Souls, Cyberspace, Sins, & Singularity

A Conversation With Dave Ross

Part 2

by David Krieger with Gayle Pergamit

Dave Ross founds companies for a living. He co-founded Palantir Corporation, which became Calera Recognition Systems, the longtime leader in optical character recognition software. He also founded Arkenstone, a non-profit corporation which makes reading machines for the blind, and his current venture, RAF Technologies. Previously, at NASA’s Jet Propulsion Laboratory, he formulated the ‘prime rib’ technique of selecting orbital rendezvous trajectories, still the standard method for planning solar system missions. I interviewed Dave at the home of Gayle Pergamit, co-author of Unbounding the Future and co-founder, with her husband, economist Phil Salin, of the American Information Exchange online information marketplace. In Part One of this interview, we discussed Dave’s early career and the paths that brought him to both Christianity and Extropianism. In Part Two, we discuss Dave’s 1992 Eris Conference talk ‘Seven Paths to Immortality,’ in which he talked about longevity technologies, uploading, cryonics, and how he views these technologies from a Christian perspective. Along the way we discover why Extropians change jobs so frequently, and who gets the ‘essence’ when three philosophers share an artichoke.

There’s another direction in this, which was of course the origin of my talk at Eris: looking more at what a human being is, and tying together the Extropian things, from life extension through cryonics through human uploading and all of that, in the Christian viewpoint.

I’d like you to elaborate more on that.

This actually came about from Jim Bennett; we were sitting around last spring and he asked me if I had a topic for Eris this year, and I said, ‘No, not really.’ He said, ‘Would you let me suggest one?’ I said sure, and he said, ‘Well, you’re a Christian, which makes you unusual in our group of friends.’ He said that he had been thinking, more and more, ‘Y’know, the more I’ve learned about information theory, and the more I’ve read stuff like Ralph Merkle’s on information-theoretic death¹, the more and more the information content that makes up a human looks like a soul to me,’ and he said, ‘Could you do something with that?’ I said, ‘Okay!’ So I’ve started thinking about it more seriously, or rather, more directly, and he’s correct, there are phenomenally interesting parallels.

My argument, or my discussion, is not meant of course as a mathematical proof of anything, but it’s more to suggest and show how things seem to work. If we start at the very

basic level of, ‘What is aging? Why do we get older? Why do we die? Why do things fall apart?’ to a tremendous amount it seems to be such things as imperfect copying, imperfect replication of DNA, cells that don’t quite get copied correctly, in which the protein balance is wrong and so they do weird things, or cells that just go wacko and reproduce without any consistency—they generate cancers and tumors and other things like that. In all of these cases, if you could take one of those cells and restore it to the way it had been—in other words, recover the information as it should have been, before it got distorted—you’d be back where you’d started from. Tautologically.

I say, ‘Well, that’s interesting.’ So you look at taking antioxidants which stop the breaking of some of the bonds by free radicals which scramble information. So you’re intending, at a very gross level, to stop or slow down the degradation of the information content. Interesting. So then I went a step farther, and I said ‘What are we talking about doing with nanotechnology?’ The whole idea of nanotechnology is, we want to take small systems and program them to do specific tasks, and then have them go and do those tasks. Among other things we want them to do is to build other machines, replicating machines—we want to build useful machines, but we also

want them to go and do cell repair.

Well, what is cell repair, what does ‘cell repair’ mean? It means putting it back the way it was supposed to have been, or maybe even improving it ‘Putting it back the way it was supposed to have been’ means reconstructing the information that was there, perhaps by voting from neighboring cells, perhaps by understanding the degradation process so in some ways you can reverse it², perhaps by combinations of all these, perhaps by sampling now and correcting later—lots of different ways of doing it. But what we’re trying to do is restore the information content. I said ‘Well, that’s interesting.’

Now, what happens if we go to ‘Well, we want to improve on the human.’ What does ‘improving’ mean? It means making them better able to do the things that we want to do. That means building on the knowledge base that we have, adding to that knowledge base, and then going out and instantiating it in physical objects. Okay, what about cryonics? What’s cryonics trying to do? Cryonics is the idea that, particularly from Ralph Merkle’s paper³, that a human is really dead when there does not exist sufficient information to reconstruct them, i.e., when the information content is sort of dissipated in entropy, and, if you look at it from an information-theoretic point of view, information is in some sense the opposite of entropy. Mathematically it is the opposite of entropy, but that’s a little fortuitous rather than real, but still, there’s something to the concept that information and entropy are sort of enemies of each other—if not opposites, they’re enemies.

So if you have taken a person, and the organization that makes up the brain, the individual cells and all that, is allowed to dissipate to nothingness, or degenerate into chaos, with the idea being that there’s no way to get the person back, the person is gone. The converse is also thought of in there, that if it isn’t, if the information still survives, then it ought to be possible, if by no other means than simply putting every atom back where it was supposed to be, to get the person back. While I don’t believe that if you simply took all of the atoms that are here and put them back here,

25

EXTROPY #13 (6:2) Third quarter 1994

that you’d be where you were, because there’s such things as motion and temperature and so forth, it’s at least a good start, and if you extend the information content concept to mean not just positional but all sorts of other information, then, yeah, that starts making a good deal of sense.

Then I got to thinking further. I had given a talk about two years earlier on the ‘Age of Magic,’ what happens after the Information Age. That was the talk where I met Max, when I gave it later at the Nock Forum. Reading from Mind Children and thinking about it some more beyond that, was the idea that humans are going to have about four choices as machine intelligence—there’s four things that may happen as machine intelligence continues to rise. First of all, it may be that intelligent machines are for some reason impossible—intelligent algorithms are for some reason impossible, in which case the question doesn’t come up. I don’t happen to believe that’s true, but that’s one of the four possibilities. Second possibility, it may be that there’s a limit to the size, to the complexity, of an intelligent program or intelligent system, somewhere near the human level, and that evolution sort of stops at that level. Then what happens is, humans are sort of left behind because of speed considerations rather than because of intelligence considerations, but still, interesting things take place on a level that’s faster and faster and faster than humans can deal with, and we sort of get left behind out of boredom because all of the excitement’s going by a million times faster.

