The Jesus Machine


Grey has just put the last of the pieces together in his head when he hears the first scream from outside the shaft. The timing is uncanny. He has spent a week and a half taking photographs and drawing schematics and dispatching hesitant mana pings into likely-looking receptors in the surface of the artifact. He has observed the series of inexplicable and then unnerving and then mind-wrenching real-world effects of powering the thing up. He has spent that time patiently avoiding leaping to the conclusion. He has tested every step and ruled out every alternative. And he has, after ten-and-a-half days, for one moment, permitted himself to entertain the possibility that what he's looking at is what he feared at first glance he was looking at, and at that instant he hears what sounds like the world above being Cleaned, and it's almost a relief.

Magic is a science. It's a science as advanced as any other, as quantum field theory or general relativity, and there have been outlandish claims and vague or self-fulfilling prophecies and forged or questionable evidence, but there is no concrete evidence, none, that anybody knew anything about it until Suravaram Vidyasagar cast uum for the first time in 1972. There were no ancient astronauts. There were no real witches at Salem. Jesus wasn't a mage-- nothing that he or Muhammed or the Buddha are ever claimed to have done matches up with modern magic. Artifacts from the pre-magic era are one hundred percent non-magical; magical artifacts "from that era" are one hundred percent forged. And the reason for this, which most of the world accepts, is that magic wasn't in reach then. Nobody had the knowledge of physics that was necessary, because the physics that was necessary hadn't been discovered and locked down yet. It was luck, arguably dumb: magic could have been discovered decades earlier or, equally, could have remained unknown long past the present day.

But the artifact that Gareth Grey's high-altitude thaumomagnetic survey found in the far east of the DRC, which it's taken his expedition a month to reach and dig down to, could turn all of that on its ear. It's the rabbit in the Cretaceous, the Casio wristwatch in the coal seam. It's a magic ring as big as a Stonehenge megalith, and it's embedded in rock fifty million years earlier than it could ever exist. There have been geological explanations suggested for its position, but they aren't convincing. And there's certainly no theory in the whole of human knowledge for who could have made it. It can't exist. It's beyond inexplicable. It represents a discontinuity in Grey's rational universe, a false statement which, thanks to the principle of explosion, logically implies the destruction of all reasoned thought. All statements are now false and true.

Aliens. Time travel. A practical joke?

But the thing is huge. Magic rings in modern engineering are of course as big as they need to be: at CERN Grey saw one wide enough to fly a light aircraft through the middle. But no magic ring is this fat and surely no practical magic ring weighs this much. Based on the geophysical scan (most of the object is still buried) it's thirteen metres wide and two thick, with the cylindrical hole through the middle just a metre and a half wide. If the thing's solid metal, it could weigh anything up to four thousand tonnes - assuming the metal can be identified. It could be moved. It would take an army, but it could be moved. An army isn't out of the question. An army could use it. More efficiently than anybody, in fact, Grey thinks.

The metal is shiny and silvery which rules out, oh, copper and osmium and precious little else. Except for a small, faint etched plus sign, it's smooth all over the exterior and the flat top face (again, as far as they've excavated). About forty percent of the cylindrical hole's inner surface has been uncovered. Into the inner rim, big, bold sigils are etched and there are more, smaller sigils etched into surface of those sigils, and a third iteration etched into those in turn. The writing is so dense and sharp as to be incoherent; regardless of how well-preserved the metalwork seems to be, uncovering enough of it to activate the ring has been painstaking. Grey feels dwarfed by its complexity. Way back in the Stone Age of computing, when the available resources meant that programs had to be minuscule and neatly folded and unreadably, irreducibly complex just to fit, let alone run, this is the kind of code that Grey saw and wrote. These sigils are at the same density and at the same degree of interconnectedness. Inadvisably tightly coupled and organic.

That's the word Grey is afraid of, "organic". If the artifact was grown and machined by some thrashing, senseless neural net, this is what you'd see. If the problem the thing was designed to solve was so complex that the solution wouldn't be recognisably designed at all, this is what you'd see. When he powers it up, Grey feels like an ant trying to understand the console of the starship Enterprise. And when he powers it up, it still, after N years, works.

The first time he put a jolt through it, a few insect bites disappeared from his arm. It was a whole day before he noticed.

There are more screams. The excavated shaft is a steep diagonal corridor cut into the side of a thickly rainforested mountain. The first part was easy, a thin upper layer of poor soil (as red as Mars and about as fertile), dry leaves and ants. After that it was a dangerously dusty, noisy, mechanical job using heavy machinery which was badly-maintained to start with and which had become visibly even worse for wear on the journey into the Congolese interior. The shaft is about fifteen metres long, with plenty of head room, but barely wide enough for one person to slide past another. The chamber at the bottom where the ring is being excavated is a little bigger, with room for a proper light and a camping chair. Given another month without interruptions, Grey would have had the whole roof pulled off, exposing the ring to fresh air for the first time in - depending on your hypothesis - tens, hundreds, thousands or millions of years. But Grey made them cut to the core directly. He wanted to see it. And that means he's trapped alone at the bottom of a hellishly hot, lung-scratchingly dusty hole in the ground, with no weapons beside an archaeologist's brush, a face mask and a mattock.

