Singularity, Schmingularity

Several science fiction authors have been writing about an idea called the singularity. The theory is that technology is being developed at an accelerating pace, one which will lead humanity to a point which will be incomprehensible to people living today. And, since the pace of technological development accelerates exponentially, this will happen much sooner than we think. In the terms of literary movements within science fiction, this is cyberpunk taken to the extreme. I believe the first story to present this idea in its full form was Vernor Vinge’s Marooned in Realtime, and it was recently explored in more detail in Charles Stross’s Accelerando.

I think the idea of the singularity has led to some good stories, but I don’t find the idea to be particularly plausible. I don’t think the pace of technological change is accelerating, and I don’t think there is any reason to think it can continue to increase.

In terms of how technology affects daily life, the biggest changes happened more than a generation ago. The locomotive and the steam ship, and then the airplane, dramatically increased the speed at which people were able to travel. First gas lights and then electrification freed people from the day/night cycle. The automobile permitted people to live away from their work. Washing machines and vaccum cleaners significantly cut down on the time-consuming drudgery of housework, and freed women to work outside the home. The television permitted people to retreat into their homes for entertainment. Vaccines against smallpox and polio freed millions of people from early death and disability. In general many improvements in medicine and agriculture kept more people alive, leading to a worldwide population explosion that started fifty years ago, and has started to slow down, at least in the second derivative, only recently.

All of the these things happened at least a generation ago. Recently we’ve seen computers and the Internet. There has been a steady improvement in medical technology, but nothing nearly as dramatic as the polio vaccine. Otherwise, recent technological improvement has merely been a steady refinement in existing technology. I don’t want to belittle this improvement, but its effect on daily life is minimal compared to the effect of, say, the automobile. Computers and the Internet are certainly a significant technological innovation, but I don’t think they change daily life as much as, say, the television.

So the argument that the rate of technological innovation is increasing, much less that it is increasing exponentially, seems very weak. We can see some significant developments potentially coming in the future, such as designer organisms, nanotechnology, or fusion power. And, no doubt, new things we haven’t even thought of yet. But none of these developments are going to happen soon. And when and if they do happen, they aren’t going to make the other developments happen any sooner.

The only new technology which I can imagine leading to a real increase in the pace of development would be true artificial intelligence. An artificial intelligence would not necessarily be any smarter than we are, though it might be. Either way, it could be focused and replicated. That could accelerate the pace of change. (I discount another possibility, that of uploading human minds into software, because I think it is quite implausible.) But there is no reason to think that true artificial intelligence is going to arrive any time soon. The lesson of the last fifty years of research into AI has been one of clearing away our misconceptions about the nature of intelligence, but while the elimination of negative ideas is valuable it has left us with almost nothing in the way of positive ideas. HAL 9000 still seems as far away today as the year 2001 did when the movie was made in 1968.

Even if we should develop a true artificial intelligence, the singularity still seems remote, because I believe that humans can only handle a certain degree of complexity. People resist things they can not understand. New things are not widely adopted until they become familiar. Also, new things often threaten entrenched interests, which produce another type of resistance. I believe these human characteristics set another limit on the pace of technological change. I see no reason to expect that these characteristics will change in the future.

To give the singularity its due, it is undeniably true that life is becoming increasingly complex. However, this complexity is being hidden. As a computer programmer, I would say that it is being hidden by a set of abstract interfaces which present a simplified view which is all we need to understand in practice.

To explain this, let me start by saying that as recently as the Renaissance it was possible for one exceptional person to understand almost everything there was to know about technology. People like Leonardo da Vinci or Isaac Newton could know almost everything there was to know about the technology of their day. They didn’t know everything–they didn’t know the myriad details known by a successful farmer or blacksmith–but they could have learned them.

That is not possible today. Today, any technological object contains a remarkable depth of complexity. I recently bought a child’s plastic swimming pool, and I was reminded that this simple object was constructed out of a petroleum product extracted from the ground, was shaped in a factory probably in China, was shipped around the world through a series of middleman, to eventually arrive at a store where it was displayed, promoted, and sold to me, using an electronic cash register which computed the change. Every one of those steps incorporates considerable complexity, to the point where there are humans who specialize in each one of them, focusing on nothing else. All that complexity is completely hidden from me by the relatively simple interface of the store. A modern Leonardo da Vinci could learn every one of those steps in detail. But then let’s move over to, say, the clothing aisle; he might learn all those steps in detail as well. But could he also master the details of the cosmetics aisle, and the electronics aisle? And we haven’t even left the store yet.

The truth is that modern life is far too complicated for any one person to understand in detail. And yet, it doesn’t matter. That is because we have developed abstractions which permit us to to understand what we need to know, while hiding what we do not need to know. And while these abstractions are vast simplifications, they are not normally dangerous simplifications; in almost all cases, we really do know what we need to know (I think that the modern financial markets may be an exception here, but that is a topic for a different essay). We can expect life to continue getting more complicated, but we can also expect to continue to develop useful abstractions to hide that complexity.

So I think the idea of the singularity, while an interesting one, is simply very unlikely to happen. Certainly not in our lifetimes, nor in the lifetimes of our children.


  1. Airs - Ian Lance Taylor » Apocalypse Now said,

    March 31, 2008 @ 9:49 pm

    […] Perhaps a better way to get at this is the notion popular with some SF writers of the singularity. As I’ve written before I don’t find the idea to be very likely. But perhaps one way to describe the singularity is the point in time beyond which life becomes unpredictable and perhaps incomprehensible. By that definition, I think the singularity happened early in the 20th century. Don’t expect some remarkable singularity to happen in the future–it already happened in the past. […]

  2. jimb said,

    April 7, 2008 @ 1:56 pm

    Speaking of AI:

    Have you read Jeff Hawkins’ On Intelligence? He lays out a theory of how intelligence works in humans that’s pretty plausible. The book’s worth reading even if you don’t buy it, because there is still a lot of interesting stuff there. His group has recently released a free-beer software library (surely patented, non-commercial use only) that implements the algorithm he believes the neocortex runs.

    Speaking of singular events:

    I’ve been struck lately by the huge significance of the invention of contraception. Before contraception, a lot of the societal strictures we laugh at now really did make sense: a pregnancy outside marriage, if brought to term, was quite likely to lead to a life of poverty and loneliness. Given the the strength of physical attraction, there was a real argument for parents and society exercising strict control over children and young adults.

    How many of our values and mores descend from the need to prevent people from having sex outside marriage, and prevent them from having sex too often within marriage? Does the list include some we don’t recognize as directly related to sex? By greatly reducing the associated risks, those values become safe to abandon, even if doing so does dismay the older generations.

    I’m especially interested in seeing which of the old-fashioned romantic ideals turn out to rest on needs that remain in force: I don’t expect the generation of children born in this decade to be completely free of scruples, even by our standards, regarding how they treat their SOs.

  3. Ian Lance Taylor said,

    April 7, 2008 @ 3:25 pm

    I haven’t read Hawkins. Thanks for the pointers.

    I agree that contraception makes a big difference. There are also general cultural differences here. In Sweden it is becoming increasingly common to have children without getting married, apparently in the expectation that the children will be raised by the community of friends even when the parents split up. I have two cousins once removed born to unmarried parents, and indeed their parents are no longer together. There is no social oppobrium. This does not contradict your point; in Sweden even children with no parents will not lead lives of poverty, though I don’t know about loneliness.

RSS feed for comments on this post · TrackBack URI

You must be logged in to post a comment.