I challenge you: your best alternate to materialism

[dlorde]

Here's a link to the comic: http://xkcd.com/505/

...If the guy moves rocks around to spell "Eat at Joe's", there's no simulated universe. But if he moves them around in just the right way (in some way that he knows will produce a simulated universe, I guess), then you get a simulated universe. It's also unclear what, exactly, is producing the simulation: the person's brain? The moving rocks themselves? The brain and the rocks? The special pattern the person is following? Which of these are necessary to simulate consciousness? Which are sufficient?

You'd have similar difficulty identifying what produces the information processing capability (and consciousness) in a human brain if you focused at the level of the atoms and molecules in the neurons. It's a matter of multiple levels of abstraction and scale.

The point of Randall's rock-moving XKCD is that the fundamental substrate is irrelevant for information processing (see A Bunch of Rocks Explained) - whether it's binary cells of cellular automata, the atoms and molecules of a biological brain, or the rocks of Randall's universe simulation, if a sufficient number of elements interact according to a consistent set of rules, their patterns of activity can constitute information processing.

In the case of the rocks, the rules and interactions are implemented by Cueball (although it could be done by an automaton); with cellular automata, the rule implementor is usually a simple program running on a microprocessor; and in a biological brain the rules are the implicit laws of physics & chemistry.

 

[Fudbucker]

You'd have similar difficulty identifying what produces the information processing capability (and consciousness) in a human brain if you focused at the level of the atoms and molecules in the neurons. It's a matter of multiple levels of abstraction and scale.

I was waiting for someone to bring that up. You are correct sir, the same problem with rock-consciousness arises with neuron-consciousness: why is some pattern of neural activity associated with a conscious experience while another pattern is just noise?

I think in both cases, you run into a problem of arbitrariness. THESE brain states are correlated with the conscious experience of toe-stubbing. THOSE brain states are correlated with experiencing a beautiful sunset. I think if you were an alien studying our brains you would wonder what the underlying rule is for why X brain state correlates to Y conscious experience. Which, of course, shows up in the moving-rocks-comic. Why is rock movement pattern XYZ so unique? Why does that arrangement simulate a universe while pattern ABC is just rocks moving around?

The point of Randall's rock-moving XKCD is that the fundamental substrate is irrelevant for information processing (see A Bunch of Rocks Explained) - whether it's binary cells of cellular automata, the atoms and molecules of a biological brain, or the rocks of Randall's universe simulation, if a sufficient number of elements interact according to a consistent set of rules, their patterns of activity can constitute information processing.

I agree, it should be irrelevant. Whatever is functionally equivalent to a working brain should be conscious. But this leads to absurd results, so by reductio ad absurdum, it can't be true. Either that, or simulating a universe by moving rocks around in some way isn't absurd.

In the case of the rocks, the rules and interactions are implemented by Cueball (although it could be done by an automaton); with cellular automata, the rule implementor is usually a simple program running on a microprocessor; and in a biological brain the rules are the implicit laws of physics & chemistry.

I highlighted the point where I think the absurdity reaches it's climax. You can indeed replace the guy moving the rocks around with an automaton (or even a random process). You can replace the rocks with any particles that can be moved around and arranged. You therefore had the real possibility of conscious meteor swarms, clouds, sand dunes...any collection of particles that is being moved around has a real (might be really low, but real) chance of being conscious. It's possible then that we're surrounded by these flashes of consciousness, as huge amounts of particles are being randomly arranged around us all the time.

 

[dlorde]

... why is some pattern of neural activity associated with a conscious experience while another pattern is just noise?

That's how it is with patterns. One pattern of carbon atoms gives you graphite, another gives you diamond, another gives you graphene; One pattern of activity in Conway's Game of Life emulates a Turing Machine, another emulates Conway's Game of Life; one pattern of neural activity is associated with a conscious experience while another pattern is just noise.

The first two systems (carbon and GOL) are simple enough that we can explain in detail why each set of patterns produces the results it does. The third is too complex to be explained at present, although we understand parts of it.

I think in both cases, you run into a problem of arbitrariness. THESE brain states are correlated with the conscious experience of toe-stubbing. THOSE brain states are correlated with experiencing a beautiful sunset.

See above (and below). We do have some high level understanding of the whys; we are missing large chunks of the detailed hows. The brain is a very complex set of interacting systems.

Why is rock movement pattern XYZ so unique?

Any pattern that is different from all other patterns is unique. That's what unique means.

Why does that arrangement simulate a universe while pattern ABC is just rocks moving around?

Why does one computer program produce Grand Theft Auto while another produces Microsoft Word, and another just loops indefinitely, doing nothing?

The difference is in the way their patterns of activity can result in various kinds of information processing (or not).

Whatever is functionally equivalent to a working brain should be conscious. But this leads to absurd results, so by reductio ad absurdum, it can't be true. Either that, or simulating a universe by moving rocks around in some way isn't absurd.

The absurdity is in the impracticality of the rock simulation example, not the underlying principle. The flexibility of application of the principle isn't absurd if you understand its implications. It might seem absurd to think that a bunch of elementary particles, or atoms, or chemicals, could somehow all interact to make a complex living creature - if you didn't already have evidence that they can and do. It might seem absurd to think you could simulate the national economy using hydraulics, but it's been done.

...any collection of particles that is being moved around has a real (might be really low, but real) chance of being conscious. It's possible then that we're surrounded by these flashes of consciousness, as huge amounts of particles are being randomly arranged around us all the time.

No. Unless you're considering extended cosmological timescales in which (hypothetically) almost anything can happen (e.g. Boltzmann Brains), the number of ways the particles required to constitute an information processor with the capacity of a conscious brain can be arranged are so vast compared to the number of arrangements that might constitute such a processor, that it is, to all intents and purposes, impossible. To suggest that such a random arrangement of particles could then maintain sufficient coherent structure, and have access to and process sufficient information of an appropriate kind, over a sufficient time to achieve a momentary awareness, really is absurd.

Information processing systems are dissipative structures that must be constructed and maintained by manipulating entropy gradients; they develop over time, they don't arise spontaneously at random (except at hypothetical Boltzmann Brain timescales).