He blames himself for calling attention. He could have dropped the matter, let the odd readings stay odd and unseen forever and continued charting the grander topology of low-grade natural background magic on the African continent. But he wanted to see something that nobody else had ever seen before. He wanted to find buried treasure.

Grey stares up the shaft for a long moment, squinting at the light, waiting for somebody to appear at the mouth. They're here for the ring. There's no other conceivable reason for anybody to be this far into the rainforest. Which means it's as valuable as he fears.

There are two more shouts and the sound of running and the sound of bodies falling. There's no gunfire, though. He looks down at the walkie-talkie in the dust and considers trying to contact somebody up above for help, perhaps to organise a distraction. But then he counts up the screams he's heard-- there can only be at most one or two of the expedition left, and they'll be running for their lives. The noise from a crackling walkie-talkie could give them away. Could he wait until dark? If he switched off the lamp down here, could he jump whoever came down to investigate? He hefts the mattock in one hand. It's balanced all wrong for use as a weapon; he'd knock holes in his own limbs by accident. Maybe he could throw the mattock out of the shaft, so that it would land elsewhere, distracting whoever is out there for long enough for him to make a run for the Jeep. How many people are out there, anyway? Surely no more than four or five. Otherwise the whole camp would have been swamped in seconds, not minutes.

Possibilities tumble. Having convinced himself that there's no way out of the situation alive, Grey wonders if he can still win by bringing the shaft down on his own head, or by destroying the artifact. But he oversaw the beaming and propping too carefully for the former to be an option, and as for the latter... it's a four-thousand-tonne doughnut of metal. As delicate as its engravings are, he'd need strong acid to do anything worse than a slight dent.

And having come to the end of that thread, having realised that he's not just dead but beaten, he remembers how far he is from home.

He exhales hard and clenches his teeth. "Damn."

"Penny for your thoughts, Dr. Grey," says a voice. At the mouth of the shaft, Grey sees the silhouette of a bald man in a dark, loose-cut suit. The man has no visible weapon. He holds onto the roof of the entrance for support, for a casual effect.

"How many of my people have you killed?" Grey demands.

"The four drivers. The two engineers. The guide and his brother. The blonde geologist and the dark-haired geologist she liked. And the fellow who carried your amulets. That's everybody." The faceless man states it as fact.

"That's everybody," agrees Grey. He hurls the mattock overarm, as hard as he can. The trajectory is long and flat enough to avoid the ceiling and walls of the shaft. Then there's a smash cut. Subjective time skips.


--flat on his back, waking up. His head is as clear as a bell. He's an early bird, but as far back as he can remember he's never woken up so cleanly, with so much clarity. Certainly not after being knocked unconscious. He detects no head injury. Gas? He sits up in the mud and squints at the light. He's still in the shaft, lying with his head and upper body in the partially excavated doughnut hole, and it's still daylight outside.

The mud, he discovers when he brushes it from his hands and hair, isn't wet dirt. It's gore. Grey recognises brain matter and chunks of skull. The other half of the doughnut hole, which is still plugged up with stone, is splattered with an entire head full of blood.

Well. That answers some questions.

Earlier, tentative experimentation had shown that the machine could resolve cuts and scars and, in one of the engineers, a years-old mild limp of uncertain origin. It was also shown to restore eyesight, which Grey found moderately troublesome, since his only sunglasses were prescription. But that was the limit. There was no chance that he was going to ask somebody, in a spirit of scientific inquiry, to deliberately break a bone and see if the machine could fix it. That would have been madness. He stands by the decision, even now, standing in the splatter pattern of his own detonated head.

So it can fix brains, from a standing start. Presumably he had to be nearby for it to record a pattern to work from. It would be impossible, surely, to put a human mind back together properly without some kind of "known-good" template to work from...

Grey cackles. "Impossible".

He thinks back to the smash cut and wonders how he'd detect other inconsistencies in his own memories under these circumstances. Did the machine restore his mind fully, or partially? There are other questions still open: can it do anything about mental illness? What about extreme age? What about non-humans?

There's no bullet or hole. He doesn't remember a gunshot. Not even a silenced round. Not even a weapon being raised.

He climbs to the mouth of the shaft and pokes his head out, blinking. The sunlight is so bright that it feels as if it has physical weight. Below the shaft mouth is the wide, mucky orange path that leads down to the camp site, a wide clearing littered with tents, parked vehicles and diesel generators. From where he's standing, Grey counts seven corpses. The bullet wounds that he can see look like clean chest shots. But he can't see the man. He can't see anybody living.

Quietly and with extreme caution, he sidles down towards the camp, through the trees rather than down the obvious path. If he can reach his tent, he can unlock the small metal trunk and reassemble his rifle. Then he can use it to keep his nerves subdued while he works out a plan.

"I said, what do you think?" asks the same voice, this time from behind. Grey sags, the effort of stealth wasted. He turns. The bald man is emerging from the same excavated shaft, which should be empty, but Grey is too tired to absorb this extra oddity. The man's tall and very young. Closer to a boy, in fact. Early twenties at the latest. He's young and he stands casually, with his hands in his pockets. His linen suit is loose and its jacket is unbuttoned: Grey sees no gun. He looks around and still sees no accomplices.

"You can have it," says Grey. "I don't care what you do with it. Just give me enough time to use it to bring my people back."

The youth smiles faintly and shakes his head.