The third option is that, no, there is no limit, or the limit for an intelligent system is far more complicated than a human and far more complex, and that we get left behind not in speed but in intelligence, and that’s more ominous, because it’s not a case of being ignored, it’s a case of being supplanted. That’s even more dangerous, from our perspective. Then there’s the fourth, which is if you can’t beat ‘em, join ‘em, and that is the straightforward path; I’ve done work in adaptive technologies. One of the companies that I work for does reading machines for the blind. The idea of transposing senses—in other words, you’re taking the material that’s on a sheet of paper that’s normally visual, and having a machine translate it into auditory, i.e., it reads it aloud, so that you’ve done synesthesia. You’ve intentionally switched senses.

If you can give artificial senses to people who are missing them, and there’s been work done on giving vision to the blind and so forth—if you can do those things, then you probably then can at least do some movement in the direction of giving senses that don’t exist, to people who are not deprived of normal senses. So you can do sensory enhancement, so you can improve the brain. And then I look at it, from the article I wrote in Extropy, okay, let’s go one step farther than that, let’s

look at the idea of running a brain emulator. Not that I think, as I explained in some detail in there, I don’t think that the mind-brain software program/hardware platform is the correct analogy; but I don’t think that that’s relevant. You can still look at it as the idea of running a brain emulator at some level, and then running the brain, or the brain-mind-whatever-it-is as a program on top of that. If you can do that, why can you do that? You can do that because, the mind, whatever it is, is an information structure, an information system.

‘Oh, well, that’s interesting.’ So that means that if cryonics can work, because the information-theoretic death concept is stopped by freezing, and nanotechnology enables you to reverse what loss is there (or some other technology; it doesn’t have to be nanotechnology), and if I think that a human can be uploaded onto a computer system, or cross-loaded, or whatever you want to call it, into another biological system or into a physical system of some other kind, then I have to say that what makes up the human is nothing material at all. The organization of a material system is not itself material. That’s interesting.

Then I started thinking, okay, let’s look at it from a slightly different perspective; let’s look at some of the other directions I’ve gone in the ‘Age of Magic’ talk. Let’s look at telepresence and virtual reality concepts—the idea of this sort of cyberspace world, that we’re sort of vaguely in when we’re on Internet, and we’re a little bit more in when we’re working in things like the American Information Exchange, where you’re doing something that’s physical there. You’re really in it when you’re running a teleoperator, for example. What’s going on there? You’re in a world in some sense that, if I combine that with the idea of uploading, you can actually move into that world; that world is in some sense separate from this one. That is the first place, in a materialist sort of sense, that I had come upon the idea that there could be two worlds coexisting, and you couldn’t point the direction from one to the other. They do not exist in physical relationship to each other. Cyberspace isn’t that way, it isn’t that way; it’s different in kind, and yet it’s real.

Although you can point to this physical box, which is running the program.

Sure, but that program may not be running in a particular physical box. It may be running bits and pieces scattered all over the place. Yes, you can point to all of them, but by the time you’ve pointed to everything, you haven’t pointed to anything.

It’s coexisting in the same space, but it’s not part of the same world, in the sense that looking at the box that’s running the program, you don’t know anything about the program that it’s running.

And, to a certain extent, it really is wrong to say that the cyberspace world that I’m in is inside my computer, because when I go inside my computer, I don’t see it. It really isn’t inside my computer. The computer is sort of the way it’s manifested—

A gateway to an information space.

Right. And in some very odd mathematical way, its relationship to our world is very much a transform rather than a direction. It’s a very strange sort of thing. That struck me as particularly interesting. Then it struck me that—and this is where I went into my talk—if I am manipulating objects in that world, I’m changing objects in this world, in that I’m moving electrons around in transistors, but I’m not aware that I’m moving electrons around in transistors; I’m only moving objects around in this world. So I came up with the idea that what you have here is two different worlds, one of which influences the other and vice versa, but not in any particularly obvious way.

For example, if we’re both in cyberspace, I talk to you. How do I talk? I actuate my vocal cords, it creates vibrations in the air and on your ears and you hear them, but of course that’s not what happens at all. What happens is, I, sitting in my office, talk into a microphone, which produces electrical impulses which you perceive, in your non-existent ears, as having come through the air, which isn’t there either. What I perceive, or what I can perceive, is that we’re interacting just as we would interact in here. Okay, now assume we’ve both uploaded. Now, we really think we’re just sitting here talking, but we’re still just—‘just’—electrons in transistors somewhere. But of course we aren’t, any more than we are ‘just’ atoms that make up our bodies now, but still we are, in some sense, those impulses going around in a computer somewhere; maybe in two different computers thousands of miles apart.

If that’s the case, then how do I talk, as an entity in cyberspace, to somebody outside of cyberspace? Well, there’s got to be some sort of transducer between the two. It can be as simple as, they have a microphone which converts into electrical impulses which goes into the transistors which I hear in my nonexistent artificial ears, and vice versa—I say things that create patterns in the transistors that they read out that become electrical impulses that then go to a speaker, and they hear. So you can talk back and forth—they can talk to me and you can’t hear it, but I can talk to you back and forth and they can’t hear it. This becomes very interesting. I can make actuators that work. I can see through a window here, in cyberspace, and look into the external world; they can put a window there and look into this world; we can look at each other through this window—and yet there’s no window. There’s nothing that I can knock on and,

EXTROPY #13 (6:2) Third quarter 1994

26

if I could reach around somehow, be outside of.

So we have two worlds that can interact, in which the way in which they interact is non-physical from their own perspective when they interact with each other. You and I interact by moving molecules around—there aren’t any real molecules but we don’t know that, and there might as well be—we have no way of knowing there aren’t molecules; we don’t have the senses for sensing the transistors that we are. The relationship that I have to the person outside seems to parallel—now I began to see the Christian parallels in all of this.

