Podcast about our universe being a computer simulation

[jenpohl]

I'm not sure if this is the right place to be asking this, but I'm looking for a specific episode of a podcast. I was sure it was RadioLab until I failed at finding such an episode. I remember that it was an interview in which the interviewee argued that it was most likely that our universe was a computer simulation run by a more advanced society. Anyone know which one this was? Thanks!

 

[FattyCatty]

I'm not sure if this is the right place to be asking this, but I'm looking for a specific episode of a podcast. I was sure it was RadioLab until I failed at finding such an episode. I remember that it was an interview in which the interviewee argued that it was most likely that our universe was a computer simulation run by a more advanced society. Anyone know which one this was? Thanks!

This sounds like it talks about what you are describing. There is a link to an interview from 2007 on bloggingheads.com that might be what you heard.

Are You Living In a Computer Simulation?
On this website you can peruse the debate that followed the paper presenting the Simulation argument. The original paper is here, as are popular synopses, scholarly papers commenting on the first paper, and a couple of replies to these comments. simulation-argument.com/

 

[little grey rabbit]

From the link of FattyCat I read this one
http://www.nytimes.com/2007/08/14/science/14tier.html?pagewanted=1&_r=1

It brings to mind Bertrand Russel's first cause argument (or variants of tortoises on the back of something else)

The First-cause Argument

Perhaps the simplest and easiest to understand is the argument of the First Cause. (It is maintained that everything we see in this world has a cause, and as you go back in the chain of causes further and further you must come to a First Cause, and to that First Cause you give the name of God.) That argument, I suppose, does not carry very much weight nowadays, because, in the first place, cause is not quite what it used to be. The philosophers and the men of science have got going on cause, and it has not anything like the vitality it used to have; but, apart from that, you can see that the argument that there must be a First Cause is one that cannot have any validity. I may say that when I was a young man and was debating these questions very seriously in my mind, I for a long time accepted the argument of the First Cause, until one day, at the age of eighteen, I read John Stuart Mill's Autobiography, and I there found this sentence: "My father taught me that the question 'Who made me?' cannot be answered, since it immediately suggests the further question `Who made god?'" That very simple sentence showed me, as I still think, the fallacy in the argument of the First Cause. If everything must have a cause, then God must have a cause. If there can be anything without a cause, it may just as well be the world as God, so that there cannot be any validity in that argument. It is exactly of the same nature as the Hindu's view, that the world rested upon an elephant and the elephant rested upon a tortoise; and when they said, "How about the tortoise?" the Indian said, "Suppose we change the subject." The argument is really no better than that. There is no reason why the world could not have come into being without a cause; nor, on the other hand, is there any reason why it should not have always existed. There is no reason to suppose that the world had a beginning at all. The idea that things must have a beginning is really due to the poverty of our imagination. Therefore, perhaps, I need not waste any more time upon the argument about the First Cause.

What if it turned out that the posthumans were themselves a simulation of the postposthumans?

 

[Doghouse Reilly]

From the link of FattyCat I read this one
http://www.nytimes.com/2007/08/14/science/14tier.html?pagewanted=1&_r=1

It brings to mind Bertrand Russel's first cause argument (or variants of tortoises on the back of something else)



What if it turned out that the posthumans were themselves a simulation of the postposthumans?

That is addressed in one of the links on the guy's Web page.

Reading this theory, it is striking how many completely unreasonable assumptions are made in order to justify speculation of things such as "simulations" of entire conscious entities and realities. Never is it discussed whether such things are possible; in fact, the assumption within the theory is that they are, and that any technological civilization that doesn't become extinct will eventually develop such technology. It's almost as though the author believes a sort of "woo" that tells him that anything is possible as long as we wait long enough and develop the right technology.

These ideas are rather large and unjustified leaps. It negates any relevance all subsequent discussion within the paper might have had for me. It's great as science fiction, but why would there actually be any scientific attention given to a theory predicated on so many unfounded assumptions?

 

[Kevin_Lowe]

I think you need to read the actual article again.

