I’m in Pinedale, Wyoming, visiting my daughter Isabel and doing some cross-country skiing among the aspen trees. They have great patterns like eyes on them. Today’s pictures are all from Pinedale, which is a nice example of an RR “real reality” far richer than any VR “virtual reality” we’re ever going to see.
In the last couple of weeks I twice noticed people online questioning my scientific accuracy when I claim in Postsingular that it’s bogus to talk about porting humanity into a complete virtual model of Earth. So today I want to explain some of the reasoning behind my claim.
Arguments for Vearth are sometimes start by talking about an imaginary substance wittily dubbed “computronium” by one of my favorite writers, Charles Stross. In Accelerando, he says computronium is “matter optimized at the atomic level to support computing.”
Although it’s a cute idea, I think computronium is a fundamentally spurious concept, an unnecessary detour. Matter, just as it is, carries out outlandishly complex chaotic quantum computations just by sitting around. Matter isn’t dumb. Every particle everywhere everywhen is computing at the maximum possible rate. I think we tend to very seriously undervalue quotidian reality.
[The ice fishing derby on Lake Fremont in Pinedale]
In an extreme vision—which is the one I disparage in my novel Postsingular—Earth is turned into a cloud of computronium which is supposedly going to compute a virtual Earth—a “Vearth”—even better than the one we started with.
This would be like filling in wetlands to make a multiplex theater showing nature movies, clear-cutting a rainforest to make a destination eco-resort, or killing an elephant to whittle its teeth into religious icons of an elephant god.
[Alley near the Teton Court Motel]
Ultrageek advocates of the Vearth scenario like to claim that nothing need be lost when Earth is pulped into computer chips. Supposedly the resulting computronium can run a VR (virtual reality) simulation that’s a perfect match for the old Earth.
As I’ll explain below, this is factually incorrect. Before getting into that, I might also ask why someone would passionately want to believe that we can be translated from flesh into bits? There’s something ascetic and life-hating about the notion. It’s a bit like a religious belief; one thinks of the old “work now, get rewarded in heaven” routine.
[Game heads on display in Ridley’s Grocery (formerly Faler’s)]
Anyway, let’s get back to my main point, which is that VR isn’t ever going to replace RR (real reality). We know that our present-day videogames and digital movies don’t fully match the richness of the real world. What’s not so well known is that computer science provides strong evicence that no feasible VR can ever match nature.
[Girl holding a fish as if it were a stuffed animal]
This is because there are no shortcuts for nature’s computations. Due to a property of the natural world that I call the “principle of natural unpredictability,” fully simulating a bunch of particles for a certain period of time requires a system using about the same number of particles for about the same length of time. Naturally occurring systems don’t allow for drastic shortcuts.
For details see The Lifebox, the Seashell and the Soul, or Stephen Wolfram’s revolutionary tome, A New Kind of Science—note that Wolfram prefers to use the phrase “computational irreducibility” instead of “natural unpredictability”.
[On a ridge above Fremont Lake on a climb led by Sherpa Iz-teng]
Natural unpredictability means that if you build a computer sim world that’s smaller than the physical world, the sim cuts corners and makes compromises, such as using bitmapped wood-grain and cartoon-style repeating backgrounds. Smallish sim worlds are doomed to be dippy Las Vegas/Disneyland/Second Life environments.
[A lo-resolution lodging near Pinedale]
But wait, answer the true-believer ultrageeks, if you do smash the whole planet into computronium, you have potentially as much memory and processing power as the intact planet possessed. It’s the same amount of mass, after all. So then we could make a fully realistic world-simulating Vearth with no compromises, right?
Wrong. Perhaps you can get the hardware in place, but there’s the vexing issue of software. Something important goes missing when you smash Earth into dust: you lose the information and the embodied software that was embedded in the world’s behaviors. An Earth-amount of matter with no high-level programs running on it is like a powerful new computer with no programs on the hard drive.
[Non-bit-mapped wood grain]
Ah, says the VR true believer, what if the nanomachines first copy all the patterns and behaviors embedded in Earth’s biosphere and geology? What if they copy the forms and processes in every blade of grass, in every bacterium, in every pebble, and so on?
But, come on, if you want to smoothly transform a blade of grass into some nanomachines simulating a blade of grass, then why bother pulverizing the blade of grass at all? After all, any object at all can be viewed as a quantum computation! The blade of grass already is an assemblage of nanomachines emulating a blade of grass. To the extent that you can realize an accurate VR world, the exercise becomes pointless.
