Why are almost all my quotes labeled "keystone"?

[The GS] can't be simulating itself, you just agreed with that. — fishfry

I said it couldn't simulate itself exactly. I didn't say it couldn't simulate itself.

Faster does not help when it comes to computation.

It is necessary for a VR, but not a simulation, all of which is pointed out in my topic. It's why a sim can be done with pencil/paper and a VR cannot. Still, Bostrom needs a fast computer because a simulation with paper and such would have humanity go extinct before a fraction of a second was simulated. Bostrom is not making an 'in principle' argument.

Going faster can never let you compute more things than you could with pencil and paper. If going faster makes a difference, then the difference is not computational. It's something else.

True only in principle. In reality, each number written on a paper will likely rot away before it is needed for the next step. The guy with the pencil will die, as will all of humanity. So will the superfast computer (it cannot run forever in practice), but it will have gotten a lot further than the pencil team, and a lot further than any TM, however pimped out you make it.

That's a different kind of computability: the ability to get it done before the demise of the thing doing the computing.

You agree with me on this point then, am I correct?

I agree with all your points on the definition of computability, but I wasn't talking about that.

This was about whether my mind somehow extends to Ms. Pac-Man's. I think it's an important point, not just semantics.

OK. I seem to be blowing it off to semantics, and I made MsPM an extension of me, not an extension of my mind. I consider myself to be conscious, not just a body that contains something that is.

It's a thousand percent different. It's apples and rutabagas. A simulation of gravity is the execution of an approximate mathematical model.

Bostrom's view is that a sim of a person is also the execution of an approximate mathematical model. That this conflicts with your opinion means that your opinion is incompatible with what Bostrom hypothesizes.

The GS's simulation of us is exact. We ARE the simulation. This seems to be a real point of difference, not just semantics.

If you mean that the thing simualted (us) is exactly the same as us, that is tautologically true, yes. But I'm saying that the simulated 'us' cannot be an exact simulation of a person in the GS world.

Then what is the thing being simulated?

Yet again, the thing being simulated is 'ancestral history', whatever that means.

You mean there's a real me

Bostrom does not suggest that there is or ever was a real fishfry in the GS world. You are part of the simulation, and that's it. The history being simulated is quite different than the one that actually happened way in the past history of the GS world, although the initial state of the simulation presumably had similarities to some actual past state of the GS history. Bostrom gives no indication of when this initial state was likely placed. Last Tuesday? A minute ago? 50000 years ago when humanity just started looking like us?

You are misconstruing what Bostrom and other simulationists believe, then. They're not saying we're an approximation. They're saying our exact reality is being instantiated by a computer.

Again, tautologically true. But our reality is the causal result of an approximation of some past GS state.

Why would anyone run an ancestor simulation? We don't, why should our future selves?

Apparently 'because they can' and we don't because we can't. But visionaries have always had a lot of trouble guessing what purposes would be served by future high computing capacity. Anyway, I don't buy that reasoning because it's only there because the hypothesis needs it to hold any water.

I've learned a lot about MOND and dark matter from her.

And she exquisitely tore apart a lot of the woo surrounding the delayed choice quantum eraser since that experiment is so often billed as an example of reverse causality. The one I tore apart had to do with general relativity, which I don't even know that well, but I know enough to show the assertions in the video to be bunk.

I have just been made aware, via flannel jesus, that an LLM has learned to play chess by training on nothing more than the records of games in standard chess notation.

That works great for opening, perhaps for 20 moves even. But eventually it has to get to a position that it hasn't seen in its training data, and then what? It can't just auto-complete with more text, since the text given would likely not be a legal move. So I'd like to see an article about how it proceeds from a middle-game. Turns out that the LLM is often more capable than I give it credit for. Scary.