 

[Fudbucker]

That's how it is with patterns. One pattern of carbon atoms gives you graphite, another gives you diamond, another gives you graphene; One pattern of activity in Conway's Game of Life emulates a Turing Machine, another emulates Conway's Game of Life; one pattern of neural activity is associated with a conscious experience while another pattern is just noise.

There are physical explanations for why arrangements of particles produce different substances. There is no physical explanation for why one pattern of rock-movement simulates a universe while another doesn't. There aren't any electrons being shared; the rocks don't form a crystalline structure. It's taken on faith that rocks moved around this way simulate a universe, while rocks moved around that way don't.

In the case of software, there is an interpreter to assign meaning to the various patterns of binary switches the computer is instructed to do. There's no physical explanation for why binary switch pattern ABC plays chess while pattern CBD plays checkers. You need someone (or something) that can assign meaning to the different pattern of switches. Where is the interpreter in a random rock moving machine that moves enough rocks around so that it simulates a universe? There is no interpreter. The whole thing falls apart without one.

This is the point where Myriad got hung up, so much so that Myriad claimed the person moving the rocks around could not be substituted with a machine making random rock placements. Myriad realized you need an interpreter to give meaning to the movements of the rocks, but any pattern of rock movement that a person can do can be reproduced by a random process, so Myriad was forced to conclude there is some special pattern of rock movement that a person could do that could not duplicated by a machine moving rocks around randomly. And that claim is absurd.

The first two systems (carbon and GOL) are simple enough that we can explain in detail why each set of patterns produces the results it does. The third is too complex to be explained at present, although we understand parts of it.

Again, this deals with the realm of physical processes. Consciousness is supposed to result from computational processes.

See above (and below). We do have some high level understanding of the whys; we are missing large chunks of the detailed hows. The brain is a very complex set of interacting systems.

Right, but there won't be a physical explanation for why neurons XYC produce conscious experience ABC, just as there is no physical explanation for why rock pattern ABC simulates a universe. The explanation is supposed to be computational, not physical.

Any pattern that is different from all other patterns is unique. That's what unique means.

That's tautologically true and misses the point. All patterns are unique (unless it's a repitition of a previous pattern), but why does this one pattern result in a simulated universe? What makes that pattern so special?

Why does one computer program produce Grand Theft Auto while another produces Microsoft Word, and another just loops indefinitely, doing nothing?

Because the person observing the computer interprets it that way. A computer out in the woods with no one observing it doesn't produce anything (except some noise and radio waves). It's just a collection of electrons, moving switches, fans, etc. You need an interpreter for software to be meaningful.

I'm reminded of the debates Piggy would get into on this point- in order for a simulation of a tornado to exist, you need a person to look at whatever's doing the simulating and say, "Hey, that looks like a tornado". Without that, there's no simulated tornado. It's not like there's a little tornado in the computer. A tornado simulation is just a bunch of pixels turning off and on in a certain way (maybe some noises being produced).

To tie this all back to rock-consciousness, if a simulation requires an observer, where is the observer in the system of a machine randomly moving rocks around in such a way that a simulated universe is produced?

To tie THAT back to the brain, where is the observer that gives meaning to the patterns of neural activity taking place? I guess you could claim that the system observes itself and attaches meaning to its own activities, but that's a horribly circular claim that I don't think would stand up.

The difference is in the way their patterns of activity can result in various kinds of information processing (or not).

And information processing requires an information processor. In other words, simulations don't exist without something to process the information being displayed.

The absurdity is in the impracticality of the rock simulation example, not the underlying principle. The flexibility of application of the principle isn't absurd if you understand its implications. It might seem absurd to think that a bunch of elementary particles, or atoms, or chemicals, could somehow all interact to make a complex living creature - if you didn't already have evidence that they can and do. It might seem absurd to think you could simulate the national economy using hydraulics, but it's been done.

Physical systems aren't the same as computational systems, so there's no absurdity in the idea of particles bonding in certain ways and producing different substances.

In the case of economics, you run into the same problem as simulations- there is no economics without conscious beings that can give meaning to the various pieces of paper being passed around. Machines randomly passing paper around might duplicate exactly the same movements of people passing the same paper around, but a bunch of machines operating randomly isn't an economy, anymore than a bunch of pixels on a screen lighting up a certain way is a tornado.

The problem of rock-consciousness isn't scale. It's the idea that a simulation can arise from a random process (which is possible), and not require an observer to give the simulation some kind of meaning (which is impossible). Without someone to look at how the rocks are being arranged, it's just rocks moving around on a desert floor, no matter how they're moved.

No. Unless you're considering extended cosmological timescales in which (hypothetically) almost anything can happen (e.g. Boltzmann Brains), the number of ways the particles required to constitute an information processor with the capacity of a conscious brain can be arranged are so vast compared to the number of arrangements that might constitute such a processor, that it is, to all intents and purposes, impossible. To suggest that such a random arrangement of particles could then maintain sufficient coherent structure, and have access to and process sufficient information of an appropriate kind, over a sufficient time to achieve a momentary awareness, really is absurd.

1. My point was not probability-based. I'll grant you that the odds of a shifting sand dune becoming conscious are really really small, but the possibility still exists. That means there's a possible world (in the modal sense) of conscious shifting sand dunes. I don't agree with Barehl usually, but he is right to discount even the possibility of conscious sand dunes, storms, etc. as absurd, no matter what the odds are. That's where the absurdum kicks in- any materialistic explanation of consciousness that allows for even the possibility of conscious sand dunes needs to be discarded.

2. We don't really know what the odds are. How many rocks do you have to arrange to simulate a universe? How may patterns are there that result in a simulated universe? How many shifting sand particles are needed to give a sand dune the approximate conscious awareness of, say, a bat? How many patterns of sand movement exist that can produce a moment of conscious awareness? For all we know, there are countless different ways sand particles can move around to produce conscious experience. Maybe every landslide that ever happens is a brief moment of consciousness. Why not? Who assigns the odds on these things?

Information processing systems are dissipative structures that must be constructed and maintained by manipulating entropy gradients; they develop over time, they don't arise spontaneously at random (except at hypothetical Boltzmann Brain timescales).

Who's doing the manipulating? If it's not someone with knowledge of computation, what gives a pattern of entropy manipulation a specific meaning? We're back to the problem of unobserved simulations.