Just as she is, Nature embodies superhuman intelligence. She’s not some piece of crap to tear apart and use up.
***
By the way, I have written a full-length essay expanding on some of these topics; it’s called “The Great Awakening,” and will appear in Asimov’s SF magazine in August, 2008, and in the anthology Year Million, edited by Damien Broderick, out from Atlas Books in August, 2008 as well.
For my answers to the many comments on this post, see my next blog post.
Much of the Earth’s mass is not currently doing useful simulation work — for each ton currently doing quantum computing to be a cloud, there are tons and tons of rocks whose quantum computing to be a rock isn’t really useful to humans… you could argue that in a perfect world you’d want to simulate all that stuff. But why?
You could still go mining — computer games already simulate detailed physics for things you are near and less detailed for further away things. So when you tunnel through the rocks, the detail is filled in. They just do not bother simulating the sound of trees falling in forests where there are no people around..
Likewise, the effects of billions of tons of rock which is currently inside the Earth can be approximated, and then you’ve ended up with billions of spare tons of computronium and still have the same human habitat..
And that’s ignoring the fact that a simulated Earth probably wouldn’t actually require all the features that the real Earth comes with. To pick a relevant recent example, we could omit the earthquake generating engine provided by the magma without anyone missing it very much…
There is another way in which you can compress the simulation, and again it’s one we already use.
We can do car crash simulations which obtain good results. The results may not be exactly the same as those obtained from crashing real cars, but then neither are the results of two car crashes — we just consider the “real” world to be the “correct” answer. Inside a simulation, “good enough” becomes “correct” — in the case of crash simulations we consider the result to be real enough to be able to design physical artifacts based on the data gathered from virtual ones
Crucially, the “world” computes its result in a tenth of second using several tons of metal. The “simulation” computes its result in (maybe) hours using a few grams of silicon.. We may not care if the simulation of the Earth runs at half speed on half the mass of the Earth. We’ll be inside it and won’t notice and the outside universe does not seem short on available time…
I’m not saying the idea of building a duplicate Earth in a computronium core is particularly sane, I’m just disputing your contention that it necessarily requires at least as much computronium mass as the Earth itself.
Katie, I talked about some of your points in my follow up post on this topic. The short answer is that the gain you’d get by not simulating the inside of the Earth is just a linear gain (by, like, a “mere” factor of billion), and what we are looking for in order to really crush a computation is an exponential gain by, like, a factor of ten to the thousandth power.
Re. crash simulations, they are very rough and approximate. Like you don’t get the subtle dents and tears in the metal, the cracks in the paint, the full spectrum of sound, and so on. “Real compared to what?”
First off, I’d like to jsut say that I’d likely qualify as a singulitarian. This is mostly because of subjects I write about That said, these are beutiful ’shots, I say this because I can’t get enough of experiencing things. However, I also subscribe to more Kantian concepts of reality; My reality isn’t yours and vice versa. I have 20/100 vision (terrible) but my vision doesn’t seem blurry to me. I can hear my name in conversation from across the house, but I don’t know what someone elses ‘bad’ hearing sonds like.
I’m trying to imply that ‘reality’ is a unique result of our own sensory intake. To say that it is ‘more real’ than what a computer can give is pretentious. That is because it implies we’d try to simulate the intangible concept of reality. We can only see an incredibly small band in the electromagnetic spectrum, our minds work in 3 dimensions (many leading theories clock in over a dozen) and we think blindingly slow.
Assuming reality is a dynamic, malleable thing (hypothetically, so I don’t get rebuked right off the bat because of conflicting views) it could expand into something truly posthuman. We could see the solar winds, feel the weather systems of an entire continent, memorize the exact makeup of a single blade of grass. We don’t wholly experience reality as is, simply by taking things for granted and the like. To say we have “Real Reality” right now seems to me very pretentious and shortsighted.
So anyway, no worries, not trying to wage war here, I just like debate. However, I distaste being typecast as an “ultrageek.” Buzz terms are definitely not a way to win over the opposite party. So, anyway, I know this is out of date, but I read it, so I’m making my two cents about it known aswell.
Aaron: Certainly I’m all for sensory amplification via cyborg devices.