But eventually it has to get to a position that it hasn't seen in its training data, and then what? — noAxioms

And then it continues to make (usually) legal moves which are approximately as good as its general skill level predicts they should be.

https://adamkarvonen.github.io/machine_learning/2024/01/03/chess-world-models.html

I also checked if it was playing unique games not found in its training dataset. There are often allegations that LLMs just memorize such a wide swath of the internet that they appear to generalize. Because I had access to the training dataset, I could easily examine this question. In a random sample of 100 games, every game was unique and not found in the training dataset by the 10th turn (20 total moves). This should be unsurprising considering that there are more possible games of chess than atoms in the universe.

If this world is simulated, the "real" world must be very like this one — Ludwig V

I'm unsure that's true. The fine-grained nature of the world we live it might just be a function of adaptive creative algorithms which feed off of past events, in the simulation. This would also explain the wildly increasing complexity across time.

The fine-grained nature of the world we live it might just be a function of adaptive creative algorithms which feed off of past events, in the simulation. — AmadeusD

I guess that's so. But that would mean that the simulation is a reality of its own, independently of the "real" reality. (As a story has its own logic, even though it is just a story) Still, the algorithms are part of reality - they are not simulated, are they? - they wouldn't really be algorithms if they were simulated. So the simulaton may be different from the real world in all sorts of ways, but it needs to be built from and in the real world.

I said it couldn't simulate itself exactly. I didn't say it couldn't simulate itself. — noAxioms

Well yes, by my definitions "couldn't simulate exactly" is synonymous with couldn't simulate.

Again, we have this ongoing equivocation of the word simulation. I agree with you that when I program my computer to simulate gravity or the weather, the simulation is not exact. It's an approximation.

But when the GS simulates my consciousness and the experience of my senses, that is exact. There is no other thing "being" simulated. The simulation is the only version of reality.

But you claim Bostrom has already considered that point, and I am not in a position to disagree. So I could be wrong.

Faster does not help when it comes to computation.
It is necessary for a VR, — noAxioms

Then whatever it's doing is not computational. If you execute Euclid's algorithm faster, it is still Euclid's algorithm and has no capabilities (other than working faster) than it did before. It does not acquire more side effects or epiphenomena or "emergences" like consciousness or realism.

This point is essential. And it's not deep. It's just a restatement of the definition of computability, as opposed to complexity, where execution speed matters a lot.

but not a simulation, all of which is pointed out in my topic. It's why a sim can be done with pencil/paper and a VR cannot. — noAxioms

If that is true, then VR is not computational. Because -- by the definition of computability -- speed doesn't matter. If a supercomputer computes Euclid's algorithm, and when you run the same code using pencil and paper it doesn't compute Euclid's algorithm, then the supercomputer did not compute Euclid's algorithm.

This is so fundamental to computing that it must be agreed to. If running an algorithm fast does something that running the same algorithm slowly doesn't do, then whatever it's doing is not computational. Else it could do the same computation slowly. That's the definition.

Jumping into a faster complexity class does not let you compute more things than you could before.

Still, Bostrom needs a fast computer because a simulation with paper and such would have humanity go extinct before a fraction of a second was simulated. Bostrom is not making an 'in principle' argument. — noAxioms

What does running an algorithm fast do that running the same algorithm slowly doesn't? You need to explain this clearly please so that I can understand.

Going faster can never let you compute more things than you could with pencil and paper. If going faster makes a difference, then the difference is not computational. It's something else.
True only in principle. — noAxioms

True by the very definition of computation.

In reality, each number written on a paper will likely rot away before it is needed for the next step. The guy with the pencil will die, as will all of humanity. So will the superfast computer (it cannot run forever in practice), but it will have gotten a lot further than the pencil team, and a lot further than any TM, however pimped out you make it. — noAxioms

Even if I agree 100%, the definition of computability specifically ignores matters of time, space, energy, and resources.

That's a different kind of computability: the ability to get it done before the demise of the thing doing the computing. — noAxioms

No. It's not. It's still computability. It's a different complexity class at best. Those are the definitions. I did not make them up. A function is computable if there's a TM that computes it. Time doesn't matter.