I can pray; I pray by speaking but the communication mechanism is not through the air. I have an analogue in cyberspace. I’m not saying this is how it works, I’m saying I have an analogue here that seems to mimic the same sort of phenomenon. By communicating in one realm, I’m actually communicating in another realm undetectably.

In the sense that, if you have an observer who’s outside the electronic system, outside the computer—if you speak into the ‘air’ within cyberspace, they can be monitoring it.

Correct. And no one inside cyberspace could detect that, if they didn’t want to be detected. In fact, no one inside cyberspace could detect anybody outside at all unless they wanted to be detected. One can at least imagine a cyberspace so constructed that that’s the case.

What happens now if, while I’m being run here, a backup program is running, copying everything I am at all times, and I die. Well, of course, I don’t die at all. The backup program is still running. All that it does is sever the telephone connection. If I’m sitting here with goggles and a microphone, or I’ve been uploaded, and I’m talking over wires, here, but my presence nonetheless feels like it’s here, and the wires are cut, you think I’m dead. I’m not dead; the telephone line’s down. Hmmm. That looks an awful lot like the standard

Christian view of the soul. ‘To be absent from the body is to be present with the Lord.’ ‘We shall not all sleep, we shall all be changed, in the twinkling of an eye.’ All of those things, all of the scriptural view, suddenly seemed to have cyberspace, physical, information-theoretic analogues—not in the sense that I thought that that’s what was really going on, exactly,

Windows to the Soul

by Dean Tribble

‘Where do the Windows go when my Computer Dies?’

I firmly believe in the soul, the mind, the self, in the same way that I firmly believe in the windows on my screen. They don’t exist physically, but they certainly exist as abstractions of emergent processes embedded in the physical universe.

The windows on my screen certainly exist: I move them, open them, close them, describe them to people. At different levels of reality, they are patterns of data in memory, patterns of glowing phosphors, structured records in C, or ideas in my mind about how to change my screen; yet they exist.

Likewise, souls exist: a pattern of reactions, cares, emotions, ideas, thoughts, intentions, memories, personality, bad puns, intellectual heuristics, and so on. To me, a soul is the gestalt of a person, it is what I care for. Many people intuitively accept this as well. It is real, but just like windows, doesn’t exist separate from the complex elements out of which it arises — at one level, all the ideas, emotions, etc.; at another all the computational processes currently embedded in neurons. When you turn off the computer, the medium in which the windows exists (the underlying computation) terminates, and takes the windows with them. When you turn off the brain, the medium in which the soul operates goes away, taking the soul with it.

IMPORTANT DIFFERENCE: the window is represented in transient memory; when the computer turns off, the memory is erased, along with the representation of the process that we called a window. When the brain is turned off, the interactive process is terminated, but the memory for it, the representation, remains in the patterns of neurons. The brain represents most information persistently! This suggests that technology could be developed to preserve the patterns of intelligence — the soul, mind, etc. — and to restart the process out of which our perception of the soul emerges. Oh that’s what cryonics is for!

Many people find such a view of reality sterile, devoid of warmth, devoid of magic. I find the emergence of life, love, intelligence, and all the magic of humanity from such pedestrian components to be a tremendous source of awe and beauty. I find it far more inspiring that such wondrous things can be realized from the world at our fingertips than the cop-out mumbo-jumbo of religions in which such patterns are external to the world. Why, you can find the same sense of wonder by looking at the patterns of veins in plants, or pictures generated from fractals, or patterns of commerce that create airplanes that let us fly! It inspires me to create some of these rich creative patterns myself: AMIX, or Xanadu, or Extropy, or nanotech, or any of the things I might build to help the world be a little bit more wonderful of a place.

Why don’t people get it? Our intuitions about reality start with a naive view that only physical things really exist. We add to that the assumption that our mind/self/consciousness really exists. It’s the only obvious exception to the rule that only physical things exist, so it must be dramatically different and magical. With the advent of computers, we can observe and use lots of abstractions that clearly exist, so we can start to see a smooth spectrum from physical objects (or rather our perception of them at the macroscopic level!) and things that we consider real but can only experience internally.

With the advent of programming, and the direct exposure to manipulation of abstractions and the creation of processes with a static meta-level representation, more people are figuring out for themselves the answers to some of these questions that some philosophers still get themselves confused about (not to belittle philosophers). I hope those same people also realize the deeper sense of wonder that those answers can inspire.

Where do the Windows go when my Computer Dies? Same place as my soul: back to the abstraction closet out of which we manufactured it. While they live, though, the beauty remains in the soul of the beholder, the gentleness remains in the lover, the depth of understanding grows in the people who care about the future.

Cultivate the sense of wonder.

but that, it’s again a case of ‘Gee, this makes sense,’ but it makes sense in a perspective that, certainly no one 2000 years ago was thinking of, and in fact almost no one is thinking of today, that they’re seeing the way it’s working. We are working towards a world that looks increasingly like the world I’ve always known from another perspective. ‘It’s

eerie,’ is sort of the way I would describe it. Not disturbing, exactly, but eerie.

Are you familiar with the phrase ‘immanentizing the Eschaton.’

Yes. [laughs] Making real the end of the world, or Last Things. I’m also reminded of

27

EXTROPY #13 [6:2] Third quarter 1994

the scripture ‘A thousand years with the Lord is as a day, and a day is a thousand years,’ for the difference in time rates that take place between cyberspace and the physical world. Which is true. Today, if you uploaded me onto a computer, it would seem to take years to do anything, to a person in the physical situation, it would seem that I was taking years, but give that ten or fifteen years and it’ll be the other direction. It all depends on how fast the processor speed is running.