 

[Nonpareil]

Here's a link to the comic: http://xkcd.com/505/

I think there's a whole lot more wrong with rock-consciousness than a missing energy source.

With that specific model of rock-consciousness, yes. Because XKCD is a webcomic meant for entertainment which occasionally has actual tidbits of scientific information contained in it, much in the same way that The Daily Show isn't actually a news channel.

Specifically, the rocks themselves there aren't doing any computing. The character's mind is. But a computer could be built out of rocks - or any other material - and there is no reason that, given sufficient time and precision in engineering, one could not make a Turing-complete computer out of any substance whatsoever.

It might run slowly. In a practical instance, the rocks may wear down quickly or be too imprecise. But, again, there is no reason that, given sufficient resources, we could not build a computer out of any substance to run any program at all, even if it does run slowly.

But it is taken for granted by some people here (who should know better), that if you move rocks around you can simulate conscious people. So the next time you see some kids in a dirt-clod fight, tell them to show a little respect for the fleeting conscious entity that may have just popped into existence.

Consciousness is computation. Computation does not care about the medium through which it is performed, but throwing a couple of dirt clods around does not a computer make, let alone an integrated system.

Please stop beating that straw man. He never did anything to you.

I agree, it should be irrelevant. Whatever is functionally equivalent to a working brain should be conscious. But this leads to absurd results, so by reductio ad absurdum, it can't be true.

The reductio ad absurdum argument is not actually an indication of falsehood. It is just a form of argument. It can just as easily be a case of someone committing the actual fallacy of argument from personal incredulity.

Such as it is here.

Either that, or simulating a universe by moving rocks around in some way isn't absurd.

No more so than simulating it through any other medium. Which is to say, it won't happen in actuality unless we're simulating a very small universe, whatever the medium.

I highlighted the point where I think the absurdity reaches it's climax. You can indeed replace the guy moving the rocks around with an automaton (or even a random process). You can replace the rocks with any particles that can be moved around and arranged. You therefore had the real possibility of conscious meteor swarms, clouds, sand dunes...any collection of particles that is being moved around has a real (might be really low, but real) chance of being conscious. It's possible then that we're surrounded by these flashes of consciousness, as huge amounts of particles are being randomly arranged around us all the time.

And this is why we don't treat the Daily Show as our actual source of news, folks.

Again, this deals with the realm of physical processes. Consciousness is supposed to result from computational processes.

Computation is physical.

To tie this all back to rock-consciousness, if a simulation requires an observer, where is the observer in the system of a machine randomly moving rocks around in such a way that a simulated universe is produced?

Are you seriously arguing that a computer ceases to compute if there is no one at the monitor? You are equivocating between having results and assigning meaning to the results.

To tie THAT back to the brain, where is the observer that gives meaning to the patterns of neural activity taking place? I guess you could claim that the system observes itself and attaches meaning to its own activities, but that's a horribly circular claim that I don't think would stand up.

Well, no. That's exactly what being an integrated system means. I refer the readers of this thread once again to the work of Christof Koch.

An integrated computational system cross-links all the data it has collected. The more heavily cross-linked the system, the more integrated it is, and the more meaning each instance of data carries.

Physical systems aren't the same as computational systems

Yes, they are. Computers are physical entities. What are you on about?

1. My point was not probability-based. I'll grant you that the odds of a shifting sand dune becoming conscious are really really small, but the possibility still exists.

If an arrangement of sand has randomly assumed some form that is capable of the computations necessary for consciousness, it is no longer a sand dune. This is just your dirt-clod strawman all over again.

Maybe every landslide that ever happens is a brief moment of consciousness. Why not?

Because a rockslide is not an integrated system.

This is not complicated.

 

[Fudbucker]

With that specific model of rock-consciousness, yes. Because XKCD is a webcomic meant for entertainment which occasionally has actual tidbits of scientific information contained in it, much in the same way that The Daily Show isn't actually a news channel.

Yet there are still many people who believe the news content of the Daily Show is accurate, just as their are people who who believe the content of the rock-consciousness comic is accurate.

Specifically, the rocks themselves there aren't doing any computing. The character's mind is.

Wait, so are you saying the character's mind is a necessary condition for simulating a universe of conscious beings? How does that work? What about Dlorde's point that the person could be replaced by an automaton? Is the simulation of the entire universe of conscious beings happening inside the rock-moving person's brain? That's a mighty big brain, to hold a simulation of an entire universe in it...

But a computer could be built out of rocks - or any other material - and there is no reason that, given sufficient time and precision in engineering, one could not make a Turing-complete computer out of any substance whatsoever.

I agree, which leads you to the rock-consciousness comic, which leads to conscious sand-dunes, which is absurd, and so there must be a problem with step 1.

It might run slowly. In a practical instance, the rocks may wear down quickly or be too imprecise. But, again, there is no reason that, given sufficient resources, we could not build a computer out of any substance to run any program at all, even if it does run slowly.

We seem to have lost the "character's mind" requirement that you thought was necessary earlier on for rock-consciouness. You can indeed build a computer out of almost anything, and there's no requirement that a "character's mind" exist for the computer to function properly (i.e., to simulate the universe by moving rocks around). A random machine could do the same job, if it moved enough rocks around.

And that brings us back to the original objection- if machines go around randomly moving rocks around...so what? How do you get a simulated universe of conscious beings from moving rocks around? What, exactly, is being simulated, how is it being simulated, and who is there to interpret the simulation?

Consciousness is computation. Computation does not care about the medium through which it is performed, but throwing a couple of dirt clods around does not a computer make, let alone an integrated system.

Please. Computation occurs when abacus beads move around. The quadrillions of particles moving around from a dirt clod in motion are orders of magnitude more complex than an abacus.

The reductio ad absurdum argument is not actually an indication of falsehood. It is just a form of argument. It can just as easily be a case of someone committing the actual fallacy of argument from personal incredulity.

Such as it is here.

I admitted that- the possibility of flashes of consciousness from the random arrangement of particles moving around us might not be absurd. But if you want to argue that materialism leads essentially to panpsychism, don't be surprised if people look at you funny. Sand dunes and cloud bursts aren't the sorts of things that should possibly be conscious. The burden of proof is on you to convince us why these things, which don't remotely resemble the only conscious things observed so far (brains) should be considered to be real possibilities of momentary flashes of consciousness.