It postulates that either most civilisations never get to be truly advanced or those that do don't run many ancestor simulations (possibly because ancestor simulations are impossible) or that it is likely we are simulated.

In any case in principle I see no reason why simulating being like us isn't possible. We're just chunks of matter, and we can simulate the behaviour of chunks of matter. It's just a matter of putting together the processing power, and the world's processing power is increasing exponentially as we produce more and more chips that are more and more powerful. The only reason I can see to think it's impossible is if you are some kind of dualist who believes we have souls, and souls are silly.

It seems that you are under the mistaken impression that the argument is meant to lead you to the conclusion that option 3 (that we are likely to be simulated) is true. It's not. It's meant to lead you to the conclusion that one of those three options is true but you don't know which, and it's up to you what probability you want to place on each of them.

 

[little grey rabbit]

At the moment I would say I can not see we could simulate something that would be self-conscious.

But I could be wrong here.

 

[Ziggurat]

In any case in principle I see no reason why simulating being like us isn't possible. We're just chunks of matter, and we can simulate the behaviour of chunks of matter.

I'm not sure you appreciate how poorly we actually do this. Simulating physical systems is an exponential problem, meaning that the required computations scale exponentially with the number of objects in the simulation. And when the terms in your exponent are themselves > 10²⁰, well... no, it isn't clear that such a thing is physically possible.

It's just a matter of putting together the processing power, and the world's processing power is increasing exponentially as we produce more and more chips that are more and more powerful.

There's no reason to think that exponential growth will continue indefinitely, and good reason to think it won't.

But suppose that we are simulated. If that's the case, we cannot assume that the true physical reality that is simulating us obeys the same physical laws as the "reality" we are simulated in. It certainly wouldn't need to, and if I had the computing power to simulate an entire reality, simulating alternate realities would seem a whole lot more interesting to me. And once we allow for different physical laws (as we must), all bets are off.

But at the end of the day, the possibility that we're inside a simulated reality is basically no different than Solipsism, a philosophy that gets to be boring very quickly.

 

[Kevin_Lowe]

I'm not sure you appreciate how poorly we actually do this. Simulating physical systems is an exponential problem, meaning that the required computations scale exponentially with the number of objects in the simulation. And when the terms in your exponent are themselves > 10²⁰, well... no, it isn't clear that such a thing is physically possible.

For simulation purposes I'm sure you don't need to track every atom, unless some scientist is looking at that particular atom that particular second.

There's no reason to think that exponential growth will continue indefinitely, and good reason to think it won't.

Of course it won't - this is a finite universe. However we've got no idea whether the hard limits on computational power in our solar system will be hit before or after ancestor simulations are possible.

But suppose that we are simulated. If that's the case, we cannot assume that the true physical reality that is simulating us obeys the same physical laws as the "reality" we are simulated in. It certainly wouldn't need to, and if I had the computing power to simulate an entire reality, simulating alternate realities would seem a whole lot more interesting to me. And once we allow for different physical laws (as we must), all bets are off.

This is wooly thinking. You are saying "The simulation argument doesn't work, because we could be simulated by a civilization in a universe very much unlike ours". But if we are being simulated by such a civilization, then we're simulations. So if the only way you can poke a hole in it is to assume that (3) is the case, I don't think Bostrom has any cause to be worried.

But at the end of the day, the possibility that we're inside a simulated reality is basically no different than Solipsism, a philosophy that gets to be boring very quickly.

[/quote]

On the contrary, the simulation argument involves looking at the universe we appear to live in, not merely speculating that it might be a delusion.

If it turns out that something or other does prevent lots of civilizations from reaching the level of being able to simulate their ancestors, that makes (1) more likely. If it does turn out that ancestor simulations are computationally impossible or that there's something much more fun to do with all that computing power, (2) becomes more likely. If it turns out there is no cosmic filter killing off advanced civilizations or stunting their development, and ancestor simulations turn out to be feasible, that makes (3) more likely.

 

[Sideroxylon]

From the link of FattyCat I read this one
http://www.nytimes.com/2007/08/14/science/14tier.html?pagewanted=1&_r=1

It brings to mind Bertrand Russel's first cause argument (or variants of tortoises on the back of something else)



What if it turned out that the posthumans were themselves a simulation of the postposthumans?