I agree with all your points on the definition of computability, but I wasn't talking about that. — noAxioms

Ok you agree. That's good. So if I write some code, and when I run it slowly it computes Euclid's algorithm; and when I run it fast, it computes Euclid's algorithm and whistles Dixie; then by the definition of computability, which you have now agreed to, whistling Dixie is not a computable function. It it were, the slow algorithm would get the same output as the fast one.

That's the only point I'm making. But it's important, because you claim that running the algorithm fast makes a qualitative difference. And I am pointing out that the difference, whatever it is, can not be a computation alone. It's doing something extra.

.

.
OK. I seem to be blowing it off to semantics, and I made MsPM an extension of me, not an extension of my mind. I consider myself to be conscious, not just a body that contains something that is. — noAxioms

Yes you are, MsPM isn't. And it's not just semantics. Ms. PM does not inherit your humanity, sentience, qualia, or experiences.

Bostrom's view is that a sim of a person is also the execution of an approximate mathematical model. That this conflicts with your opinion means that your opinion is incompatible with what Bostrom hypothesizes. — noAxioms

Wait. There's an abstract mathematical model of a human and any particular sim is only an approximate instance? Is that what you are saying Bostrom is saying? That's more like Tegmark, that we're all mathematical structures.

If you mean that the thing simualted (us) is exactly the same as us, that is tautologically true, yes. But I'm saying that the simulated 'us' cannot be an exact simulation of a person in the GS world. — noAxioms

So there's a simulation of a person AND there's a real person being simulated? Now you have TWO mysteries instead of one. I'm a simulation and there's a real me above that? I don't believe that.

Yet again, the thing being simulated is 'ancestral history', whatever that means. — noAxioms

I'm afraid I am not able to process this. Every simulation entity S has a "real" counterpart S' that's the thing being simulated? This is not a coherent ontology.

Bostrom does not suggest that there is or ever was a real fishfry in the GS world. You are part of the simulation, and that's it. — noAxioms

You have just finished telling me that I am a simulation. For sake of discussion I accept that.

Then you tell me that I'm only an approximation of a real person.

And NOW you tell me that there is no real person.

I don't follow your chain of exposition here.

The history being simulated is quite different than the one that actually happened way in the past history of the GS world, although the initial state of the simulation presumably had similarities to some actual past state of the GS history. Bostrom gives no indication of when this initial state was likely placed. Last Tuesday? A minute ago? 50000 years ago when humanity just started looking like us? — noAxioms

I'll stop responding point by point here because I no longer accept the coherence of the thesis being proposed. I'm a sim fishfry and there's a "real" entity fishfry who's being simulated, but who isn't reall there. My parser has stopped here, I need context and explanation before I can go on.

You are misconstruing what Bostrom and other simulationists believe, then. They're not saying we're an approximation. They're saying our exact reality is being instantiated by a computer.
Again, tautologically true. But our reality is the causal result of an approximation of some past GS state. — noAxioms

Ditto, will defer comment. My brain threw a "Makes No Sense" exception.

Apparently 'because they can' and we don't because we can't. But visionaries have always had a lot of trouble guessing what purposes would be served by future high computing capacity. Anyway, I don't buy that reasoning because it's only there because the hypothesis needs it to hold any water. — noAxioms

Ok. This was about ancestor simulation. I've always thought that was a weak part of the argument. What we do is invent video games that use different physics and are nothing like us at all. We don't run ancestor simulations. "Ok I'll be James Madison and you be Dolly."

And she exquisitely tore apart a lot of the woo surrounding the delayed choice quantum eraser since that experiment is so often billed as an example of reverse causality. The one I tore apart had to do with general relativity, which I don't even know that well, but I know enough to show the assertions in the video to be bunk. — noAxioms

I've seen videos where someone debunks every other relativity video on the Internet, from the Science Asylum to the annoying (my personal opinion) British PBS guy, to the guy from Fermilab, Don Lincoln, and a few others. I don't know enough physics to disagree with any of them.

That works great for opening, perhaps for 20 moves even. But eventually it has to get to a position that it hasn't seen in its training data, and then what? — noAxioms

That's the astonishing thing. It plays pretty well even then, in games whose length exceeds the length of any of its training data. That was the scary and surprising part.