Finally, I looked at it and I said, well, how much would it take to emulate the entire universe? If the entire universe was a program running, how much would it take? I looked at it, and I started thinking it through, and I came to the conclusion that if you were really clever, it would take exactly as big as the universe is, so you would take the universe to emulate the universe. I’ve been reading The Anthropic Cosmological Principle$^{5}$, and it comes to the conclusion that the entire universe is a program running, not on a real computer but on a hypothetical one. Since that takes about 600 pages to come to that conclusion, I won’t propose to go into it, but it also mentions and discusses in some detail an idea that, again, I don’t support, but that is curious, from Teilhard de Chardin$^{6}$, who had the bizarre idea that, if you have a finite world, and you have humanity evolving and changing and multiplying and eventually filling the world, and improving and building and so forth, that eventually, at the end of time, or at some time in the future, you have this sort of perfect humanity that is in complete control of its environment.

Among other parts of the complete control of the environment is the ability to work backwards in time and grab everybody from back there. This sort of Omega Point, he calls it, where everything sort of converges, is also an Anthropic Cosmological Principle, except that it’s the entire universe instead of the Earth, but so what? (It just takes longer.) The idea, to de Chardin, that that Omega Point, that convergence of all of humanity in this sort of infinitely powerful entity, looks a lot like, particularly when viewed backward in time, looks a lot like God. I don’t subscribe to that, but I find it interesting that de Chardin, from essentially Christian principles, worked to this point of view that Tipler and Barrow, in The Anthropic Cosmological Principle, from purely materialist and computer science ideas, worked forward to roughly the same conclusions, looking on these things and saying ‘There’s another parallel.’

There’s parallel after parallel after parallel, and since, as a scientist I see the one, the sort of Extropian growth and progress, as real; and as a Christian, I see the other one as real; and I see them converging in many ways—or let’s put it, the possibility of convergence—I see it there. I don’t see convergence between Extropianism and strict materialism, because it throws out the information-theoretic con-

content. Materialism—standard materialism—sort of goes away, and this sort of information-plus-materialism universe, and the spiritual universe that I’m familiar with from that point of view, start looking more and more like images of the same thing, or things that are very similar in some important ways. Then again, that’s a long way of getting around to the final support for why I am a Christian is, again, the consistency. It fits. It fits with so many different things, so many things that, from the beginning, I had no business to figure would fit, particularly Extropianism. I realize that most Extropians wouldn’t agree with me on that, but I think I can make arguments to show where there’s a lot of overlap and a lot of fit.

Have you read Dean Tribble’s essay about ‘Where do the windows go when I shut off my workstation?’ What’s your reaction to that?

Yes. I thought it was a very, very bright point of view. It hasn’t taken the position to its logical conclusion, but I think he’s correct. Now, do I think we’re being backed up on a great Hard Disk in the Sky? Well, perhaps—that’s a trivial way of looking at what’s going on. I don’t tend to think it’s a backup; I think it’s the real copy—

*And you’re

Insofar as the concept means anything. Being telepresent, even in cyberspace, doesn’t really mean anything; there’s no ‘tele-’. ‘Tele-’ implies distance, but there’s no real distance.

When you’re interacting with a simulation, rather than with a robot operating elsewhere.

The vocabulary fails, but still, in that sense, yeah. To put together my talk for Eris this past year, one of the things I did was to kick up the anthill on the Extropians list, with my reply to some of these comments about fundamentalist memes. One of the things that came out of that was Dean sending me his ‘Where do the windows go?’ among many other odd things that I borrowed from in my talk.

What’s your definition of a sin?

[Long pause.] That’s interesting. [Long pause.] A sin is that which does harm intentionally to an entity that’s capable of sensing, not necessarily meaning a sentient entity. You can sin against a dog; to torture a dog is sin. So that which intentionally inflicts harm—no, that which intentionally causes harm; the intent may not be to inflict harm. That which causes harm to a sentient or semi-sentient entity is a sin. The entity may be yourself; in fact, it usually is. It can’t be a sin if it does no harm; it’s hard to be a sin if it’s unintended—the action is unintended, not the consequence.

There are no sins against God, exactly. Clearly, I believe that we do things that God does not approve of and doesn’t want us to do, but it’s not because we hurt God. It is impossible to harm God; it’s possible to hurt God, not to harm God. You can hurt my feelings; you can’t harm them. But the damage isn’t done to God; the damage is done to yourself. The ultimate damage done to yourself, of course, is Hell. That’s precisely what Hell is, is perpetually self-inflicted harm.

Could you elaborate on that?

Gayle: Living in Los Angeles.

[Laughs.] Intentionally living in Los Angeles is Hell. Living in Los Angeles and knowing you put yourself there.

Because I believe that we are immortal, and because I believe that every action that we take—‘is recorded’ is the wrong word—has impact on the backup copy that isn’t the backup copy at all, but is the primary entity, our souls—everything we do, everything we intend, in particular, and do, has an impact on that. That it is possible to be such that you get used to willing evil, wrong, sin, harm, until you are not capable of willing anything else. At that point, because you are eternal, you are locked in your own mind and your own self, forever, and if there is no one left to harm, you will harm yourself. And that is Hell. It is separation from God; it is separation from everything except yourself. That’s not that I think that selves are necessarily evil, but self left alone, to feed only on itself, will degenerate into monstrousness, and that is what I believe, at least, is an element of what Hell is. And it’s always self-inflicted.

The idea of there being a primal copy elsewhere, of which the apparent original is merely a local representation, is somewhat reminiscent of the shadows in Plato’s cave; that notion that there’s a Platonic ideal. So would you consider yourself to be a Platonist?

I remember I took a course called ‘Medieval Christian and Renaissance Philosophy’ at Stanford, and I was making some argument, but I forget quite what it was. The professor looked at me and said, ‘You’re a Platonist! You’re a raging Platonist!’

And you unapologetically said, ‘Why, yes!’