Grey conceals his anger. He decides to play the boy's game, to buy time. "It's obviously a doctor. I suspected from the Red Cross symbol on its hull. It's the mechanical realisation of the abstract concept: a machine which makes people better. The most complicated medical device ever created, a million times more complicated than any medical device I've ever seen and a thousand times more complicated than the human body it's designed to fix. And... it can't exist. I can't even conceive of magic so advanced. No human can, no matter the IQ. It can't exist. I'm a mage and I know magic isn't like this."

"But what do you think?"

"What do I think about what?"

"What do you think happens next?"

"Obviously you and whoever else is with you are going to kill me and take the machine."

"What if I didn't do that?"

Grey blinks. "...We would need to get it to a laboratory," he says. "Because one isn't enough. If we put the thing at the most accessible point on Earth and formed a human processing system ten times as complicated as Mecca, and forced people through the machine one at a time, one every two seconds, for the rest of time, it wouldn't be enough. It wouldn't register statistically. It wouldn't make a dent in any of the rates. Which means we need to make more. Millions more. This is... it's Outside Context Medicine."

"And then what would happen?"

Grey stares into a distant possible future. "Medicine as we know it would-- it would become magic. Everything we know about medicine would be revolutionised. We'd write libraries about what the machine does to people, the difference between broken and fixed people. And then we would throw away those libraries because we'd never need them again because everybody would live to a hundred and twenty without trying. If you lived inside a machine you could live for eternity. And if there's a way that the machine can reverse telomere shortening, then everybody on Earth could live forever just with periodic visits. You could have eternal youth. For everybody."

"And then what?"

"And then?" Grey concentrates. "There would-- there would be no Malthusian catastrophe. There wouldn't need to be. Because you don't need food and water anymore. You visit the machine. Malnourished? Visit the machine. You come out the other side fed and watered. Food becomes a luxury item. The capacity of the planet becomes a function of physical space. Maybe if the technology can be adapted, the whole of the world could be pervaded with this restorative power. You wouldn't need to eat, or drink. Or even breathe. You wouldn't need air anymore. You'd-- You'd have to rediscover death."

The bald youth reflects for a long moment, and then asks, "A likely story, do you think?"

Grey smiles darkly. "Of course not. None of it."

The youth says, "Here's what we think: A major medical research company pays for the rights to study, own and operate the machine. At great length and expense, they duplicate it. They want a return on their investment. They make eight machines, embed them in purpose-built medical establishments in world cities and sell the best medical care that is theoretically possible to only those able to afford millions of U.S. dollars per visit. When it becomes clear what the organisation is sitting on, it becomes the target of heavyweight litigation, industrial espionage and eventually overt physical attacks. A man is denied access due to perceived war crimes; another man, also a perceived war criminal, is admitted. Unrelated tensions boil over at the same time, amplifying the situation. A full European War erupts.

"But in fact, what's more likely is that the machine proves unduplicable. Its location on neutral territory in, for example, the Hague, the Netherlands, becomes the nucleus of a community of ill and dying pilgrims desperately queueing for one-time exposure to a machine which cannot physically process one in a hundred of the patients who need its treatment. A second city is founded on the streets of the first. First crime consumes both cities, then disease, then violence. In the final series of riots, the facility is stormed and the machine captured by a dozen different groups in a single week. Eventually the Dutch military end the conflict by permanently disabling the machine.

"But even that's an outside chance because, in the first place, you're never likely to get it out of the DRC unchallenged. Eight African nations including the Democratic Republic of the Congo itself become aware of the machine's existence and initiate a decades-long, interminable land war to claim it. Western nations become involved and the war in turn claims millions of lives and ends with the tactical atomic bombing by the United States of the installation where the machine is being held. Even though the machine was believed to have been rendered unrecoverably inoperable years earlier, the bombing is regarded as the greatest humanitarian catastrophe of all time.

"Except that that might not happen either. Let's say the U.S. wins the war. They capture the machine and take to the bunker underneath the White House, where only the President, his family and his cabinet are permitted access to it. Medical technology is deliberately stalled and never reaches the pinnacle it should.

"And yet, for anybody to leave the machine unexploited is implausible. We spin more numbers and simulations and we see the machine being reverse-engineered, and the principles it applies being adapted for purposes other than the immediate, perfect restoration of living and dead humans. Mr Grey, you've seen how easy it is to heal. Can you imagine how easy it'll become to kill?

"The truth will inevitably be somewhere in the middle of all of these possibilities, but I'm sure you understand the common theme. Death surrounds this machine, like a curse. Death and leverage. The mother of all MacGuffins."

Grey imagines how easy it would become to kill. You wouldn't need a gun anymore. You could create a bullet and give it motion. You could simply "correct" a living human body to a living body with a hole in it.

"And you see," the youth concludes, "that you have to let us take it and put it somewhere safer."

"Take it back, you mean," Grey says.


"It must have been in a hell of an accident to wind up inside a mountain," Grey says. "How did you lose it?"

The young man shrugs.

"But how did you know we'd found it? I chose my crew for loyalty. Until yesterday, I didn't utter a hint to them about what I thought the thing actually was. And I know none of them satphoned home."


"Then who are you?"

"I can't tell you."

"But you're going to kill me."

In response, the youth nods towards the shaft mouth. "It's still a risk. You understand."

Grey does.

