Nanotechnology without Genies -- closing thoughts

 

I bought Engines of Creation on September 4, 1986, as soon as I saw it on the New Books table at the Stanford bookstore, and read it straight through. This was one of the great conversion experiences of my life. I believed that the world as we know it would come to an end within a couple of decades, to be replaced with "a world made new." I was the kind of person who gave copies of Engines to my friends and family, and talked about it to anyone who would listen. (In other contexts such people are known as evangelists, cult zombies, pests, etc., but of course none of that applies here ;-)

At this point I need to interrupt myself. When I first published this site, I wrote the following two paragraphs:

I joined the Foresight Institute as soon as I heard about it. My collection of Foresight Updates goes all the way back to issue #1. I was one of the first to buy into Eric Drexler's vision, and in the late 1980s and early 90s I spent much of my time thinking about how to make it happen. At the beginning of 1993, I signed up as Senior Associate #2 at Foresight. However, the turning point came about that time. I had already started coming to my senses in the fall of 1992.

The starting point was the observation that robotics at the nano level can't progress any faster than robotics in general. That was the first step towards 'calibration.' At the open mike session on the last day of the 1992 Foresight Conference, I made some remarks that were the germ of the arguments I am going to present here. Over the course of the next year I deprogrammed myself. By the fall of 1993, I was no longer a believer. The arguments presented here are based on notes written in 1992-3. They didn't start coming together into their present form until December of 1998, and I didn't fully appreciate the central importance of the interface problem until February of 1999, just a few weeks ago. It took a long time to think this through.

A few weeks after writing the above, I reviewed my notebooks from the late 80s and early 90s, and found that the truth is more complex. (Writing history from memory is generally not a good idea.) Apparently I never did believe in molecular manufacturing, except maybe in the first few months after I read Engines. I started "coming to my senses" a long time before 1992. There is a page dated November 10, 1988, which is almost an outline of Geniebusters. Most of the basic arguments against molecular manufacturing were already there, on that page (but not the arguments about AI). And yet I still drove to Palo Alto (400 miles) on several occasions to attend meetings, and I wrote checks for more than $2,000 over a period of years. I didn't believe, but I also did believe.

I also found things like this in the 1988 notebook:

You could have entire regions of zero entropy, spreading, crystallizing everything... You could have nanocreatures, robotic fish, that communicate with each other and are programmed to do parallel processing. They keep spawning more robotic fish, and connecting with each other... Trillions of robotic fish all linked together would be a supermind capable of anything.

This stuff goes on for many pages. In recent years I have come to think of myself as a nanocritic, and I tend to forget about my own nanofollies of 1988-92.

It was in 1993 that I started questioning AI. In January of that year I wrote:

The problem of creating an artificial intelligence is the same as the problem of creating an intelligence. Of any kind. Period. Teaching a robot is not going to be any easier than teaching a human being. And as for having trillions of robotic fish communicating with each other, that is the same problem as human beings communicating with each other. Ultraintelligence has to start with us. It is not something outside of us. It is something that proceeds from us. It is an extension of us.

At that point the pieces of the puzzle were all there in plain view, and I should have been able to put them together and draw a conclusion. Nevertheless just a few days later I wrote a letter to Bob Boyer in which I talked about nanotechnology leading to the end of the world. I didn't use the word "singularity," but that was the idea.

Even now, that idea is still seductive. It's easy enough to make fun of the Singularitarians, but at the same time it's almost impossible not to believe that we are headed for some kind of cosmic cataclysm in the near future. I already believed that when I was seven years old.

Belief systems are very strange. Getting rid of them is not easy.


Second Thoughts

June 29, 2000

In fact, I'm still open to the idea that we are in the midst of a great transformation or metamorphosis.

A couple of days ago I saw an article in Nature, which treated Engines of Creation with contempt. The article says there is usually no such thing as bad publicity, but Engines is so bad it has been an embarrassment to many people in the field. I quoted this at the end of section 28. Of course when I first read these remarks, I smiled with satisfaction. But on second thought, I think the author of this article is just a journalist who would sneer at any new idea. I don't want to be associated with such people. Engines of Creation is wrong, but it's not trivially wrong. It deserves to be taken seriously. I have more in common with Eric Drexler than with philistines who dismiss his ideas without coming to grips with them.

Tonight I was surfing and by a circuitous route I found myself reading the Encyclopedia Britannica article about the history of Iran. You have the Hellenistic period, the rise of the Parthians, the Sasanian period, and so on, century after century. Is history going to go on from our present situation, century after century, one period giving way to another? Two thousand years from now, will there be an encyclopedia article that describes our own time as just one period in a long series?

I don't think so. As Vinge says,

Perhaps it was the science-fiction writers who felt the first concrete impact... More and more, these writers felt an opaque wall across the future. Once, they could put such fantasies millions of years in the future. Now they saw that their most diligent extrapolations resulted in the unknowable... soon.

There certainly is a "wall" in front of us. Science is rushing toward some kind of climax.

In the last seven or eight years I have typecast myself as a nanocritic. It's time to break out of that role.

I am not repudiating anything I wrote on the original geniebusters site. I still think Eric Drexler and Vernor Vinge are totally wrong about what's happening, and what's going to happen. They aren't even looking in the right direction. They both think that the singularity will emerge from Marvin Minsky's AI lab. This is an illusion.

The 20 - 30 year time frame is probably also an illusion. It seems to be a moving frame.

But the question is not whether some momentous change is happening, the question is what is happening. What kind of situation are we in? The worst thing you can do is to have a belief about this. I am in danger of getting locked into a "steady state" belief, which is no better than believing in the Singularity. We have to stay open to events as they unfold.

We should also remember the Merkle Maxim (originally due to Alan Kay): the best way to predict the future is to create it.

The operative question is, which research agenda is going to bear fruit?

Let's stop arguing and get back to work.


back to the table of contents

home