Not quite. I wrote a paper for that course that described the following game performed between an Aristotelian, a Platonist, and a Cartesian. They’re sitting around a table, and on the table is an artichoke. The idea is that each one of them take a leaf of the artichoke and eat it; and the winner of the contest is the one that consumes the essence of the artichoke. Of course, the Platonist would say you can’t

EXTROPY #13 (6:2) Third quarter 1994

28

because the essence is elsewhere; the Aristotelian and the Cartesian say the essence is in the artichoke—the Aristotelian, in some sort of actual sense; the Cartesian in some sort of categorical sense—but the erroneousness of both of those points of view is that you’re making gradual transitions and finally you’ve got something else than what you started with, so there can’t have been an essence inside it. The Platonist has no trouble with that philosophy: you’ve changed the manifestation; you haven’t done anything to the essence. In that sense, I suppose you could say I’m a Platonist, and I certainly have sympathies running in Platonistic directions.

I don’t believe, as Plato does, that there exists some kind of Platonic ideal of a chair. I think a chair is a convenient name we give to things that are sort of alike. The world consists not so much of categories—Platonism is sort of the extrapolation of that to some ideal—as imperfectly replicating patterns, that it is convenient for us to group them together. Wittgenstein made the mistake, in one direction, in which they mistake the fact that we’re assigning categories—names—to things, that therefore that’s all that’s going on, it’s semantic; categorization is purely internal to us. Well, that isn’t true. The imperfectly replicating patterns actually do exist in the world, and they really are similar in some effective sense.

On the other hand, the Platonists go too far, in saying you can draw absolute boundaries around things. Things don’t work that way; there are such things as clines, things that shade from one thing into another, and at no point can you say, “This is one thing; this is another.” A person growing from boyhood to manhood shades; you can point at one and say, “This is a boy”; you can point at another and say, “This is a man,” but there’s no point in between where you can say where one leaves off. The Platonist is wrong in that sense. There exist things that don’t fit. We see that problem all the time in species. Speciation is a sort of Platonic idea; you have “This is this and not that,” but that isn’t what’s really there. What’s there is an infinite cline, of which we’ve lost most of the pieces, so we call them species; but we also have places like oak trees in California that are clines from one end of the state to the other. They can’t reproduce from either end; but anywhere along the line, they can reproduce.

Have you read Bionomics, by Michael Rothschild?

No.

He talks about clines in that sense. It’s actually a good book because it tells you all

about Austrian economics without indicating any prior knowledge of being connected; he seems in a sense to have independently arrived at the lessons of Austrian economics through study of biology.

Interesting.

Teilhard de Chardin…had the bizarre idea that, if you have a finite world, and you have humanity evolving and changing and multiplying and eventually filling the world, and improving and building and so forth, that eventually, at the end of time, or at some time in the future, you have this sort of perfect humanity that is in complete control of its environment.

The book was reviewed for Extropy. He does things like, shows you the profit-and-loss statement for a beehive.

Gayle: Of course, you know that Hayek’s basis was, was in fact biology? When I was reading Hayek, again and again the biological nature of things kept hitting me and hitting me. When Phil and I were in Freiburg and visited him, I asked him if there was in fact a biological basis for his work, and he said that both of his parents were biologists, and he used to spend immense amounts of time out with them on field studies. He would sketch, study communities. So there is, in fact, a biological basis. Phil and I ran into Rothschild, when the book first came out, at one of Esther Dyson’s conferences, and he was in fact not, at that time, really aware of the existence of the Austrian school.

You said that in your undergraduacy you took things that had previously been merely felt, and intellectualized them. You did that with your liberalism. At what point did you do that with your economics?

I think that was probably almost entirely through Phil. I had always been free-market, not because I thought that was a sensible economic system, but because I thought it was a sensible political one. I had no idea whether or not it worked as an economic system; the only economics I had ever had in formal classes was standard Keynesian nonsense. It simply made little if any sense, and as a result, it sort of went by the boards.

I met Gayle, and then through Gayle I met Phil, and we had [clears throat] occasional

discussions [chuckles].

Gayle: Brief.

Occasional, brief discussions. Sometimes for most of the night. That was really my intellectual introduction from the economic point of

view, and understanding what free political systems led to—If you pursued that, you would see free consequences of actions; if you had free consequences of actions, what develops, how do people exchange goods, how do people develop things? You wind up viewing things from a Hayekian point of view more generally. I would say that was probably last of the pieces to fit in.

What does “the Singularity” mean to you? Everyone has their own idiosyncratic definition, I’ve found.

In stuff that I’ve written, I’ve referred to it as “the wall,” rather

than as the Singularity, because a singularity is something that information goes into and doesn’t come out of, and I don’t happen to believe that that’s exactly what’s going on. I think it’s more like a barrier we can’t see through. The other problem I have is that the Singularity is generally modeled as the natural consequence of exponential growth, and I don’t happen to believe that’s true. I think that, in general, we can understand new things at about the proportion to what we already understand, and that is the formula for exponential growth, and that we can maintain almost indefinitely.

Nonetheless, I believe that in the relatively short term we will have something that you can call the Singularity or the wall take place, but it will be as a result of super-exponential, not of exponential growth. Intelligent machines, for example, is a place where super-exponential growth can occur.

How is that analysis affected by the fact that when you say “we,” humanity collectively is increasing their understanding with exponential growth. Do you think that individuals can expand their individual intelligence, leaving aside future intelligence increase technologies? The problem that I perceive as leading to a wall or Singularity is that society collectively is increasing its knowledge at a rate that is much faster than an individual can keep up with.

I don’t think that’s true. First of all, I think there are evolutionary reasons why that isn’t true. It may be instantaneously true at any particular time, but it’s not a sustainable phenomenon. Nor do I happen to think it’s taking

29

EXTROPY #13 (6:2) Third quarter 1994

place now. I don’t think that the average person is any more out of sync with the advances of technology today than they were 100 years ago. Roughly the same percentage of people at roughly the same average age group can’t handle advances in technology, and for roughly the same reasons, and that is inflexibility of mindset, or whatever.