Its posthumans all the way down.

 

[Ziggurat]

For simulation purposes I'm sure you don't need to track every atom

How do you know how much you need to track? Sometimes crude approximations work, but in many, many, MANY cases, those crude approximations break down completely. How do you know when an approximation is good enough? Well, in general, you DON'T, unless you try it without the approximation.

And I really don't think you understand the sort of complexity we're talking about here. There are about 10¹¹ neurons in the human body. Even if you want to ignore the details on each cell and treat them as macroscopic objects, right there you're talking about 10²² interactions you need to keep track of. That's an absurdly large number, but it's STILL inadequate. Because you can't treat neurons as simple objects, because they aren't. They have all sorts of lovely interactions with things like electrolyte levels, blood oxygen content, neurotransmitters, and drugs, all of which involve quantum chemistry if you actually want to get them right. And you need to get them right if you want the simulation to be physically accurate, because they have quite observable effects. And of course, we're still just talking about one brain. Not even the body it's in. Not even the world that surrounds it, which is feeding it information. Not even all the other people and other organisms, around them, all of which can only be treated with even a semblance of accuracy by doing calculations that are unimaginably complex.

It's not gonna happen. It can't happen. Computational power will hit a wall long before it does.

 

[Kevin_Lowe]

How do you know how much you need to track? Sometimes crude approximations work, but in many, many, MANY cases, those crude approximations break down completely. How do you know when an approximation is good enough? Well, in general, you DON'T, unless you try it without the approximation.

And I really don't think you understand the sort of complexity we're talking about here. There are about 10¹¹ neurons in the human body. Even if you want to ignore the details on each cell and treat them as macroscopic objects, right there you're talking about 10²² interactions you need to keep track of. That's an absurdly large number, but it's STILL inadequate. Because you can't treat neurons as simple objects, because they aren't. They have all sorts of lovely interactions with things like electrolyte levels, blood oxygen content, neurotransmitters, and drugs, all of which involve quantum chemistry if you actually want to get them right. And you need to get them right if you want the simulation to be physically accurate, because they have quite observable effects. And of course, we're still just talking about one brain. Not even the body it's in. Not even the world that surrounds it, which is feeding it information. Not even all the other people and other organisms, around them, all of which can only be treated with even a semblance of accuracy by doing calculations that are unimaginably complex.

It's not gonna happen. It can't happen. Computational power will hit a wall long before it does.

This is a silly argument - you don't need to simulate every atom in a 1980s-era pocket calculator to run a software simulation of the important aspects of a 1980s-era pocket calculator.

This web page purports that if we can keep the current rate of exponential growth up then we'll have computers with equivalent processing power to the human brain in 2020, and after that it's "just" a software problem. I see no reason to believe we are going to hit hard limits to computational growth in the next few decades.

 

[Ziggurat]

This is a silly argument - you don't need to simulate every atom in a 1980s-era pocket calculator to run a software simulation of the important aspects of a 1980s-era pocket calculator.

What are the important aspects of a virus? What are the important aspects of the various strains of bacteria that inhabit my gut? What are the important aspects of protein folding?

This web page[/url] purports that if we can keep the current rate of exponential growth up then we'll have computers with equivalent processing power to the human brain in 2020, and after that it's "just" a software problem.

Not even close. It's not good enough to make an artificial brain that can do the same thing that a real brain can do, it needs to respond the same way too. And that means responses not only to nerve stimuli, but also chemical stimuli. As in, give it a drug, and it needs to respond appropriately to that drug. Do you have any idea how hard a problem that is? Hell, we can't reliably fold single proteins with a computer yet. We cannot calculate how a drug will interact with receptor proteins or ion channels or whatever. Maybe we'll begin to by 2020, but the problem of simulating a physical brain is many orders of magnitude harder than the problem of simulating an idealized one. But only the former will suffice. And even then, it's not enough. The environment must be simulated too, and the environment includes effects which cannot easily be simplified either.