It can't just auto-complete with more text, since the text given would likely not be a legal move. So I'd like to see an article about how it proceeds from a middle-game. Turns out that the LLM is often more capable than I give it credit for. Scary. — noAxioms

I think the article in @flannel jesus's article(s) go over some of that, I haven't read it all yet.

But that would mean that the simulation is a reality of its own, independently of the "real" reality. — Ludwig V

I want to agree and disagree with this. By most definitions of 'reality', yes, a simulated world would be a reality of its own, but it being called a simulation is an explicit admission of it being dependent on the deeper reality running the simulation, just like saying 'God created the universe' makes the explicit relation of the universe being dependent on the god. Neither case is that of a 'universe on its own'.

Well yes, by my definitions "couldn't simulate exactly" is synonymous with couldn't simulate.

Again, we have this ongoing equivocation of the word simulation. I agree with you that when I program my computer to simulate gravity or the weather, the simulation is not exact. It's an approximation. — fishfry

The above two comments seem to contradict each other. By your definition, a simulation isn't one unless it is exact, and then you give examples of simulations that are not exact.

But when the GS simulates my consciousness and the experience of my senses, that is exact.

You also said that consciousness is not computational, and therefore the GS cannot simulate via computation, a conscious thing. That puts you into a position to not dictate whether or not those holding a different opinion would say that exactness is required or not.

Then whatever [VR] is doing is not computational.
...
If that is true, then VR is not computational.

So pacman does not involve computation. Hmm....

If you execute Euclid's algorithm faster, it is still Euclid's algorithm and has no capabilities (other than working faster) than it did before. It does not acquire more side effects or epiphenomena or "emergences" like consciousness or realism.

Agree, but I was talking about VR when I said that the rate of computation is essential. None of your examples above are VR examples.

Other examples is any other kind of real-time programming such as a self driving car. A car cannot function if its processing uses paper and pencil. A certain minimal rate of computation is required, or the task cannot be done. Computing slowly isn't enough if you take 3 months to see the stop sign. You seem to assert that what a self-driving car does therefore cannot be computation, but it very much is.

What does running an algorithm fast do that running the same algorithm slowly doesn't?

I already told you: It gets it done before the computer ceases computing. A human with a pencil lives maybe 50 years (with the pencil) and accomplishes what a computer can do in under a second, and computers tend to last longer than a second before they fail. A computer can come up with an answer while the answer is still needed. In a simulation, there are no deadlines to meet (except getting something done before the computer fails), but in any kind of real-time programming, it must be completed before the output is needed by the consumer of that output.

Even if I agree 100%, the definition of computability specifically ignores matters of time, space, energy, and resources

Ok you agree. That's good. So if I write some code, and when I run it slowly it computes Euclid's algorithm; and when I run it fast, it computes Euclid's algorithm and whistles Dixie; then by the definition of computability, which you have now agreed to, whistling Dixie is not a computable function. It it were, the slow algorithm would get the same output as the fast one.

If it whistles Dixie, it is computing something different. Both should have identical output. Euclid's algorithm isn't a real-time task.

That's the only point I'm making. But it's important, because you claim that running the algorithm fast makes a qualitative difference.

Only to a real-time task, and none of your examples are one.
.

Wait. There's an abstract mathematical model of a human and any particular sim is only an approximate instance?

There's a model of physics, and any sim is only a computable approximation of that. Bostrom says that a human is a product of physics, and thus can be functionally simulated given a sufficient level of detail, which is still classical.

That's more like Tegmark, that we're all mathematical structures.

Same model, different supervenience, if I get my terminology straight.

So there's a simulation of a person AND there's a real person being simulated?

I don't know what you think it means for a real person to be simulated. Bostrom suggests a sim of ancestral history, which means that random new people get born, and these people do not in any way correspond to actual people that might have existed in the history of the GS. Much depends on what period of history they choose for their initial state.

Now you have TWO mysteries instead of one. I'm a simulation and there's a real me above that? I don't believe that.