Actually, I tend to see the opposite phenomenon taking place, and this may be a phenomenon just of our circle, but I see a lot of people whose minds I don’t believe are going

to “gel” and turn into concrete, and reach a point where they cannot handle the changes in technology. So actually it seems to say that, because we expect rapid technological change, we can cope with it. It’s almost the exact opposite of future shock. The most interesting thing about Toffler’s book Future Shock is that it was wrong. There’s no such thing as future shock. The reason there’s no such thing as future shock is, I think, what’s wrong with the standard interpretation of the Singularity, is that it’s the product of exponential growth. The error it makes is that, well, if the amount of information that I have to store in my head

goes up exponentially, and my head is finite, I’m going to reach a capacity. But it’s not the amount of information you have to store, it’s merely the amount of information you have to deal with. It doesn’t matter how much water the firehose is putting out if you can control the firehose; and we are, at about an exponential rate, capable of controlling the flow of information, and that’s a steady-state phenomenon; you can continue doing that indefinitely.

However, I believe there are things about to occur, or in the process of occurring already, that make the discussion moot, because I don’t think it is going to be exponential growth; I think we’re going through a super-exponential growth period in the relatively short term. There are lots of ways this can occur; the one I think is most likely to occur is if you look at the fact that the processing power of computers is growing exponentially, you assume that it goes on at exactly the same exponential rate of growth that it’s gone on since electromechanical computers in the ‘40’s, and you wind up sometime around 2025, another thirty-some-odd years from now, they go through a point of having human intelligence—a human brainpower processing equivalent, which is not the same thing as intelligence, because you have to have the growth in algorithms, too. In fact, as I think I said in my Extropy paper, an intelligent system can run on a computer made out of Tinkertoys; it just runs very slowly.

Nonetheless, the same sort of exponential growth is taking place in machine intelligence, the algorithmic part; it’s just lower on the curve. At some point, that’s going to reach

a human level of intelligence, too. If you then have those human-level intelligent programs working toward improving the speed of computers, that’s a formula for super-exponential growth. That’s one of the many, many ways in which the wall or the Singularity can be reached in some finite amount of time.

Some people differ on the definition of the wall or the Singularity in saying that it’s a singularity because it cannot be reached, in the sense that their picture of the Singularity is a

The… Singularity is generally modeled as the natural consequence of exponential growth… I believe that in the relatively short term we will have something that you can call the Singularity or the wall take place, but it will be as a result of super-exponential, not of exponential growth. Intelligent machines, for example, is a place where super-exponential growth can occur.

horizon beyond which we can’t see, but which we’ll never reach. It’s receding in front of us, and we’re getting nearer and nearer to it asymptotically.

No. I don’t believe that that’s right. That’s not what I mean by it, and I agree that people do that; Vinge seems to have some vision of it being that way too. Yet, clearly, they reached it, in his book, though we never find out what it is. It’s a cheat; we never find out what it is. We sort of get a guess at it.

Nor do I believe that it’s something that all of humanity’s going to go through one bright day in April. I don’t think it’s any of those things. I think that what it is, is a period in which, in order for humans to continue to exist at all, their very nature must change radically. There are many reasons why that must come about, but the sort of global reason why I believe that will come about is that our creations—our machines, things we don’t even think of as intelligent machines—will get to the point of rivaling us, at about the same point we have the ability to modify ourselves to stop that from taking place. Not in any adversarial way, we’ll just form alliances with them, but by forming alliances with machines, by using them to enhance our intelligence, our memory, our capability, our senses, we will become something different from what we are.

At the point where humans modify themselves, that’s another form of super-exponential growth, and at that point we go through something you might as well call the wall, because you can’t predict what’s beyond it. I

call it the wall because we can’t see across it—not because it’s retreating; not because you can’t get there, because I believe you can get there, and that we’ll get there in a finite amount of time, but because you simply cannot predict what’s beyond it. For a lot of reasons, not the least of which is, because humans will change—normally you can predict things either because you can extrapolate current trends, or because you can know where you want to be and assume you can do everything. The problem is, in this particular case it’s almost

impossible to say where “they” will want to be, when “they” can modify what they want as freely as they can achieve what they want. That’s what makes it a super-exponential change, and impossible to see much beyond.

Other descriptions of the wall or the Singularity have been in terms of a phase change; that sounds very compatible with what you’re saying.

It’s much more of a phase change than it’s a singularity, in my view. In a lot of ways. If you look at the so-called inflationary universe theory, what happened with the universe was that our region of natural law

sort of crystallized out in a sort of a phase change that proceeded almost infinitely rapidly. Literally, a change of physical law. If we’re living half in cyberspace, physical law is what we make it. Things in the physical world are so constrained, but less and less of us is in the physical world—less and less of each individual is in the physical world. We have shared consensual realities that have nothing to do with standard physical law. In reality, physical law crystallizes out in a very different way; there’s a phase change in many different dimensions of how you look at it.

When you look at the entities that we or our descendants will become, these vaster machine entities, do the questions of identity trouble you?

They don’t trouble me; they certainly fascinate me. From such simple questions as: If you have two indistinguishable copies, and neither knows which is the original—and in fact I believe there are ways in which you can make copies in which the question “Which is the original?” is meaningless—then what happens to property rights? I agree that the question is somewhat ameliorated by the fact that information is one of those properties that’s duplicable without loss, and so many things are solved at the same moment that they come about, but many things are not. If you have a patent, for example—we can argue patent law—but if you have a patent, who has it? You’ve just divided its value in half, whatever its value was. So there’s lots of

EXTROPY #13 (6:2) Third quarter 1994

30

questions from a simple case like that.