I see no reason to believe we are going to hit hard limits to computational growth in the next few decades.

A few decades isn't enough. Not nearly enough. Not when we can't even fold a single protein reliably.

 

[Kevin_Lowe]

What are the important aspects of a virus? What are the important aspects of the various strains of bacteria that inhabit my gut? What are the important aspects of protein folding?

Rhetorical questions aren't a substitute for an argument.

Not even close. It's not good enough to make an artificial brain that can do the same thing that a real brain can do, it needs to respond the same way too. And that means responses not only to nerve stimuli, but also chemical stimuli. As in, give it a drug, and it needs to respond appropriately to that drug. Do you have any idea how hard a problem that is?

Worst case scenario, you pause the simulation for a minute or two to create a general solution for the effects of alcohol, or nicotine, or whatever that will be good enough to pass muster.

Hell, we can't reliably fold single proteins with a computer yet. We cannot calculate how a drug will interact with receptor proteins or ion channels or whatever. Maybe we'll begin to by 2020, but the problem of simulating a physical brain is many orders of magnitude harder than the problem of simulating an idealized one. But only the former will suffice. And even then, it's not enough. The environment must be simulated too, and the environment includes effects which cannot easily be simplified either.

Repeating this same silly argument doesn't make it any less silly. You don't need to simulate a pool cue and an eight ball at the quark level to simulate a pool cue and an eight ball well enough to fool human-like senses, and the same applies to everything else you are going to come up with.

A few decades isn't enough. Not nearly enough. Not when we can't even fold a single protein reliably.

You still don't get it. You don't need to simulate the universe we think we live in, you just need to simulate one that will pass for it to fallible simulated humans. 99.99...9% of the time the exact state of the quarks in my arse cheeks don't matter for that purpose and hence would not need to be simulated in exact detail.

 

[bokonon]

It postulates that either most civilisations never get to be truly advanced or those that do don't run many ancestor simulations (possibly because ancestor simulations are impossible) or that it is likely we are simulated.

All the civilizations for which we have any evidence are nowhere near advanced anough to run "self-aware" simulations.

Suppose we make the unwarranted assumption that such a civilization exists. We must assume that not only is such technology possible, but that it has been developed, and the civilization developing it considers running such simulations a worthwhile exercise. This, too, seems highly unlikely.

As far as I'm concerned, all this "brain in a jar" speculation is pointless magical thinking, no different than earnestly debating how many angels can dance on the head of a pin. It's religion for those who worship at the altar of technology. With the right special effects, it can make Keanu Reaves simulate entertainment for an hour or two, but there's no justification whatsoever for taking it seriously.

 

[Kevin_Lowe]

All the civilizations for which we have any evidence are nowhere near advanced anough to run "self-aware" simulations.

Suppose we make the unwarranted assumption that such a civilization exists. We must assume that not only is such technology possible, but that it has been developed, and the civilization developing it considers running such simulations a worthwhile exercise. This, too, seems highly unlikely.

I think you too need to read what Bostrom actually wrote again a few more times until you understand it.

He's arguing that either most civilisations never get to be truly advanced or those that do don't run many ancestor simulations (possibly because ancestor simulations are impossible) or that it is likely we are simulated. He isn't advocating any one of those three options in particular.

The argument doesn't rely on any of the unwarranted assumptions you list.

 

[bokonon]

I think you too need to read what Bostrom actually wrote again a few more times until you understand it.

He's arguing that either most civilisations never get to be truly advanced or those that do don't run many ancestor simulations (possibly because ancestor simulations are impossible) or that it is likely we are simulated. He isn't advocating any one of those three options in particular.

The argument doesn't rely on any of the unwarranted assumptions you list.

Nonsense. Let's take his "arguments" in order.

1. Most civilizations never get to be "truly" advanced.

Well, there's a bold proposition. Obviously, all of the civilizations of which we have knowledge fall into this category. I'd say we can stop right here. But does he? He does not.

2. Those civilizations that do get to be "truly" advanced don't run such simulations.

Argument 2 assumes argument 1 is false. It therefore assumes that there may be some civilizations which are "truly" advanced. We have no evidence that such civilizations exist, so argument 2 relies on the UNWARRANTED ASSUMPTION that such civilizations may exist.