No more so than simulating it through any other medium. Which is to say, it won't happen in actuality unless we're simulating a very small universe, whatever the medium.

The point isn't about the constraints on the comic (big universe or small), but whether the comic is to be taken seriously at all. People here take it seriously.

And this is why we don't treat the Daily Show as our actual source of news, folks.

Computation is physical.

No, computation results from physical processes. Computational isn't any more physical than the Pythagorean theorem. You don't believe there's a physical Pythagorean theorem floating around somewhere, do you? Can computation exist in a possible world where there are no minds? Picture random machines moving abacus beads without there ever being a mind to interpret the results. Is that an example of computation? I think the answer to that is obvious: no, it's just abacus beads moving around. Minds are necessary conditions for computation to occur.

Are you seriously arguing that a computer ceases to compute if there is no one at the monitor?

Of course I'm going to argue that. All a computer does is move switches around in a predetermined fashion. That's not computation, any more than an abacus falling from a plane, with the beads randomly moving around is computation. There is no computation without someone around to do the computing.

You are equivocating between having results and assigning meaning to the results.

Not at all. I'm claiming computation is evaluating (assigning meaning) to the results. You can have all the results you want. They won't mean anything unless there's someone there to interpret those results. Remember, the results from any computational process are just arrangements of matter (switches, rocks, abacus beads, etc). Without an observer, the results are meaningless.

Well, no. That's exactly what being an integrated system means. I refer the readers of this thread once again to the work of Christof Koch.

LOL, I don't think Koch is who you want to be referring to on this particular subject:

"I think consciousness, like mass, is a fundamental property of the universe. The analogy, and it’s a very good one, is that you can make pretty good weather predictions these days. You can predict the inside of a storm. But it’s never wet inside the computer. You can simulate a black hole in a computer, but space-time will not be bent. Simulating something is not the real thing.

It’s the same thing with consciousness. In 100 years, you might be able to simulate consciousness on a computer. But it won’t experience anything. Nada. It will be black inside. It will have no experience whatsoever, even though it may have our intelligence and our ability to speak."
http://www.technologyreview.com/news/531146/what-it-will-take-for-computers-to-be-conscious/

 

[Belz...]

I think in both cases, you run into a problem of arbitrariness.

Given that it depends on your definition of consciousness, yeah.

 

[Nonpareil]

Yet there are still many people who believe the news content of the Daily Show is accurate, just as their are people who who believe the content of the rock-consciousness comic is accurate.

They're wrong. What's your point?

Wait, so are you saying the character's mind is a necessary condition for simulating a universe of conscious beings? How does that work?

No. I was pointing out why the character dragging rocks about in the comic is not actually creating a computer. The rocks are not part of a single unified system. The only computer present in the comic is his brain. The rocks in the comic are no more a computer - and no more capable of consciousness - than if he were simply taking notes on the simulation running in his head.

Because that's exactly what he's doing, only using rocks instead of ink.

What about Dlorde's point that the person could be replaced by an automaton?

Then the only "brain" present is whatever is driving the automaton.

Is the simulation of the entire universe of conscious beings happening inside the rock-moving person's brain? That's a mighty big brain, to hold a simulation of an entire universe in it...

Yes.

Again, this is why XKCD is a comic strip and not an actual scientific article that you can quote like holy writ. I'm not sure why you're so hung up on this.

I agree, which leads you to the rock-consciousness comic

Except it doesn't. See above.

which leads to conscious sand-dunes

Except it doesn't. See above.

which is absurd, and so there must be a problem with step 1.

Except that this is nothing but the argument from personal incredulity (with a bit of strawman, as mentioned previously).

Reductio ad absurdum, again, is not actually an indicator of a logical fallacy. It is just a format that one's argument can take.

And that brings us back to the original objection- if machines go around randomly moving rocks around...so what? How do you get a simulated universe of conscious beings from moving rocks around?

The same way you would get one using a standard PC, but with rocks in place of electrons or switches or whatever you want to call it.

What, exactly, is being simulated, how is it being simulated, and who is there to interpret the simulation?

That depends entirely upon the parameters of the simulation. Even something as simple as SimCity is a simulation of a universe with conscious beings in it. It's just not a very good or in-depth one.

Please. Computation occurs when abacus beads move around. The quadrillions of particles moving around from a dirt clod in motion are orders of magnitude more complex than an abacus.

Which does not in any way change what I said. Dirt clods are not computers. Computers do not care about the substrate on which they are based.

I admitted that- the possibility of flashes of consciousness from the random arrangement of particles moving around us might not be absurd.

It absolutely is.

But if you want to argue that materialism leads essentially to panpsychism

I do not.

Sand dunes and cloud bursts aren't the sorts of things that should possibly be conscious. The burden of proof is on you to convince us why these things, which don't remotely resemble the only conscious things observed so far (brains) should be considered to be real possibilities of momentary flashes of consciousness.

No, it isn't, because I have never claimed any such thing.

The point isn't about the constraints on the comic (big universe or small), but whether the comic is to be taken seriously at all. People here take it seriously.

I can't speak for other people, but I find it much more likely - in the case of users like dlorde and PixyMisa - that it is you who failed to understand what was being said.

No, computation results from physical processes.

Computation is a physical process taking place within the confines of a closed system called a "computer". This isn't arguable. We can watch the switches move and tell you why.

Can computation exist in a possible world where there are no minds?

All minds are computation. Not all computation is minds. So yes.

Picture random machines moving abacus beads without there ever being a mind to interpret the results. Is that an example of computation? I think the answer to that is obvious: no, it's just abacus beads moving around.

Because an abacus is no more a computer than a sheet of paper that you have written numbers on. There is no computation occurring in the abacus. It is a visual aid for the computation taking place within your head.

If my PC existed in a world without minds, though, engaged in an eternal game of StarCraft against itself, yes. It would be computing.

I have no idea why you think that this is in doubt.

Minds are necessary conditions for computation to occur.

No.