The youth continues, "All I can say is that we're the ones who ran the numbers. Of course, nobody can accurately predict the future, but after ten thousand high-fidelity simulations of the same events, some outcomes turn out to be more probable than others, and then when we go through our courses of action, this is the firmest recommendation."

Grey's eyes widen. He puts his hands out, mind racing, heart racing. "Wait. No, wait!"

The youth pulls his right hand out of his pocket and says "One more, please," seemingly to nobody before pointing at Grey with his first two fingers.

"How good are your simulations? What kind of fidelity? Was I a component? Could you be?" Grey delivers all four questions in about three seconds.

There's a pause. The boy doesn't lower his hand, but Grey's got his attention.

"How would the simulation begin?" Grey continues. "If you were simulating this course of action, how would it begin? It would begin with the decision being made, right?"

A longer pause.

Grey says, "It would begin exactly like this. No matter the option, no matter the outcome. The course of action to be tested is X. So they create a simulation in which the course of action selected was X. In the simulation, you are given the order. In the simulation, a simulation of you carries X out. And as it plays out, the simulators observe the results. They collect the results from X and Y and Z and stack them all together. And then pick the best one and they go out into reality and do it once, for real.

"Think about it. You, you, know all of the possible outcomes if nobody does anything, if nobody interferes," Grey says. "Because those runs were run. But you don't know what the other options of interference are. You don't know about the other hypotheses. About Y and Z and the rest. That's the only way that you could prove that this isn't a hypothetical, because that information is the only information that's guaranteed not to be available inside a hypothetical. And you don't have it. And this can't be the right solution because it doesn't make sense. Killing to prevent more killing? Killing to prevent a medical revolution that could save literally every life? This has to be one of the faulty hypotheses. And that means that you and I don't exist. This isn't happening. So it doesn't matter if you don't kill me."

"Then it doesn't matter if I do--"

"But the point is that X doesn't have to happen here. Make this the one where you gained unexpected self-awareness and disobeyed orders. Put your... absence of gun away. Maybe that's what they really want to see happen. They order you to do X but they want me to overrule you with Y so they can see how Y plays out. They want to see what happens if... you let me try to save everyone." Grey locks eye contact with the youth. Grey tries to make himself believe that he sees a flicker of doubt in there.

There is the longest pause.

"No," says the boy. "This is real."

He shoots Grey. No gun. No bullet. He just opens a hole in Grey's heart.


Next: Space Magic

Back to Ra
Back to Fiction
Back to Things Of Interest

Discussion (37)

2012-07-14 14:54:33 by qntm:

Interesting fact: I am less confident about this title than any I've ever used.

This story puts forward my "hypothesis hypothesis", which states that our universe is merely a low-fidelity simulation of events which *could* occur in some higher, more accurate, "real" universe. The simulation in which we live is one of many, being run in order to locate the best possible outcome for the "real" universe. This means that, by definition, events transpiring in our universe will never be for the best, whereas events in the real universe do always turn out perfectly (good is rewarded, evil is punished, et cetera).

The metal that the machine is made of is rhenium.

2012-07-14 16:26:34 by YarKramer:

The title certainly got my attention.

Hmm. I wonder how this fits into the rest of the story. Presumably these guys are at least related to the same guys who may or may not put Laura through her paces until she fails to go up against a robot.

One wonders why these guys didn't simply go back in time to before these people discovered it in the first place, thus bypassing the need to kill anyone. (This wouldn't make for an interesting story, but still.) Maybe they're from a future in which Laura *didn't* learn to stop killing people, and this "philosophy" spread, so the society at large tends to shoot first and ask questions later?

2012-07-14 16:56:49 by Baughn:

Solomonoff Induction, isn't it?

You're hardly the first one to try this.

2012-07-14 18:21:26 by qntm:

I didn't know what Solomonoff induction was until you mentioned it, but if I'm reading this correctly, it's one of the coolest things ever.

Also, no.

2012-07-14 20:13:56 by eneekmot:

Scary, this is the first time we've seen combat magic that isn't non-lethal (referring to the time Laura was mugged). I wonder when we're going to get to the science of weapons of mass destruction. From what we've seen of magic so far conjuring a few milligrams of antimatter would likely be the closest thing to a bomb.

2012-07-14 22:28:52 by qntm:

Oh yes! Also: this chapter is largely a commentary on my own indecision as to how the chapter should go down. All the events that the unnamed bald fellow describes are outcomes that I toyed with presenting literally as the outcome of Grey's discovery. I couldn't decide which direction to go in, so I decided to use that and snuff the story at its root before it became too complicated.

2012-07-15 00:29:50 by DanielLC:

"... because that information is the only information that's guaranteed not to be available inside a hypothetical"

No. It's the only information that's guaranteed not to be /correct/ inside a hypothetical. They'd just give him made up numbers, saying that this one is the best option.

"Mr Grey, you've seen how easy it is to heal. Can you imagine how easy it'll become to kill?"

Being able to heal someone with a four thousand tonne artifact is pretty impressive. Being able to kill someone with one is not. Am I missing something?

If the simulations are good enough to create sapient beings, the artifact is worthless. Even if you use it as a substitute for eating, it would take orders of magnitude more resources to support a flesh-and-blood human than a simulation.