This is no different in principle than the unwarranted assumption that deities inhabit the heights of Mount Olympus. Either they do or they don't. If they do, maybe they don't interfere in the lives of mortals. In other words, a fitting premise for speculative fiction, but hardly worthy of serious consideration.

3. It is likely that we are simulated.

Depends on the unwarranted assumption that both argument 1 and argument 2 are false.

Option 3 is only plausible in the trivial case that the hardware for the simulation is all the matter in the universe, and the simulation is powered by all the energy in the universe. In that case, either there is a "creator" which designed the simulation and started it running, or there isn't. Again, a valid starting point for theological fantasy (or, taking just the world as the simulation, "Hitchhiker's Guide to the Galaxy"), but to me it seems silly to spend much time seriously contemplating such possibilities.

Maybe mankind one day invents time machines, and we are presently surrounded by undetectable travelers from the future who either do or do not shape the course of our lives. Maybe angels are real, and prayer convinces them to alter natural laws to favor outcomes we desire. Maybe there is an alternative universe in which we all snort cheese through prehensile tails. Maybe not. I'm not arguing one way or the other.

His "three arguments" are only interesting if there is a possibility that the third option is true. Civilizations don't become that advanced? Boring. Civilizations become that advanced, but they have better things to do? Boring, and completely speculative, since we have no evidence that civilizations become that advanced. We're simulated? Interesting, as the premise for fantasy and fiction. It isn't, in my opinion, a hypothesis that deserves serious consideration.

 

[Kevin_Lowe]

Nonsense. Let's take his "arguments" in order.

1. Most civilizations never get to be "truly" advanced.

Well, there's a bold proposition. Obviously, all of the civilizations of which we have knowledge fall into this category. I'd say we can stop right here. But does he? He does not.

If you're convinced based on a sample of one not-yet-mature civilization that the majority of civilizations never develop substantially further than we are now, then indeed you can skip the rest of the argument.

You also have some very strange thought processes I do not pretend to follow.

2. Those civilizations that do get to be "truly" advanced don't run such simulations.

Argument 2 assumes argument 1 is false. It therefore assumes that there may be some civilizations which are "truly" advanced. We have no evidence that such civilizations exist, so argument 2 relies on the UNWARRANTED ASSUMPTION that such civilizations may exist.

You're still missing the point. Of course (2) assumes (1) is false, because (1), (2) and (3) are meant to be mutually exclusive.

In any case, in a universe that's as old as ours there have been billions of years for other civilizations to develop. To my mind the belief that such civilizations may exist is the only sensible one, and the belief that such civilizations definitely do not exist is just bizarre.

This is no different in principle than the unwarranted assumption that deities inhabit the heights of Mount Olympus. Either they do or they don't. If they do, maybe they don't interfere in the lives of mortals. In other words, a fitting premise for speculative fiction, but hardly worthy of serious consideration.

That's a profoundly stupid comparison in my view.

3. It is likely that we are simulated.

Depends on the unwarranted assumption that both argument 1 and argument 2 are false.

That's the whole point. (1), (2) and (3) are mutually exclusive. This is not an error that you have cleverly discerned, it's the point of the argument.

Option 3 is only plausible in the trivial case that the hardware for the simulation is all the matter in the universe, and the simulation is powered by all the energy in the universe. In that case, either there is a "creator" which designed the simulation and started it running, or there isn't. Again, a valid starting point for theological fantasy (or, taking just the world as the simulation, "Hitchhiker's Guide to the Galaxy"), but to me it seems silly to spend much time seriously contemplating such possibilities.

No.

His "three arguments" are only interesting if there is a possibility that the third option is true.

No. (1) is if anything more interesting than (3), since we are rapidly approaching the point at which we will have machines capable of simulating a brain in a vat, and shortly afterwards we will have machines capable of simulating that plus some kind of environment for the simulated brain.

If something's going to bring technological progress to a halt one way or another in the next few decades, which is what it will take, then I want to know about it.