Of course I'm going to argue that. All a computer does is move switches around in a predetermined fashion. That's not computation

Of course. What makes it computation is that the movement of those switches computes mathematical values. That is what computation means.

Not at all. I'm claiming computation is evaluating (assigning meaning) to the results.

And you're wrong.

There isn't anything else to say. That isn't what computation means, and your attempt to equivocate between the two is flatly wrong.

You can have all the results you want. They won't mean anything unless there's someone there to interpret those results. Remember, the results from any computational process are just arrangements of matter (switches, rocks, abacus beads, etc). Without an observer, the results are meaningless.

"Without a subjective observer, subjective opinions do not exist."

Yes. So? Without a conscious observer there to stare at the monitor, a computer could still calculate the orbital period of Jupiter. It just wouldn't care.

LOL, I don't think Koch is who you want to be referring to on this particular subject

I don't particularly care what you think I should be arguing. And yes, I'm fully aware of what it is that Koch is arguing, and I assure you that I have a better understanding of it than you.

"The analogy, and it’s a very good one, is that you can make pretty good weather predictions these days. You can predict the inside of a storm. But it’s never wet inside the computer. You can simulate a black hole in a computer, but space-time will not be bent. Simulating something is not the real thing.

It’s the same thing with consciousness. In 100 years, you might be able to simulate consciousness on a computer. But it won’t experience anything. Nada. It will be black inside. It will have no experience whatsoever, even though it may have our intelligence and our ability to speak."

We are entirely agreed on this point, because simulations are not functionally identical in every way. If they were, they would be the real thing rather than a simulation. I have made this same argument - in fact, I believe that I have pointed it out to you specifically - in several different places across these forums.

A simulation of a storm will not get you wet. A simulation of a universe will not actually cause a second Big Bang. A simulation of a mind will not actually feel - because they are simulations. They are models. They are approximations, not the real thing.

 

[Myriad]

I agree, which leads you to the rock-consciousness comic, which leads to conscious sand-dunes, which is absurd, and so there must be a problem with step 1.

No. The problem is your equating the remote astronomically unlikely possibility of a certain random outcome with the capability of generating that outcome. It's like saying blind people can read printed text because if they guess the letters at random there's a chance they might guess correctly. In theory (but not within the constraints of the actual known universe), enough monkeys and enough typewriters will eventually produce Shakespeare's Hamlet. The error you're making is concluding from that that monkeys can write.

Also, I'll point out that calling the xkcd rock-universe a "simulation" is only correct from the narrator's and the readers' points of view, and only by reference to our universe. (The rock-universe may or may not be a simulation of our universe, but it cannot be a simulation of itself!) The patterns in the computation are what they are, whether or not they're also a simulation of anything else.

If it rains in some locale within the rock-universe, the rocks don't get wet. But neither do the subatomic particles that make up the hydrogen atoms in the water molecules in a rainstorm. Wetness is a perceived property of a statistical pattern in the distribution of molecular species, just as the rain in a weather forecasting computer is a perceived property of a statistical pattern in the distribution of values in an array of state data. Either they're both real, or neither. Choose as you will, but don't try to sneak in a meaningless distinction by changing the definition of "real" on a whim between one and the other.

 

[dlorde]

It's taken on faith that rocks moved around this way simulate a universe, while rocks moved around that way don't.

Cueball arranges the rocks in the required pattern (the state of the universe at some chosen time) and then manipulates them iteratively as elements of a cellular automaton. It simulates a universe because that's what he's set it up to do, just like you can set up Conway's Game of Life to simulate a Turing Machine (although being computational, in that case the simulation is also an emulation).

In the case of software, there is an interpreter to assign meaning to the various patterns of binary switches the computer is instructed to do. There's no physical explanation for why binary switch pattern ABC plays chess while pattern CBD plays checkers.

Well, yes there is a physical explanation - the program algorithms and input data (the initial binary switch patterns) ensure certain correlations between the physical activity of each game (e.g. the moves) and the switching activity in the simulation in accordance with the application of the game rules.

You need someone (or something) that can assign meaning to the different pattern of switches. Where is the interpreter in a random rock moving machine that moves enough rocks around so that it simulates a universe? There is no interpreter. The whole thing falls apart without one.

Not really. It's the old tree falling in the forest conundrum. The tree still falls, the air still vibrates at certain frequencies; nothing falls apart, it just isn't observed.

Everything needs interpreting before it makes sense - babies have to learn how to interpret the world around them, but the world is there regardless. Simulations are no different - you can leave an evolution simulator running overnight, uninterpreted, while you sleep, and in the morning you can see how far it's got - does it mean anything at 4am when there's no-one looking? Maybe, maybe not, but it's still simulating evolution, and it will still produce results even if you die in the night. In the Moving Rocks XKCD, Cueball appears to be both rule executor and state interpreter.

This is the point where Myriad got hung up, so much so that Myriad claimed the person moving the rocks around could not be substituted with a machine making random rock placements. Myriad realized you need an interpreter to give meaning to the movements of the rocks, but any pattern of rock movement that a person can do can be reproduced by a random process, so Myriad was forced to conclude there is some special pattern of rock movement that a person could do that could not duplicated by a machine moving rocks around randomly. And that claim is absurd.

I don't think a person is required. All that is needed is an initial pattern and some means of moving the rocks according to set rules, and the information will be processed. Of course, for it to be useful or meaningful to observers, they need to observe it and understand how to interpret its states - like a foreign film without subtitles...

Again, this deals with the realm of physical processes. Consciousness is supposed to result from computational processes.

Computational processes abstractly describe physical processes that proceed according to certain rules.

... there won't be a physical explanation for why neurons XYC produce conscious experience ABC, just as there is no physical explanation for why rock pattern ABC simulates a universe. The explanation is supposed to be computational, not physical.

A computational explanation is a generalised description of the way physical processes, that proceed according to certain rules, behave.

... why does this one pattern result in a simulated universe? What makes that pattern so special?

It's special because Cueball set it up to correspond to an initial state of a universe and with rules that, iteratively applied, produce patterns of activity in the rocks that correspond to patterns of activity in a physical universe.

Because the person observing the computer interprets it that way. A computer out in the woods with no one observing it doesn't produce anything (except some noise and radio waves). It's just a collection of electrons, moving switches, fans, etc. You need an interpreter for software to be meaningful.