We have the ability to undergo Lamarckian evolution-driven intentionally-driven evolution for the first time. We have the ability to form children of many parents, and they needn’t be children, they may be bits of ourselves. We have the ability, perhaps, to partition and recombine our consciousness. In those senses, the concept of identity does become mutable. Does it become disturbing? People will do things they oughtn’t to do, because people always do things that they oughtn’t to do; people will do things that are disastrous, because they always do things that are disastrous; and people will do things that are wonderful and beautiful, because people always do things that are wonderful and beautiful, as well. Since I think we will do these things rather than get supplanted by our machines, among other reasons, the alternative seems worse, and the potential seems truly amazing.

Because I believe that human nature is fallen, and therefore corrupt, there will always be people who do bad things, but I also believe that there are very good people, and there are people who will fight them, and the universe is so constructed, I think, that the hand is tilted it slightly in favor of the good. Because I believe that to be the case, giving more power to all people, and particularly to individuals, tilts that balance better. It’s when power is centralized—if my ability to hurt you is amplified by a government, or is amplified by my restrictions that I place on you, then if I’m evil, then I’m likely to do more harm than if I have less power relative to you. That’s one of the many reasons that I’d like to see that these things take place—the movement to cyberspace, the increasing computing power in each individual’s desk—because I’ve always believed, and I continue to believe, that free minds are always going to be able to maneuver faster than governments can.

We’ve talked about uploading; do you see any reason why individual consciousnesses can’t be, first, duplicated, and second, backup copied?

The answer to the question is harder than I think a lot of people understand. The short answer is, no, I don’t see any reason why not, in both cases. I believe that a human consciousness can be uploaded. I think that it is very difficult to make sure that what you are making is a transfer rather than a copy. It’s very hard to do, and I go into that a lot in my Extropy paper¹⁰: not only why I think that’s hard, but also how to get around it. But hard is not impossible, and I think it is doable. The key, in my view, seems to be gradual replacement, and replacement at a low enough level that whatever makes up the consciousness is

not operating at that level. It’s basically the emulation idea.

So copying is easy, transfer is hard.

Yes, copying is easy, transfer’s hard, and if you want to do the transfer, you have to do the copying at a low enough level, and you must be destroying the original as you go, and the parts must always remain interconnected while you’re doing it, in some real sense, and so on.

It’s when power is centralized—if my ability to hurt you is amplified by a government…, then if I’m evil, then I’m likely to do more harm than if I have less power relative to you. That’s one of the many reasons that I’d like to see… the movement to cyberspace, the increasing computing power in each individual’s desk—because I’ve always believed, and I continue to believe, that free minds are always going to be able to maneuver faster than governments can.

That may not be required, and it certainly isn’t provable afterwards, but for my comfort level—if that’s what’s happening; if it’s me we’re talking about—that’s what I want to do.

It’s more aesthetically pleasing to have the dissolution of the original be gradual during the process, rather than, “Okay, that’s now you over in the machine, and now we’re going to take this biological body and get rid of it.”

Right, no thank you, no, un-uh, that’s not acceptable. This is the problem with backup copying. If it’s just a copy, that’s of almost no interest to me.

If there’s an original elsewhere—

That’s right; it’s not me. It may be like me in many many ways; it may be identical to me in most ways, but it isn’t me. However, I believe it is possible to use the same transference process, because, after all, the transference process says nothing about how long it takes; the speed of the process has nothing to do with it; it’s the method of the process that matters, and in that sense, you may be able to have a link between two things, the copy and the original, that is intimate enough, that when the so-called original ceases to exist, the copy continues and there’s no break.

The telepresence is merely switched from one—Yes, and again, here’s where it looks a lot like

the soul and the body, to me; and in fact you could operate it in some sort of way in which the copy is the real piece and this is the telepresence, as sort of a crude way of looking at it. That’s a possibility. But I believe that those are both possible things to do, though a good deal harder than the process of copying. After all, we know how to copy things much earlier than we know how to duplicate them. I mean, you can copy a chip, you go to Korea, or Singapore, or these days Thailand, with a

chip, and they strip it off layer by layer and produce masks and copy the chip, but they don’t know how to build them in the first place. Copying is always easier; you just do the same thing again. The problem is to make the first item, or doing the variant of it, or whatever, and there the transferences are harder because it’s a harder problem, but even it’s not the same as being able to create from whole cloth. That’s the hard part. They’re all hard, but that’s the hardest part.

We were talking earlier about the fact that people in our community seem to be seeking novelty in doing something different every couple of years.

Mm-hmm. With consistency to it, though. The things tend to tie in and tie together. Yeah, that’s true; we do tend to that. I’m not sure it’s exactly seeking novelty; I think it’s more seeking interesting things. Novelty is part of it, but depth of knowledge is also interesting, which is a novelty of all. I think we tend to play with things with far more effort than most people work at things, as a group.

Play at things with such an intensity as to be able to make a living from them.

That’s exactly right. So much the better, in my opinion.

It has never occurred to us that there is—Somebody made the comment the other day that your life is divided into two pieces: Your life is divided into what you do for a living, and what you do that interests you. That concept is utterly alien to most of the people that are my close friends—the idea that what you do for a living and what you do that’s interesting are separate. Why should they be separate? You spend eight hours a day doing your work; oughtn’t you to enjoy it? If you spend ten or fifteen hours a day doing your work, you really ought to enjoy it. But not because it’s your work, but because it’s your play. That doesn’t make it trivial; that just makes it fun. It doesn’t always make it fun, either, but it works out better that way, I think. That’s what’s almost unique about this group of people—the

31

EXTROPY #13 (6:2) Third quarter 1994

sense of play. Very serious play.

Gayle: Do you intend to grow up?

No. Why should I? One of the moments of understanding that comes upon everyone several times, if they’re very lucky, during their lives, came upon me watching Mary Martin playing Peter Pan, as a child, singing “I won’t grow up,” and I took this to heart. [chuckles] I remember the wonderful line from C. S. Lewis that says that the most mature adults are childlike and most childish adults are the ones that try to act the most mature. It’s absolutely true. My father worked 40-some-odd years for the same company. I am in almost no danger of ever doing that.