My point is that we are continually carrying out reality-entangled quantum computations, far richer than anything that can be done on any digital machines in the near future—and that it’s a mistake to put too much mental energy into the digital machines around us both because they are somewhat feeble and because, viewed in the light of the history of technology, digital computation will most likely be utterly outmoded in a couple of hundred years. This isn’t “pretentious,” it’s a philosophical, scientific, and historical fact about the world we live in.
As far as I know, I didn’t call YOU an ultrageek, although if you want to take the word personally, feel free to adopt it as a banner. Having worked as a C++ programmer, I certainly have an ultrageek side myself.
Well, that’s comforting to know, I just finished a C++ class myself, still a newcomer, though. Anyway, I suppose the vote’s still out on what our future will be (and always will be, likely, which is the reason why all SF writers, myself included, are only liars with varying degrees of deftness.)
Other than that, I’ve seen it raised earlier that we may opt for all sorts of other things then a canned reality. What we could achieve in a world where we can write the rules has fewer limits than you may imply, even if the limit on definition and clarity is a major one.
Though I do acknowledge true reality is likely going to persist and be desireable. We may, however, see a similar situation in the game vs. reality relationship many people have these days.
Anyway, I’m going to wrap up by saying I think Stross isn’t a bad writer. I was a big fan of Atrocity Archives and somewhat of Accelerando. I suppose we can poke holes in his work or anyone else’s but we might as well be barking our way up fog. SF is more than anything about conjecture.
So I suppose I’m more interested in new conjecture and not as much in defacing Stross’ work or anyone else’s. The best way to deal with a concept that seems hackneyed is to dilute it with new ones. Let’s show the fantasy boysd we still have some steam!
Charles Stross is one of my favorite writers, and it was his magnificent ACCELERANDO that set me down the path of writing POSTSINGULAR and HYLOZOIC. The fact that Charlie’s work goaded me into criticizing some of the things that his CHARACTERS do is by no means a criticism of the AUTHOR’s thought and his writing.
I second that emotion! Stross is not only the autopope, but the autoprophet. IMHO he should take longer writing the books, but obviously he’s busting to write the next one and the next one, which is also good. Rudy has better characters, but then Stross is young. Relatively young. Younger than me, anyway. Where does he get off being so damn good anyway, the whippersnapper?
I’m new to this topic and Virtual Reality in general, so please bear with me. I haven’t read the author’s books or all of this blog but in haste I’ll just jump to my inquiries:
Could it not be possible to create virtual subatomic particles that function at least remotely similarly to real life subatomic particles? Could not the Hadron Collider give us the necessary clues to finally understanding the mysteries of the atom enough to transfer the physical atom into a virtual earth atom? If so, our software engineers could recreate earth and all or a portion of its geology and evolution from scratch by tinckering with virtual elements. Now that we have a genome handy and can map the entire human DNA, could we not produce a virtual baby that accurately resembles a real life baby and grows and lives according to its virtual genetic programming acting according to its virtual environment?
BTW, I recently read an article speaking about the possibility of building supercomputers that are built out into the sea and use seawater to cool them and to generate the electricity to power them. Pretty fascinating stuff!
Like I said, I am completely new to VR. Any constructive feedback would be appreciated.
A very good blog post if I may say, I do understand you are impressed by the scenery and landscape, textures of the real world with all their details. Yes at the moment the virtual world doesn’t allow such detail but as technology evolves things will change on this aspect and will start to look more realistic which is the goal of many virtual platforms I think. We can already see how the first “mirror worlds” start making their entry just with the goal of reflect the real world as much as possible as they can in the virtual world. Your impressive snow landscape will become possible in a few years from now inside a simulation enviroment.
Rudy - of course I agree with “But, come on, if you want to smoothly transform a blade of grass into some nanomachines simulating a blade of grass, then why bother pulverizing the blade of grass at all?”. But I would certainly consider transforming the blade of grass into some nanomachines simulating something else, like maybe a virtual heaven where I can have a few thousands subjective years of fun with the proverbial 70 virgins, or a few thousands subjective years of fun doing something else.
Re “Just as she is, Nature embodies superhuman intelligence. She’s not some piece of crap to tear apart and use up.”
Well, nature is certainly not some piece of crap to tear apart and use up, but it is not either something to mystically revere. I would never tear apart and use up some parts of nature like my friends or my doggy, but have no objections to doing so with other parts or nature. Building a house, and also cooking a meal now that I think of it, are examples of tearing apart some part of nature and using it up.