I thought it was interesting that was Putnam's intent, but everyone who hears about brains in vats likes to argue whether we could actually be envatted. — Marchesk

I suppose if you could show that we can't be brains in a vat even if metaphysical realism is the case then you can argue that realism doesn't entail radical skepticism, and so refute Putnam's argument.

But, yes, from what I've seen people argue on here, there does seem to be a misunderstanding of Putnam's intention. He's not saying that we could be brains in a vat; he's saying that we can't be, and so therefore metaphysical realism is false.

I suppose if you could show that we can't be brains in a vat even if metaphysical realism is the case then you can argue that realism doesn't entail radical skepticism, and so refute Putnam's argument. — Michael

Maybe so, but t's hard to see how metaphysical realism doesn't entail the possibility of some form of radical skepticism, even though I am a realist.

Even if BIVs aren't tenable, a Matrix, Star Trek holodeck or Boltzmann Brain scenario might be. You'd have to show that such a simulation isn't physically possible, and everything I've read leads me to believe that it is possible to compute a convingly realistic world. And if you're born into that world, you wouldn't know what was unrealistic anyway (relative to the actual physical world).

One possible answer to the Fermi Paradox is that advanced aliens are feeding us a simulated universe that looks empty. In that case, we'd only be wrong about the wider universe, not matters inside the solar system (extra solar light would be simulated to fool us).

You'd have to show that such a simulation isn't physically possible — Marchesk

I don't think even that would work, as it could be that the "real" world operates according to different physical laws, and the ones we're familiar with are only the laws of our simulation.

I don't think even that would work, as it could be that the "real" world operates according to different physical laws, and the ones we're familiar with are only the laws of our simulation. — Michael

Like maybe in the real world P=NP, but not our simulation?

But it seems like if you could show that it's impossible to construct a simulation in our world, then the basis for the simulation argument is undermined (because what do you mean by our world being simulated?). However, that sounds related to Putnam's argument against being able to make a radically skeptical realist assertion.

Like maybe in the real world P=NP, but not our simulation?

But it seems like if you could show that it's impossible to construct a simulation in our world, then the basis for the simulation argument is undermined (because what do you mean by our world being simulated?). However, that sounds related to Putnam's argument against being able to make a radically skeptical realist assertion. — Marchesk

I believe the hypothesis trades on logical possibility, not physical possibility.

I suppose the best you could show is that if a simulation is impossible in our world then our world can't be a like-for-like simulation.

I believe the hypothesis trades on logical possibility, not physical possibility. — Michael

Right, but it the hypothesis also needs to make sense. So you could argue that realists cannot coherently say our world is a simulation if building such a simulation in our world is impossible, because we have no way to refer to the states of affairs of the real world in which in our simulation lives.

Right, but it the hypothesis also needs to make sense. — Marchesk

Sure, but making sense doesn't depend on obeying physical laws. The notion of defying gravity à la Superman is coherent, even if not able to be done.

I heard a nice argument about the impossibility of BIVs, daemons, Matrix style arguments etc etc. It points out that the mathematics we've managed to create is completely inconsistent with a universe which, fundamentally, works on floating point (or other possible finitary approximations of) arithmetic for real numbers.

If the simulation was computed with floating point arithmetic, circles would be square-able through straightedge and compass, all numbers would be computable, irrational numbers would eventually having a repeating pattern of digits in their decimal representations and so on. Simulated universe => Computable universe => this. Not this => not computable universe => not simulated universe. One's modus ponens is another's tollens.*

The first objection would be that 'numbers are not in an analogous category to physical objects in the simulated universe' - but why would it matter? They would have to be represented with adequate precision to stop the squaring of the circle - which is infinite precision - and is a constraint placed on the physical operations we are capable of doing with compass and straightedge by their innate, 'virtual', properties. We don't live in the kind of universe where a circle is square-able with compass and straight edge - so we don't live in a simulation with finite precision.

Considering that the computer would have to be infinitely large (physically, not purely mathematically) to store every digit of Pi - to set it to the true value we have in this universe - I think we can rule out that we live in an infinitely large computer, too. So we don't live in a finitary or infinte computer simulation. Eliminate the disjunction and break free in either case (P or not-P is hard to dodge).

That might work against the computer simulation hypothesis, but not the brain-in-a-vat or evil demon hypotheses, as those are concerned with stimulated experiences, not computer programs.

Can you spell out why it wouldn't work for the brain-in-a-vat one? I really don't care about the demon.

Instead of seeing a circle because some circular object has directed light onto our eyes which in turn stimulates the relevant areas of the brain, an evil scientist directly stimulates those parts of the brain. The quality of the experience (and underlying brain activity) is the same even though the immediate causes are different. There's no prima facie reason to suggest that if the scientist is directly stimulating the brain then "circles would be square-able through straightedge and compass".

Although I wonder if your floating point number example even works for the computer simulation. The precision only needs to be high enough to fool the naked human senses. Any results from computer measurements can always be hypothesised to be fabricated by the scientist/demon/simulation.

Although I wonder if your floating point number example even works for the computer simulation. The precision only needs to be high enough to fool the naked human eye. — Michael

No, it needs to be high enough to fool human technology and math. That's why some people have speculated that physics might be able to show we're in a simulation.

So the idea is replace all experiences with exactly equivalent substitutes which come solely from stimulating the brain?

Presumably this is automated to be real time. I can't conceive of a way of doing this, nor do I think it's possible. All that changes is that we are no longer software, but we are being stimulated by software in a manner which produces equivalent lives; in other words, it's a more convoluted way of doing exactly the same thing.

Although I wonder if your floating point number example even works for the computer simulation. The precision only needs to be high enough to fool the naked human eye.

