a place on a planet like the Earth (and perhaps on many other types of planets as well), technological civilization seems like and absolutely ludicrously unlikely event. It has happened exactly once in four and a half billion years (for 0.0002% of that time), or in 530 million years in which complex life has existed on Earth (or 0.0019% of that time). So I would imagine that life is in fact very common in the universe, although almost exclusively in the form of single-celled creatures living on rocks and in oceans. Very, very, very rarely one would find a planet with some sort of multicellular life on it — simple plant-like creatures, or molds of some sort. And once in an unimaginably huge while, one might expect to find a planet where technologically intelligent life once existed. Finding a planet where technologically intelligent life exists concurrently with us seems depressingly close to a fantasy.

[Edit: My percentages were two orders of magnitude off! They were simple ratios, not percentages. Thanks to Scott for pointing this out.]