2012-07-15 05:42:22 by Moti:

"There was no chance that he was going to ask somebody, in a spirit of scientific inquiry, to deliberately break a bone and see if the machine could fix it."

Nice sort-of-callback to "The artifact was completely impenetrable...", where the scientists of course did go, ahem, much further than that.

2012-07-15 05:53:02 by LNOr:

I seem to be missing a major bit of the plot here. Grey throws his pickaxe at the guy in the doorway, apparently gets shot in the face by the invisible gun, then... what? Then the guy puts him in the healing machine, helpfully does the magic to activate it, and then lets him wake up? Just so the guy can have a conversation with the person he's planning to kill anyway? I am confused.

2012-07-15 09:48:18 by Thomas:

He doesn't need to put Grey into the machine. He's already standing in it, and the machine works on its own. The guy has to wait for Grey to get out of the shaft he's in before he's able to kill Grey.

Which brings up the blackly comedic outtake potential of Grey being gradually drowned in a liquid soup made from a couple of thousand iterations of his own head.

2012-07-15 10:33:01 by qntm:

"Being able to heal someone with a four thousand tonne artifact is pretty impressive. Being able to kill someone with one is not. Am I missing something?"

The technology used to heal people could very easily be adapted/damaged/corrupted so that it kills people or otherwise makes them ill or broken. The corrupted technology would vastly simpler, just as easy to use, much more portable, require less magic and still have the same capacity for the number of people it could be used on. Essentially, the point being made is that the thing is a new weapon of mass destruction in disguise.

2012-07-15 10:36:06 by qntm:

LNOr: no, you seem to have it exactly right. The first thing that the youth says is "Penny for your thoughts." He does indeed want a conversation with the fellow he's about to kill before he kills him. As for why, I leave that to your interpretation.

2012-07-15 11:23:17 by bysjonas:

Just like Oul's Egg, this magical machine again reminds me to The Fifth Gift, a science fiction short story by localroger, read at "".

2012-07-15 15:47:23 by Aegeus:

"...and still have the same capacity for the number of people it could be used on. Essentially, the point being made is that the thing is a new weapon of mass destruction in disguise."

The whole problem with the machine is it only works on one person at a time. And the boy's weapon appears to have the same limitation (or he could have just torn apart the whole camp in an instant). I still don't see how this is better than a gun. Aside from being really cool, anyway.

2012-07-15 18:28:42 by Daniel:

"And the boy's weapon appears to have the same limitation (or he could have just torn apart the whole camp in an instant). I still don't see how this is better than a gun. Aside from being really cool, anyway."

Without having the mechanics of it demonstrated more closely, it's impossible to say. However, my speculation would be that guns are noisy, ammunition-limited, and fire a physical projectile which can be defeated by any number of techniques (though none of those would be particularly practical in the situation the characters were in).

That said, I still don't get the implication that this miracle of medicine could reasonably become a portable mass-scale superweapon. If it is reprogramable, then sure, you could theoretically reprogram it such that its stored pattern for "healthy human" is missing several critical components, but this leads to two fairly irreconcilable issues:
First, for it to be a reasonable weapon assumes that we have the capability to 1) reprogram it, and 2) either miniaturize and duplicate it or expand its area of effect. The duplicate/miniaturize option leads to people walking around with what are effectively instant-kill melee weapons, whereas the expanded area option leads to more or less a new type of WMD. The capability to make any of these changes seems to be strongly hinted against by the youth's simulations, at which point these are unlikely outcomes at best.
Second, if the first case were to be true and the device could be weaponized, logical consistency would dictate that it could also be anti-weaponized. If it can be modified to be a city-scale instant-kill weapon, then it can also be modified to be a city-scale instant-resurrection device. Similarly, if you have simply miniaturized it, then for every soldier/criminal/whatever running around with instant-kill devices, there's going to be ten more people running around with instant-fix devices setting their antics right in a rather macabre game of whack-a-mole.

I just don't see room in the simulations for "it's weapon-capable and there's no countermeasure that can be taken"

Still, it's a good story. Thanks, Sam!

2012-07-16 01:22:45 by jalapenodude:

Notes on the "hypothesis hypothesis" [Sorry, this is going to be lengthy, since I'm using it as a an excuse to write about cosmology, which happens to be the subject I'm getting a PhD in]...

In order for the H.H. to be true, it seems to me that we need three conditions to be correct:

1) "Almost all" conscious observers will find themselves in simulated universes.
2) "Almost all" simulated universes will be created for a "purpose" (i.e. with the goal of maximizing a utility function of some kind).
3) "Almost all" of the "purposes" for which simulated universes are created will entail creating a "perfect" universe.

Let's take each of these in turn.

First, why have I put "almost all" in quotes? Well, "almost all" is a probabilistic notion, so #1 is really equivalent to the statement that (# of conscious observers in simulated universes)/(total # of conscious observers) is approximately 1. [You might argue that we should say "universes with a conscious observer" rather than "conscious observers; this won't change anything I'm about to say, provided we really mean "observable universe" when we say "universe."]

Why is this a problem? Well, in general we expect both the numerator and denominator in that fraction to be infinite. In order for the fraction to be sensibly defined, then, we need a "measure", i.e. some procedure for taking a limit. [In cosmology this is known as the "measure problem."]