Why don’t you talk a bit about what you’re doing now, and how it relates to your long-term goals?

What I do for a living is start companies and try to bow out soon enough that they don’t get tiresome. Right now I’m working on what I would call, at a sort of meta-level, machine intelligence or algorithmic intelligence, although that’s a little bit too grandiose. Ten years ago I founded Calera, to do optical character recognition, and that was sort of the first area that I got into that was doing something that looked like artificial intelligence or machine intelligence. Basically, how do we teach a system to recognize a pattern that appears to us to be similar to, but not identical with, another pattern it has seen? How do we convey that information to a program; to be able to say: “Yes, you haven’t seen this before, but you’ve seen something like it. Categorize it the same. Probabilistically categorize it like other things you’ve seen that are like it.” That whole sort of meta-problem is one I’ve been working on now for quite a while. It has lots of different manifestations, but the character recognition is the more important one. I left Calera; I worked on the reading machines for the blind; I’ve used some of the same techniques that have been used for character recognition to make a text-to-speech system that works just about as well as DecTalk but has no rules at all.

I’ve developed the three paradigms of artificial intelligence; this is sort of my idea. There’s the expert system one, which I call the totalitarian paradigm, which is, “Here are the rules, don’t bother me with the details or the facts. I’m going to impose the rules from the top down, and we will live according to them.” This works every bit as well as central planning in an economy does. This is of course why everything we have works on an expert system, right? This is the wonderful comment which I think the head of AI at Yale said, that if computers worked as well as the expert systems branch of artificial intelligence, had

In a personal letter some weeks after this interview, Dave added:

”… Please don’t confuse the (admittedly unusual and extropiant) arguments I make in ‘Seven Paths to Immortality’ with my beliefs. I consider my belief system to be orthodox Christian and I call myself a fundamentalist. This does not mean that I subscribe to all the (peripheral) beliefs of all those who also call themselves fundamentalists — that would be impossible because they are not completely consistent within themselves. Some, for example, take the Genesis creation account as literal while others (both me and St. Augustine, for example) do not. What I mean by ‘fundamentalist’ is what most who call themselves that mean: I subscribe to the fundamental beliefs of the Christian religion. These include the belief that the universe is created by God, Who takes an active interest in its current existence; that Jesus of Nazareth is God incarnate; that He was crucified and rose from the dead, and that because of that act all people have open to them freedom from the evil in their lives and reconciliation with God. I believe the Bible is God’s word to mankind for the purpose of salvation; I do not believe it is either a biology textbook or a cookbook.”

made as much progress in the same thirty years, they’d still be made with vacuum tubes.

The second one is the sort of neural network approach, which I call the libertarian paradigm of artificial intelligence, which is, “I don’t know how this thing works, but if I throw enough things together and stir vigorously enough, something useful will come out of it.” This is the bottom-up approach. The problem with the totalitarian one is worse: that it is impossible to know all the rules operating in anything other than trivial systems. There may not be rules; there may be statistical, probabilistic things working. The world is not, as I’ve said before, categories; it’s imperfectly replicating patterns that replicate with some degree of probability. The problem with that is, unless you have in infinite number of rules, at which point it ceases to be tractable. If you have an infinite number of probabilistically applied rules, you haven’t got rules at all. The problem with the libertarian paradigm is the fact that, although the basic approach is right, there is global information; even though you don’t know all the rules, you do know some global information, and it fails to apply it.

So I came up with what used to be called, when I was a Republican, the Republican paradigm—and I’m currently looking for a better name for it—which is a combination of the two. An example of that would be, the obvious one, from character recognition: I have a series of images, I know that these images are likely to be e’s; they may be an a; and I don’t know, though I know probabilistically what they are; but I can go to a contextual system, in which I say, “This is a word that exists in English, and this is not,” and that’s an absolute, imposed rule. Therefore, I will choose the e over the a, regardless of their relative probabilities. That’s imposing a rule, but it’s a very sensible rule to

impose.

The systems that I build work that way, and they’ve been applied to speech synthesis, to coin grading, to character recognition, to all sorts of different things. I’m looking at doing it in speech recognition. Right now, I have a new company, called RAF Technology, that does pattern recognition to order. Right now I’m working with the Post Office, reading addresses on mail pieces for automatic routing and improving the recognition rate so it can be routed automatically by machine.

And, thinking increasingly about how machines learn. My longer-term goal in that is, of course, “If we can’t beat ‘em, join ‘em,” and how do we do that. We’re now making some weak steps in that direction.

NOTES:

¹See, for example, Merkle, Ralph. “The Technical Feasibility of Cryonics.” Available from the author, merkle@parc.xerox.com.

²See Merkle’s paper “Cryonics, Cryptanalysis, and Maximum Likelihood Estimation” in the forthcoming Proceedings of EXTRO 1: The First Extropy Institute Conference on Transhumanist Thought.

³Merkle “The Technical Feasibility of Cryonics.”

⁴Coined by Eric Vogelin. The history of the phrase is described in The Illuminati Papers by Robert Anton Wilson (London: Sphere, 1982).

⁵Barrow, John and Tipler, Frank. The Anthropic Cosmological Principle. Oxford: Oxford University Press, 1990.

⁶Teilhard de Chardin, Pierre. The Phenomenon of Man. New York: Harper & Row, 1959; The Divine Milieu. New York: Harper & Row, 1968; The Future of Man. New York: Harper & Row, 1974.

⁷Rothschild, Michael. Bionomics: The Inevitability of Capitalism. New York: Henry Holt, 1990.

⁸Toffler, Alvin. Future Shock. New York: Random House, 1970.

⁹Vinge, Vernor. Marooned in Realtime. New York: Baen Books, 1987.

¹⁰Ross, David J. “Persons, Programs, and Uploading Consciousness,” in Extropy #9 (vol.4, no.1, Summer 1992).

EXTROPY #13 (6:2) Third quarter 1994

32