You need an interpreter for anything to be meaningful.

... A tornado simulation is just a bunch of pixels turning off and on in a certain way (maybe some noises being produced).

That was a category error in a different debate. A tornado simulation won't blow your roof off, but you can do calculations on a calculator simulation. A simulation that performs the same computations is an emulation.

To tie this all back to rock-consciousness, if a simulation requires an observer, where is the observer in the system of a machine randomly moving rocks around in such a way that a simulated universe is produced?

A machine randomly moving rocks around could not simulate a universe. I don't know where you got that from. The rocks must be moved according to a consistent set of rules.

To tie THAT back to the brain, where is the observer that gives meaning to the patterns of neural activity taking place?

There is no observer giving meaning to the patterns of neural activity (although external observers may give meaning to the physical results of that activity); the patterns of neural activity themselves constitute the experiencing observer and his/her experiences.

And information processing requires an information processor. In other words, simulations don't exist without something to process the information being displayed.

I don't see your point here - a processor is just a mechanism to manipulate data according to a set of transforms, in the Moving Rocks XKCD, Cueball plays the role of processor. In computer simulations, microprocessor(s) do the manipulation. In neuromorphic simulations, neural networks do it.

Physical systems aren't the same as computational systems, so there's no absurdity in the idea of particles bonding in certain ways and producing different substances.

I don't follow you - computational systems are physical systems that behave in certain ways.

The problem of rock-consciousness isn't scale. It's the idea that a simulation can arise from a random process (which is possible), and not require an observer to give the simulation some kind of meaning (which is impossible).

The process isn't random, it's painstakingly set up and implemented according to specific rules by Cueball, who is the observer.

Without someone to look at how the rocks are being arranged, it's just rocks moving around on a desert floor, no matter how they're moved.

Cueball is there - he's running the simulation.

The amusing (and possibly thought-provoking) conceit of the XKCD cartoon is that after describing how a universe can be simulated in a cellular automaton, Randal has Cueball raise the 'Our universe as a simulation' question by suggesting we could be entities in his simulation. It's a philosophical question with an interesting twist - some physicists have suggested that information is the fundamental 'stuff' of reality; if so, an indistinguishable computational simulation/emulation would, in principle, be possible, and thus, we could be entities in such an simulation/emulation.

1. My point was not probability-based. I'll grant you that the odds of a shifting sand dune becoming conscious are really really small, but the possibility still exists.

I seriously doubt that it's physically possible, but if the same information is processed in the same way over the same time by some system, you'll get the same results. The absurdity is in thinking that everything that is not physically impossible should be taken seriously, however unlikely. Statistical mechanics suggests otherwise.

That means there's a possible world (in the modal sense) of conscious shifting sand dunes.

Also a possible world where eggs unsmash and tower blocks undemolish themselves. Possible, but absurd.

2. We don't really know what the odds are. How many rocks do you have to arrange to simulate a universe?

Clearly a lot. Randal may have estimated it, why not ask him?

How may patterns are there that result in a simulated universe?

I don't know; any such patterns would necessarily be dynamic.

...Maybe every landslide that ever happens is a brief moment of consciousness. Why not? Who assigns the odds on these things?

Why not? because even if it's physically possible (which I doubt), it's obviously absurdly unlikely; you're describing a form of Boltzmann Brain.

Who's doing the manipulating? If it's not someone with knowledge of computation, what gives a pattern of entropy manipulation a specific meaning? We're back to the problem of unobserved simulations.

It was just a turn of phrase - the common examples of such systems are living creatures, and we tend to think of them as active agents, but at a lower level of abstraction they're just replicating physical systems driven to increasing complexity (thus maximising energy dissipation) by environmental energy/entropy gradients.

 

[Fudbucker]

It's taken on faith that rocks moved around this way simulate a universe, while rocks moved around that way don't.

Cueball arranges the rocks in the required pattern (the state of the universe at some chosen time) and then manipulates them iteratively as elements of a cellular automaton. It simulates a universe because that's what he's set it up to do, just like you can set up Conway's Game of Life to simulate a Turing Machine (although being computational, in that case the simulation is also an emulation).

You're running into the same problem that Myriad ran into. Your claim is Cueball knows of some way (some "required pattern") to arrange rocks so that he can simulate a universe. Fine. But that's not a necessary condition to produce a simulation of a universe. Cueball's knowledge of some simulation-producing arrangement of rocks doesn't preclude the possibility of a random arrangement of rocks that's exactly the same pattern as the one Cueball has in mind (and the arrangement Cueball eventually produces). There is no "required pattern" Cueball can come up with that can't be randomly duplicated.

Cueball thinks, "I have all these rocks. I can simulate a universe if I start moving rocks around this way: 101001001010101...". And in some other universe, a random rock moving machine starts placing physically identical rocks exactly the same way as Cueball. The random rock moving machine duplicates Cueball's pattern exactly. So after awhile, you have Cueball's pattern:
10010101001001010001000101010....001010
And the rock-moving machine's pattern:
10010101001001010001000101010....001010

And your claim is that Cueball's pattern is a simulation of a universe, while the rock-machine's pattern, which is identical to Cueball's pattern, isn't? Am I reading you right on that? Because that's absurd. There's nothing special about what Cueball did, nothing that can't be duplicated by a random process. As I told Myriad, it's like claiming there's a book you can write that an infinite numbers of monkeys chained to typewriters can't duplicate. Such a book doesn't exist, no matter what "required pattern" you have in mind before you start writing it. Likewise, there's no special pattern Cueball can come up with that the machine randomly moving rocks around can't eventually match.

Here's another way to think of it: imagine you're observing a computer running some software designed to simulate a tornado. Now imagine a bunch of atoms randomly come together and form a Boltzmann Computer, right next to the computer you're observing. The Boltzmann Computer's components are doing the exact same thing as the computer running the tornado simulation: microchips are functioning identically, the same pixels are turning off and on, etc. Both computers would be simulating a tornado, would they not? But one of the computers was never built, was never designed, and was never programmed. It just happens to randomly be doing the exact same thing as the computer that WAS designed, built and programmed. But it would make no difference to you- both computers would be simulating tornadoes.