Here's one possible way we could imagine doing that. First, take the "ultimate ancestor" of our observable universe. This means the entity/computer responsible for doing the simulation of *our* observable universe. (If we're not a simulation, our ultimate ancestor is our own observable universe.) I say *ultimate* because we can imagine the computer that's simulating us itself being simulated, and so on.

[An ultimate ancestor is therefore always defined only if this chain ultimately terminates--this is equivalent to saying that the (true) universe had a beginning. This is certainly true given general relativity--see e.g. the Hawking-Penrose singularity theorems--where the universe can be future-infinite but not past-infinite. If you want to entertain the possibility of universes where GR isn't true, you'll need a different measure.]

[After this point, we're doing standard cosmology--the only changes is that we're using the observable universe of our "ultimate ancestor" rather than our own observable universe.]

Next, construct the set of events (i.e. points in spacetime) whose future lightcones encompass our the observable universe of our ultimate ancestor--i.e. the set of points for which the entire observable universe of the computer running us is in the *future*. Choose the latest such point [or one of them]--i.e. a point whose future lightcone doesn't enclose any of the other points in this set. [This step isn't really that important, but it gets us a concrete point in spacetime.] Choose a reference frame [it doesn't matter which one, since we've ensured that the entire observable universe of our ultimate ancestor will always be in the future] and take a [comoving] sphere of radius R around this point. Now count the number of conscious observers who come into existence inside the future lightcone of this sphere within a time T. [This is called the "lightcone measure."] We can now compute the desired probability sensibly--count the number of observers inside simulations and divide by the total number of observers. Take T to infinity and then R to infinity, and we get the probability we want.

So, what's the problem? Well, imagine doing this for our own universe. One of the leading theories of the early universe is "inflation." [Which you should go read about on Wikipedia if you haven't before.] If inflation is correct then the universe went through a period when it expanded exponentially--eventually quantum fluctuations caused a non-inflating bubble to form within this exponentially inflating background, which reheated [it had lots of extra energy because it was no longer expanding nearly as quickly], causing a big bang and giving rise to [a volume larger than] our current observable universe.

Crucially, the point we identified *must* be earlier than the point where the non-inflating bubble that hosts our universe formed. [This is one of the reasons inflation was proposed--without it the observable universe doesn't lie within a single future lightcone, and we can't explain why points on the opposite side of the observable universe look roughly the same--the "horizon problem."] So its future lightcone contains a large amount of exponentially inflating space, in addition to our own observable universe. Now--this is the key point--if there's a constant rate of bubble nucleation, there will be *exponentially more* bubbles will form at a later time than at any earlier time--because there will now be exponentially more space for them to form in!

What does this mean for our population of observers? Ignoring ones from simulations, this means the *vast majority* of them will have formed as close to the cutoff time T as possible, because exponentially more bubbles form later than earlier.

So, how long does it take to form a conscious observer after a big bang? The standard way would seem to take several billion years--we need multiple generations of star formation to get enough heavy elements for planets to form, complex chemistry to get started, etc. But there's an easier way: we could just have a brain (microchip, etc.) coalesce out of random thermal fluctuations that has "memories" of a body, a sensory apparatus, etc. [See the wikipedia article on "Boltzmann brains."] This is obviously going to be spectacularly unlikely--but if we take R to infinity, it *will* happen occasionally, and critically it can happen immediately after the big bang. And it turns out that in the lightcone measure the exponential gain you get from waiting until *just* before the cutoff far outweighs the rarity of the occurrence. So Boltzmann brains will be "almost all" of the conscious observers--i.e. if you're a conscious observer, than with probability one in this measure you're a Boltzmann brain!

Does including the possibility of simulated universes change this? Only if each simulating entity can simulate an *exponentially large* number of simulated universes almost immediately, which seems highly implausible to me.

So at least in some measures requirement 1) fails. However, let's press on anyway: there are *some* measures in which Boltzmann brains aren't preferred, although they tend to be much more contrived than the one I've presented.

Now for condition 2). This seems to be a *sociological* assumption--people who have the power to create simulated universes will use them overwhelmingly for a "purpose," i.e. to *solve problems*, rather than just to explore to possible parameter space of universes. I'm not sure whether that's the case. Here's one argument in the other direction: people who are trying to solve problems will tend to *constrain* their simulated universes to a much larger extent--they'd just simulate the Earth starting from the recent past instead of an entire observable universe starting from the big bang. That would seem to suggest that unconstrained/"unpurposeful" universes will have many more conscious observers in them--so you'd need that to be outweighed by a *much* higher frequency of "purposeful" ones.

[Alternatively, taking the reasoning above re inflationary universes into account, it's a question of what the *first* simulated universe will be like, since the vast majority of observers will have "ultimate ancestors" who formed as soon after the big bang as possible.]

Finally, condition 3). This has two parts--first, it requires that your (Sam's) definition of "perfect" is sufficiently close to the definition of the typical entity with the power to create simulated universes--e.g. that they have roughly the same notions of "good" and "evil" as you do. Second, it requires that the "purpose" of the simulated universes is to create the perfect universe for everyone, rather than e.g. creating the perfect universe for just the individual who controls the simulations, or maximizing the profit of the corporation that controls them. Frankly, this is the part I'm most skeptical of, despite all my arguments about condition 1).