It's more that the 'simulated universe' must have square-able circles in it. This isn't a perceptual property, it's a relational property of circles, squares and the transcendental nature of Pi.

all numbers would be computable, irrational numbers would eventually having a repeating pattern of digits in their decimal representations and so on. — fdrake

Actually, I don't understand this. Assuming we live in the real world, how is it that we determine the digits of Pi? Don't we have some computer running the relevant calculations? So far it hasn't provided a repeating pattern of digits. Why can't a simulated computer run the same calculations?

No, it needs to be high enough to fool human technology and math. — Marchesk

The human technology is part of the simulation, too. I'm not sure what you mean about fooling the math.

So the idea is replace all experiences with exactly equivalent substitutes which come solely from stimulating the brain?

Presumably this is automated to be real time. — fdrake

Well, the brain isn't very fast compared to computers. It takes a quarter of a second or so to think a thought or recognize an object. Responding to a startling sound is much faster (50 milliseconds), but it's still slow compared to computers which can operate on nanosecond time frames.

The human technology is part of the simulation, too. I'm not sure what you mean about fooling the math. — Michael

The fact that we can compute PI to huge numbers of places means the simulation has to be able to do that. And that we can devise physics experiments that can measure the amount of time it takes light to cross the length of an atom means the simulation has to be able to accommodate that.

That requires way, way more compute power than fooling the naked eye.

It isn't a matter of producing the correct digits. Pi is a computable number. It's this property, it is a real consequence of the fact that Pi is transcendental. It literally puts a constraint on what is possible using a compass and straightedge.

Well, the brain isn't very fast compared to computers. It takes a quarter of a second or so to think a thought or recognize an object. Responding to a startling sound is much faster (50 milliseconds), but it's still slow compared to computers which can operate on nanosecond time frames.

The speed we think and act probably puts some bounds on their informational content. But the speed alone tells us nothing about how hard it would be to simulate human experience, or to provide real-time equivalent stimulations to a brain (assuming the brain can indeed be stimulated to produce these things without sensorimotor constraints and the nervous system at large... which is unlikely).

The fact that we can compute PI to huge numbers of places means the simulation has to be able to do that. — Marchesk

So why is that a problem? The simulation runs the same program that we run to calculate the value of Pi.

And that we can devise physics experiments that can measure the amount of time it takes light to cross the length of an atom means the simulation has to be able to accommodate that. — Marchesk

The simulation only needs to simulate what we see. What we see is the device and its human-readable output.

It isn't a matter of producing the correct digits. Pi is a computable number. It's this property, it is a real consequence of the fact that Pi is transcendental. It literally puts a constraint on what is possible using a compass and straightedge. — fdrake

I don't understand why this would be a problem for the simulation. If our computers can calculate Pi without ever repeating digits then the simulation can calculate Pi without ever repeating digits.

The simulation only needs to simulate what we see. What we see is the device and its human-readable output. — Michael

The computer needs to be able to compute the result of any experiment we might think to devise in a convincing fashion. That goes way beyond simply fooling the human visual system.

The speed we think and act probably puts some bounds on their informational content. But the speed alone tells us nothing about how hard it would be to simulate human experience, or to provide real-time equivalent stimulations to a brain (assuming the brain can indeed be stimulated to produce these things without sensorimotor constraints and the nervous system at large... which is unlikely). — fdrake

That might be so for BIVs, but it won't be so for holodecks, since holodecks feed our sensory organs instead of our brains. Imagine the ST universe where a whole civilization lives inside a large holodeck. And that leads to another possible answer to the Fermi Paradox.

All advanced civilizations end up inside simulations, because they're far more appealing than exploring space.

@Michael

I don't understand why this would be a problem for the simulation. If our computers can calculate Pi without ever repeating digits then the simulation can calculate Pi without ever repeating digits.

The computer could produce an arbitrarily accurate approximation of Pi, given sufficient computing time. The computer would not have infinite memory however - that makes it inconceivable, it would have infinite size -, and it requires infinite memory to store all of Pi. Our reality doesn't have an arbitrarily accurate approximation of Pi - this would be a rational number, a fraction - it has Pi in all its delicious transcendental infinite glory. And because Pi is transcendental, people in the real world cannot square a circle. The latter being a physical process.

A finitely sized computer could only store finitely many digits of Pi. This would make Pi rational, so the circle would be square-able; it isn't, so we're not in one of those.

@Marchesk

That might be so for BIVs, but it won't be so for holodecks. Imagine the ST universe where a whole civilization lives inside a large holodeck. And that leads to another possible answer to the Fermi Paradox.

I literally can't imagine what that would be like in any coherent way. I suppose these arguments aren't very good at convincing the unimaginative.

I literally can't imagine what that would be like in any coherent way. I suppose these arguments aren't very good at convincing the unimaginative. — fdrake

Have you watched any of The Next Generation or Voyager? It's not uncommon for some of the crew to spin up a holodeck program that's as sensory rich as the real world, and there be a malfunction where they're trapped in the program and can't exit.

In one, a simulated character who had become aware made a simulation of the entire ship to fool the crew members into thinking they had exited the holodeck.

I'm not a mathematician, so most of this is going over my head. Perhaps you could explain how we can show that a circle isn't square-able, and why this method would produce different results in a simulation?

I've watched NGE and Deep Space Nine. I still can't imagine a holodeck the size of the universe.

Wiki has a decent article on it..

I've watched NGE and Deep Space Nine. I still can't imagine a holodeck the size of the universe. — fdrake

If you ran a planet-sized holodeck, would we be able to know it wasn't actually the size of the universe? We could think we were sending a probe off into deep space, and it's just the program making it look that way. Even in a room-sized holodeck, they're somehow able to move around quite a bit as if they weren't constrained to a room.