Do you agree that the Boltzmann computer is simulating a tornado (for however long it runs)? If you agree with that (as I think you must), then you'll have to agree that a random rock moving machine can simulate a universe.

Or take Boltzmann Brains. When (if) they pop into existence, there's no Boltzmann Brain builder moving atoms around with a special pattern of neurons and brain activity in mind. The Boltzmann Brain just randomly pops into existence, and is conscious for as long as it can function (and if we're positing Boltzmann Brains, we can posit life-support apparatus also popping into existence that's connected to the Boltzmann Brain). So in that case too, you don't need someone like Cueball, who knows how to arrange things a certain way, in order to produce consciousness. You just need enough random stuff to happen that is the functional equivalent of a working brain.

This is an important point, so I'm going to stop here. Are we agreed that a machine randomly moving rocks around can simulate a universe?

 

[Nonpareil]

You're running into the same problem that Myriad ran into.

No. You're still running into the Daily Show problem of trying to take a joke webcomic perfectly seriously in terms of scientific accuracy.

The rocks themselves are not performing any computation. They are an abacus, nothing else. They are not a computer; they are a means that the character uses to keep track of the simulation in his head. The only actual computation happening is inside Cueball's mind. Everything else is notation. He is just writing things down using rocks. The rocks themselves are not part of a unified system and mean nothing except to him.

And yes, it's technically possible for, somewhere out there, a group of random rocks to be the same as one of Cueball's output arrangements. It doesn't matter. Again, there's no actual computation taking place. It's technically possible that, somewhere out there, there's a field of wildflowers that looks exactly like van Gogh's Starry Night.

It is theoretically possible to build a computer out of rocks. It is also theoretically possible to simulate a universe using this rock computer. Cueball has not done so.

 

[dlorde]

You're running into the same problem that Myriad ran into. Your claim is Cueball knows of some way (some "required pattern") to arrange rocks so that he can simulate a universe. Fine. But that's not a necessary condition to produce a simulation of a universe. Cueball's knowledge of some simulation-producing arrangement of rocks doesn't preclude the possibility of a random arrangement of rocks that's exactly the same pattern as the one Cueball has in mind (and the arrangement Cueball eventually produces). There is no "required pattern" Cueball can come up with that can't be randomly duplicated.

The same argument applies to any process, computational or explicitly physical. If you turn on your computer, and by some chance fluke a random series of electronic glitches causes it to exactly simulate a spreadsheet loading your tax data and working out your tax position for the year, has it processed that information or not? If it randomly generates a Stuxnet-type virus and sends it onto the internet, where it disables power stations and water supplies, are you going to say it's not a real virus, just a meaningless random collection of bits? If you write a novel and a thousand other people write a word-for-word identical novel what do you (and they) do?

And your claim is that Cueball's pattern is a simulation of a universe, while the rock-machine's pattern, which is identical to Cueball's pattern, isn't? Am I reading you right on that?

No, I haven't said anything of the sort. If they are identical sequences of events they are both simulations of the same thing, processing information whether there's anyone around to appreciate them or not. Someone could turn up at a later date, analyse the patterns in the rocks and conclude that they correspond exactly to the patterns of activity in a nearby universe (then try telling them that it's entirely random...)

Your idea of random sequences simulating ordered sequences raises interesting questions about the meaning of randomness, but doesn't go anywhere. They may be theoretically possible, but so unlikely they can be discounted. Statistical mechanics (thermodynamics) and simple probability theory tell us why.

... Do you agree that the Boltzmann computer is simulating a tornado (for however long it runs)? If you agree with that (as I think you must), then you'll have to agree that a random rock moving machine can simulate a universe.

I don't have a problem with that, as a hypothetical.

Or take Boltzmann Brains... in that case too, you don't need someone like Cueball, who knows how to arrange things a certain way, in order to produce consciousness. You just need enough random stuff to happen that is the functional equivalent of a working brain.

Boltzmann brains can be conscious - your point is?

Are we agreed that a machine randomly moving rocks around can simulate a universe?

Theoretically it could simulate anything, but I see it the way Eddington did when considering monkeys at typewriters (according to Wikipedia):

[he invited] the reader to consider the incredible improbability of a large but finite number of monkeys working for a large but finite amount of time producing a significant work, and compare this with the even greater improbability of certain physical events. Any physical process that is even less likely than such monkeys' success is effectively impossible, and it may safely be said that such a process will never happen. It is clear from the context that Eddington is not suggesting that the probability of this happening is worthy of serious consideration. On the contrary, it was a rhetorical illustration of the fact that below certain levels of probability, the term improbable is functionally equivalent to impossible.

 

[dlorde]

The rocks themselves are not performing any computation. They are an abacus, nothing else. They are not a computer; they are a means that the character uses to keep track of the simulation in his head. The only actual computation happening is inside Cueball's mind. Everything else is notation. He is just writing things down using rocks. The rocks themselves are not part of a unified system and mean nothing except to him.

I think this is wrong - the cartoon is explicit that Cueball has built a computer; Cueball is using the rocks as a cellular automaton that he has set up to emulate a computer (panel 11: "I was able to build a computer", panel 12 "... It's rocks instead of electricity, but it's the same* thing... *Turing complete"). He then programs this to be a physics simulator that can fully simulate two particles interacting, then goes on to extend this to simulating a whole universe.

So the rocks are participating in the computation as fundamental elements in the same way the cells of Game of Life participate in GOL Turing Machine simulation computations; the actual computation occurs a couple or more levels of abstraction above the cell on/off or the rocks changing position. The computer would be set up as a pattern of rocks on the CA grid, the program to run would be set up as another pattern of rocks to be 'read' by the CA computer, and the universe state data would be another pattern of rocks, also read by the computer.

Once Cueball has set up the cellular automaton with rocks in the pattern of a Turing Complete computer, and set up the universe simulator program (with the physics program as its heart) and the data for it to process, his role is to iterate the CA rules on the rocks of the CA. If he's set it up correctly, this will make CA simulate/emulate the computer, which will in turn run the program on the data, and so compute the simulation.

He doesn't say how he accesses the simulation output, but I guess he's too busy moving rocks...

