The Singularity Is Not Upon Us

Again linking to Uncertain Principles—I haven’t been there in a while and it’s worth visiting— I have some thoughts on “The Singularity, which Orzel masterfully describes as follows:

One of the Big Ideas in modern SF is the idea of a “Singularity,” a term coined by Vernor Vinge (I’m not sure if he originated the concept, but I’m pretty sure the term is his). The idea is based on the observation that the pace of technological change has increased dramatically in the last several decades, and if anything appears to be accelerating. Extrapolating out from this, Vinge (and others working with the idea) predict that there will come a point in the near future when changes occur so rapidly that humans (or our post-human descendents) will become essentially unrecognizable. The complexity of our technology is growing exponentially, so humans of the relatively near future will be so far beyond the humans of the past, or today, as to be utterly incomprehensible.

As a science fiction reader, the idea makes for some cool stories. As a scientist, I find it a little annoying. In science, you see exponential growth fairly frequently– the number of products of a given reaction, the number of photons in a laser mode, the number of atoms in a BEC: these are all things that grow exponentially. And the thing about exponential growth is this: it never lasts very long. Eventually, all the resources needed to drive the growth– reactants, excited atoms, non-condensed atoms– are consumed, and the growth levels off. A graph of the product versus time usually looks sort of like an “S” in the Star Wars font: basically flat, turning up dramatically (though not backwards), then flattening out at a higher level.

The classic example (dating back to antiquity) of the unsustainability of exponential growth is the chessboard problem: take a chessboard, and put one penny on the left-hand square closest to you, then two on the square next to it, and four on the square next to that, and so on, doubling the money at each square as you work your way across the board (you’ll have $2.56 at the end of the first row, then double back). See how far you can get– it won’t be very far. If you were to manage to cover each square of the board, you’d need 18,446,744,073,709,551,616 pennies for the final square. And if you’ve got that many pennies, I’d like to talk to you about some research problems I have in mind…

A technological example of exponential growth is the famous Moore’s Law: the speed of computer chips doubles every eighteen months. This is why the 500 MHz computer I bought in 2000 (which wasn’t top-of-the-line then) was replaced by a 2,500 MHz computer (still not top-of-the-line) this year: four years, a bit more than two doublings. Futurists are fond of extrapolating this trend (which is really somewhere between an empirical observation and an industry target) out for several years, and promising no end of marvels once we all have terahertz computers the size of a pack of cigarettes (neglecting the fact that Windows will by that time require six terabytes of RAM for best operation…).

Of course, there are all sorts of problems with this extrapolation. Moore’s original observation was that the number of transistors on a chip (which is related to the speed) was doubling every eighteen months. If you extend that for a few more doublings, you start to get down to transistors that are smaller than the size of a single atom. At which point, you start having some serious sustainability problems.

Even if you assume that some technique will be found to allow the doubling to continue, there’s also Moore’s other (often-overlooked) law: the cost of a chip fabrication plant doubles every 2-3 years. At some point, it’s not unreasonable to believe that the benefits of increased computer power will no longer outweigh the costs of maintaining the current rate of growth, and things will taper off. If you look at some of the “Moore’s Law” graphs out there, you can convince yourself that this might be starting to show up.

Of course, identifying where the turnover will take place is a tricky business. Much trickier than identifying the current trend, let alone extrapolating the trend. If you’re in the exponential portion of the growth curve, it’s hard to spot the end. But it’s a safe bet that things will level off sooner or later (though you’ll probably make more money betting on “later”).

One of the things people usually don’t seem to bring up are our previous experiences with Singularities. Perhaps it’s because they happened to us before we were “us”, so we don’t really remember them. But they exist, as precedents, and they suggest something pretty interesting.

What were our previous Singularity experiences? I would say there were three: the development of language, the development of agriculture, and the development of writing. Each of these fulfill one of the most important traits described by Vinge, which is that it renders those who came before the Singularity basically unable to grasp the people who come after it. It changes the very nature of personhood.

What’s interesting is that two of these Singularities actually involve language, and only one involves “tools” of any kind. In fact, if we were to look at ourselves the way that we look at other species and the rest of nature, that is, looking in terms of geological time, our leap from unlanguaged upright apes to blogging, hi-tech, globalized engines of ideas-expressed-through-language took only an instant. The development of writing was just a part of the development of language, seen from geological time.

Which highlights something interesting: the developments in human history that have wrought the most profound changes—the changes to our very nature, those that made us different from our evolutionary ancestors—have come upon us very slowly. Computers have changed our lives, yes, and in a very valuable way, but they have not yet changed our lives in the way that the development of language has. Language helped to reshape our brains, helped us to learn how to organize ourselves, how to make the best of our one natural advantage, our ability to make and use tools.

So looking at computers in terms of geological time, we’re still at the very beginning of whatever they’re going to do to us, or rather, whatever we’re going to do to ourselves with them. Until computers start reshaping our biology and neurology, I think they will not effect a Singularity upon us.

They could, eventually, do this, of course. Greg Egan’s novel Schild’s Ladder is not only an excellent story, but tangentially is an exploration of how we could eventually use computers to completely reshape ourselves, once they’re past the primitive flint-and-stick level of development we’re at now. But that is very likely millennia away, or at the least, centuries.

I could be wrong, it may be possible that, without us being reshaped internally in any way, we could change so that we are incomprehensible to those who came before us. But I think it’s unlikely to the point of being moot. If a normal child born of normal parents (parents genetically the same as us, as the resulting child would be) can adapt to whatever developments occur, then I think the Singularity has not come. It may be difficult to understand such a society, but then it’s difficult understanding how Medieval people thought, or how ancient pagans did; it would constitute a culture difference, not a difference on the scale of Singularity.

Whether or not Moore’s Law holds, I think it’s safe to say that the Singularity is nowhere near being upon us.

Leave a Reply

Your email address will not be published. Required fields are marked *