2012-07-16 08:04:34 by Snowyowl:

The thing about using the artefact as a weapon is that it can be reverse-engineered. In itself it may not be especially useful, except for treating your own people should they get shot, but the principles it's built on would give whoever controlled it a massive technological advantage over everyone else.

As an analogue, suppose a hospital was sent back in time from 2012 to the late 19th century. It contains items that could be used as weapons without much modification - drugs that can kill if injected, X-ray emitters, spinning motors, that sort of thing. But they'd be generally less useful than existing technology; guns and bombs existed at the time that were much better at killing people. On the other hand, the microchips in most of the equipment would easily change the tide of the First World War if they could be reverse-engineered. For example, they could be used to break enemy codes. In fact, the software in the computers probably contains methods of data encryption far superior to anything in use at the time. And the knowledge of chemistry, biology and physics that something like an NMR machine would give you might be enough to kick-start the invention of nuclear weapons.

Okay, that's a bit of a stretch. The point is, the artefact isn't just a machine for making people better. It's also an impossibly complex magical machine, and whoever controls it would be able to reverse-engineer magical techniques far in advance of anything else in use. Normally, technological progress doesn't carry this risk, because it happens everywhere at approximately the same rate. But if one group of people had powerful magic that another group didn't, it might go to their heads.

2012-07-17 01:35:24 by Gigalith:

Nitpick alert: Couldn't they turn the ring on its side, put some kind of medical roller-coaster through, and get throughput of way higher than 0.5 PPS? If the magic is instantaneous, they need only be in the ring for whatever fraction of a second. Even with just stretchers on wheels I think you could get higher speeds. Not that it would change the political situation, of course, but I think even the one-ring-only scenario isn't so bad.

2012-07-17 02:21:59 by James:

Sam wrote:
"The technology used to heal people could very easily be adapted/damaged/corrupted so that it kills people or otherwise makes them ill or broken. The corrupted technology would vastly simpler, just as easy to use, much more portable, require less magic and still have the same capacity for the number of people it could be used on. Essentially, the point being made is that the thing is a new weapon of mass destruction in disguise."

Obviously I'm not gonna argue with you on this point - out of the two of us only one knows how magic works so I'd end up looking stupid.

I will say that that doesn't really come across in the story though.

Grey's comments seem to show that he at least thinks that the healing effect requires passing through the centre - "and forced people through the machine one at a time" - but then after the various scenarios are explained he seems to change that opinion - "You could simply "correct" a living human body to a living body with a hole in it." - based on not particularly anything. Unless his assumption is that the change to a body+hole can only be achieved by passing something through the ring which doesn't seem to be the case.

Hope this doesn't come across argumentative/aggresive/needlessly pedantic/<negative adjective>

2012-07-17 13:13:29 by JoetheRat:

Yes, it would be in itself a horribly inefficient murder weapon. But the principles behind the magitechnology could be reproduced in simpler form. I think we see it in practice:

"You could simply "correct" a living human body to a living body with a hole in it."

"He shoots Grey. No gun. No bullet. He just opens a hole in Grey's heart."

The implication in the wording is more an appeal to Chekhov than Occam, but notable.

2012-07-17 13:56:05 by MHD:

Just to weigh in on all of that meaningless discussion about anthropic pinciples.

The notion of a simulated universe is meaningless under Tegmark 4 cosmology. The T4 postulates that the true laws of physics as a mathematical object, along with the initial conditions for our universe as a mathematical object, together unfolds into the universe we experience.

In essence, "real" is not a privileged state of existence.

This also implies that being in a simulated universe doesn't make any difference. Whoever is simulating is merely opening a "window" into the mathematical definition, with the use of a very large computer.

If you deliberately change the simulation, then that change can be expressed as mathematics too, and now you are looking at a different universe where said change happens at t=...

Lastly, using sentient simulations is IMHO unethical, and preventing powerful AI's from using them is an active research topic in General Artificial Intelligence. But, of course, if humans are the perpetrators, colour me unsurprised.

2012-07-18 09:07:22 by Snowyowl:

Sorry, MHD, I don't understand. How is using simulated realities unethical if they already "exist" in every relevant sense before you perform the simulation?

2012-07-22 16:51:28 by Plod:

It's not simulating in general, it's simulating minds. What do you do when the simulation ends? Turn it off? If they're sapient, that's equivalent to murder. And you can't run it forever.

2012-07-22 20:14:07 by Knut:

Sam, I still don't understand what happens in the shaft. If the youth wanted to talk to Gray before he killed him, why didn't he just have a chat in the shaft? Why blast his head open, reanimate him, and then talk to him outside? Or is this some sort of plot point? Maybe Gray wasn't perfectly remade, but the machine changed him in some way that the youth wanted to observe?

Anyway I can't wait for the next one

2012-07-22 20:20:18 by qntm:

Knut: the youth wanted Grey's perspective on things. Grey (justifiably) didn't listen and attacked the youth, who defended himself. Having neutralised Grey, the youth still wanted Grey's perspective, so he reanimated Grey and resumed the conversation.

It was also a very quick way for the youth to confirm to Grey (and for me to confirm to the reader) exactly how powerful the machine actually is. It can create matter, and (arguably more impressively) it can correctly reconstruct a working human brain from exploded gore. The alternative would have been to just explain it with words, which is less interesting.