 

[Nonpareil]

I think this is wrong - the cartoon is explicit that Cueball has built a computer; Cueball is using the rocks as a cellular automaton that he has set up to emulate a computer (panel 11: "I was able to build a computer", panel 12 "... It's rocks instead of electricity, but it's the same* thing... *Turing complete"). He then programs this to be a physics simulator that can fully simulate two particles interacting, then goes on to extend this to simulating a whole universe.

Well, it says this, but I think it's wrong.

The rocks aren't moving. The rocks themselves are at no point actually changing state to result in the next pattern displayed. Cueball is explicitly the one doing all the moving.

Take the same situation. Replace infinite space and infinite rocks with infinite paper space and infinite ink. Cueball writes down everything in single dots of ink instead of rows of rocks. Again, the dots themselves are not a computer; they are just a visual aid (or, if you like, the output) for the computation taking place inside his head.

I don't disagree that you could build a Turing-complete computer out of rocks, or that what Cueball does is not Turing-complete computation, but the computation in question is very clearly taking place inside his head. The rocks themselves are no more a computer than a sheet of paper with instructions for completing the next step written on it.

 

[dlorde]

I don't disagree that you could build a Turing-complete computer out of rocks, or that what Cueball does is not Turing-complete computation, but the computation in question is very clearly taking place inside his head. The rocks themselves are no more a computer than a sheet of paper with instructions for completing the next step written on it.

I guess it depends what you consider the computation to involve. All Cueball is doing (once he's set up the initial patterns) is to 'blindly' iterate the CA rules - which are trivially simple - by moving the stones. He could tell someone else what to do, and they could do it without having any idea that this was a CA emulating a processor running a simulation. It could also be done by a simple automaton.

It's fair to say that the whole system, Cueball included, does the simulation computation, but if Cueball is keeping track of anything in his head it's the next rock and the CA rule to execute on it. His role is like a RISC processor running microcode to emulate a microprocessor that can run a program to execute a simulation.

I suppose you could also say Cueball is computing the CA, which is emulating the processor, which is computing the simulation; but he's just a cog in the machine, he doesn't even need to know that the CA is emulating a processor.

It's the interacting patterns [of rocks] that really do the computation.

 

[Belz...]

Randal has Cueball raise the 'Our universe as a simulation' question by suggesting we could be entities in his simulation. It's a philosophical question with an interesting twist - some physicists have suggested that information is the fundamental 'stuff' of reality; if so, an indistinguishable computational simulation/emulation would, in principle, be possible.

Indeed. If, as I suspect, the universe is just composed of data points with certain parameters, there is no actual difference between the universe and a simulation of one.

 

[Belz...]

Are we agreed that a machine randomly moving rocks around can simulate a universe?

You didn't actually read his post, did you ?

 

[Nonpareil]

I guess it depends what you consider the computation to involve. All Cueball is doing (once he's set up the initial patterns) is to 'blindly' iterate the CA rules - which are trivially simple - by moving the stones. He could tell someone else what to do, and they could do it without having any idea that this was a CA emulating a processor running a simulation. It could also be done by a simple automaton.

This is true. However, it is important to note that the rocks are still not a unified system. There is nothing about the rocks themselves, or in their arrangement, that results in the creation of the next output state.

By contrast, in an actual computer (even one emulated by the Game of Life or anything equivalent), the switches are arranged in such a way that the behavior of one affects the state of the rest.

It's fair to say that the whole system, Cueball included, does the simulation computation

This is where we hit our disagreement. From what I understand - and I could be wrong, as, I repeat, I'm just an enthusiastic layman with some basic training - there is no functional difference between this system and Cueball sitting at a desk with a list of instructions on them. These instructions tell him how to write the next set of instructions, which will tell him how to write the next, and so on and so forth.

Make that list of instructions as big as you like, but it's not part of a computer. The only computation taking place is in his head.

 

[dlorde]

This is true. However, it is important to note that the rocks are still not a unified system. There is nothing about the rocks themselves, or in their arrangement, that results in the creation of the next output state.

Yes, the rocks on their own are just rocks. To be a CA, they must be moved according to a set of rules. They're the equivalent of the black/white cells of GOL. But their arrangement does influence the next output state by the application of the rules; that's how the patterns move and interact (see Rule 110, a candidate for Cueball's CA).

By contrast, in an actual computer (even one emulated by the Game of Life or anything equivalent), the switches are arranged in such a way that the behavior of one affects the state of the rest.

I don't see a fundamental difference - in a binary digital computer, patterns of bits interact; in a moving rock CA, patterns of rocks interact. The means are obviously different, but the principle is the same.

From what I understand - and I could be wrong, as, I repeat, I'm just an enthusiastic layman with some basic training - there is no functional difference between this system and Cueball sitting at a desk with a list of instructions on them. These instructions tell him how to write the next set of instructions, which will tell him how to write the next, and so on and so forth.

That's right - the CA cells could be made using rocks, flags, dots on a piece of paper, or bits in RAM. The rules could be applied by any system that can move from cell to cell and set it on or off (rock present or absent, flag flying or furled, dot present or erased, bit set or unset) according to the state of the neighboring cells.

It's all about the patterns. The computer (the Turing Machine that runs the simulator program) is a particular pattern of cells in the CA that, when iterated over with the CA rules, interacts with a patterns of cells corresponding to input data, according to program instructions encoded as another pattern of cells, to produce a dynamic output pattern that is the simulation.

If you started with a different patterns of on/off cells, and iterated exactly the same CA rules on it in exactly the same way, instead of a a Turing Machine emulation, you could have Glider Guns, or an emulation of GOL itself, or a self-replicating pattern, etc....

Make that list of instructions as big as you like, but it's not part of a computer. The only computation taking place is in his head.

I beg to differ. What Cueball does and what goes on in his head when he is moving the rocks is no different when running the Turing Machine pattern than when running the Replicator pattern, or the Glider Gun pattern, or any other CA pattern.

Once the initial state is set, he could hand over the task of moving the rocks to someone who knows only the CA rules for moving a rock - who doesn't even know what a CA is, or that it's emulating a Turing Machine, or that the TM is simulating a universe.

The computer is the Turing Machine pattern. It computes when the CA runs. The CA runs when the CA rules for setting a cell are iterated over all the cells.