2012-07-22 20:41:44 by Knut:

Also, Plod and MHD, shutting off a sapient simulation <em>is</em> equivalent to murder. The most perfectly humane kind of murder imaginable, and therefore not a problem at all. Consider what (the totally awesome) Epicurus said: When we exist death is not, and when death exists we are not. All sensation and consciousness ends with death and therefore in death there is neither pleasure nor pain. The fear of death arises from the belief that in death there is awareness.

Shutting off a sapient simulation, or killing someone painlessly, is a morally completely neutral act if we ignore the ripple effects of the murder on other sapiences.

2012-07-26 04:33:34 by DanielC:

Knut, that's like saying 'murder is okay if you can do it totally painlessly and if the murderee has no friends'.

2012-07-27 03:10:58 by Thrack:

Wont there be a universe identical to the simulation? Shutting off the simulation doesn't destroy the existence of the universe so I wouldn't think it counts as murder (with those assumptions on how the universe works). It would simply prevent you from seeing what will happen in their future.
Unless you count shutting down the duplicate universe as murder regardless of how many exact copies there are?

Hmnn... here's a thought experiment that may help resolve this; instead of dealing with universes let's make an exact duplicate of a person and store it in a computer via advanced atomic level scanning procedures. I'm not sure if it makes any difference whether or not you are running the copy in a simulated universe but if so then let's suppose we are simulating the person's surroundings to a ridiculous level of accuracy. Now, is deleting the person murder, even though it is no different from the original? Quite an interesting question really... I think I would be willing to delete my own duplicate in such a situation but others may not.

This just seems like an easier way to think about it since it uses concepts that are more concrete.

2012-07-27 03:17:23 by Thrack:

"I'm not sure if it makes any difference whether or not you are running the copy in a simulated universe [...]"

By this I mean, is it sufficient if the copy exists in a file, *not* run in a simulation? If it is a static, unchanging series of bits that perfectly describe a person? Is merely the potential for thought and self-awareness sufficient to create the same moral dilemma and make deleting it potentially murder?

2012-09-19 02:53:52 by DanielLC:

The problems with murder are:
* It's painful for the victim.
* It's painful for people who care about the victim.
* It keeps the victim from feeling happiness.

The first two problems don't apply when turning off a simulation, but the last one, the opportunity cost, does. What's more, as long as you have the capacity to simulate people, you ought to use it. Otherwise, there's no happiness generated, and lots of opportunity cost.

2013-03-03 05:34:04 by Banksfan:

so I *just* got that "Outside Context Medicine" is a reference to Excession. Awesome.

2013-07-04 18:04:46 by Eitan:

This conversation is heavily reminding me of Permutation City.

"Murder" is a legal term. If shutting down a simulation isn't consider murder under the law, then it isn't murder.

Knut, you might want to read a little more about the philosophy of death so that twenty years from now you don't take it upon yourself to 'relieve' yer mother's suffering by pulling the plug in the hospital or something. I didn't realize that people still took Epicurus seriously today, but hey, if we have people raving about Atlas Shrugged, then I guess anything's possible.

2013-07-04 18:05:30 by Eitan:


2014-02-20 06:21:15 by Creaphis:

Ignoring the direction taken by previous comments, I just want to note that I find it interesting how, in Sam's stories, scientists of all stripes seem surprisingly comfortable with the idea of being in a simulated universe. Grey, our thaumaturgist/geologist/archaologist is apparently also so well-versed in this particular branch of cosmology that he can spin new theories under threat of imminent death. I'm also thinking of "I don't know, Timmy, being God is a big responsibility" where the scientists find/create incontrovertible evidence that they're in a simulated universe and then, ho-hum, go home for the weekend.

This isn't a literary or scientific criticism. I love everything on this site. I just don't think I could roll with that punch quite so easily.

2014-07-04 10:42:12 by Watercressed:

How does "shiny and silvery" rule out osmium? The picture on Wikipedia ( fits that description.

2015-09-24 04:04:23 by jstanley:

I don't understand why Natalie "didn't need to be killed" after learning about Ra, but eight people who found a big magic ring (seven of which _didn't even know what it does_) had to die. And even then, if Exa is acting for the greater good here based on extremely high-fidelity simulations, why he is brutally murdering innocent people one by one instead of just instantaneously vaporising the entire camp?

2016-07-11 11:46:26 by john:

Did some quick math. If you stack people end-to-end and move them at a little under the speed of sound, a population of twelve and a half billion could get an average of one treatment per person per year, assuming the machine works that fast. Setting it up would present engineering and administrative challenges, but nothing obviously insurmountable. It'd essentially be a scaled-down bullet train, combined with prioritized waiting lists of the sort the NHS already uses.

It wouldn't replace conventional medicine, let alone agriculture, so long as access to the machine itself was finite. People wouldn't forget how to do triage. What would otherwise be the most hopeless cases go to into the miracle box; If there aren't enough hopeless cases in a given year to keep the machine running at capacity, you start sending people with conditions that are reliably curable but only at prohibitive expense, or the backlog of severe and otherwise-permanent disabilities. Once those are done, you start resurrecting medal-of-honor recipients and other war heroes or (former) living national treasures, followed by murder victims who might be able to identify their attacker. Otzi the Iceman is at the top of that last, both for scientific reasons and by virtue of seniority, so I don't think they'd get to the bottom before someone cleverer than me figured out the rest of the priorities.