That would be something other than 'ancestral history'. You say take a molecular scan of a real person, create a sim model of that exact arrangement of matter, put it in a small environment, and see what it does. That's far more likely than this 'ancestral' thing, but it also would be trivial for the simulated person to realize he's not the original since he's been put in this tiny bounded space, a sort of jail, when he remembers getting into the scanning machine.

Now according to your stated beliefs, that simulation wouldn't work. It is computational and you say a person isn't, so the simulated thing would not be functional at any level of detail.

Then you tell me that I'm only an approximation of a real person.

No, I did not suggest there needs to be a 2nd fishfry that is 'real'. Ancestral history simulations certainly don't produce simulated people that correspond to people in the GS world.

I no longer accept the coherence of the thesis being proposed. I'm a sim fishfry and there's a "real" entity fishfry who's being simulated, but who isn't reall there.

No, not two of you. Bostrom's sim hypothesis would have all of us being in one large simulation, and no real fishfry in the GS world. I apologize if something I posted led you to conclude that I was suggesting otherwise.
What is approximated is the physics. I can simulate planetary motion by modeling Earth as a point mass. That's a super-trivial approximation of Earth that works for seeing where it is 100 years from now, but it needs more detail if you say want to see which way the planet is facing in 100 years.

What we do is invent video games that use different physics and are nothing like us at all.

Yes, but over time, many video games keep getting closer and closer to the sort of reality we'
re used to. Not all of them. Some are still total fiction with deliberate fiction physics, if they have physics at all. They're also video games, which makes them VR, not simulations.

I've seen videos where someone debunks every other relativity video on the Internet

It's low hanging fruit to debunk various videos. There is indeed whole sites dedicated to debunking relativity in all possible ways, and it is a interesting exercise to find the fallacious reasoning in every one of the arguments. And I do know enough physics to do it to almost all of them.
The delayed choice quantum eraser isn't really an experiment having anything to do with relativity theory.

That's the astonishing thing. It plays pretty well even then, in games whose length exceeds the length of any of its training data..

News to me as well. It seems to require at least some level of what would qualify as 'understanding'.

By most definitions of 'reality', yes, a simulated world would be a reality of its own, but it being called a simulation is an explicit admission of it being dependent on the deeper reality running the simulation, just like saying 'God created the universe' makes the explicit relation of the universe being dependent on the god. — noAxioms

The meaning of "dependent" is context-dependent. The dependence of a simulation on its deeper reality is quite different from the dependence of a created object on it creator. If one thinks of some entity having created a universe, the implication is that the creation exists in its own right. Insofar as a simulation is a reality of its own in the way that a story is a reality of its own, it will not exist in its own right and remains under the control of the story-teller, even though it may have an internal logic that is not the same as the logic of reality.

I'm a sim fishfry and there's a "real" entity fishfry who's being simulated, but who isn't really there. — fishfry

..... unless you think of fishfry as an avatar. On the other hand, if I am a simulation that is not aware of the fact, I must be able to act and react in my world. In that case, I am not a simulation of anything.

Bostrom suggests a sim of ancestral history, which means that random new people get born, and these people do not in any way correspond to actual people that might have existed in the history of the GS. — noAxioms

I can think of models of the weather system that are used to predict the weather. They can be called simulations. They remain quite distinct from the actual weather. There are neither storms, nor rain, nor sunshine inside the computer. Yet the point of the exercise is that it remain as close as possible to what actually happens/-ed. (I can't imagine what the point of ancestral simulations would be, if not that.)
Once you suppose that the simulations are conscious (perhaps per impossibile), those people are not simulations. They might be clones (even if they are not clones in the classical sense). But they would be actual people, perfectly capable of behaving differently from their "originals".

The history being simulated is quite different than the one that actually happened way in the past history of the GS world, although the initial state of the simulation presumably had similarities to some actual past state of the GS history. — noAxioms

The point of the simulations would be lost if real people capable in their own right of acting and reacting in their world. It wouldn't even be a way of running an alternative history. Or is there some other point at stake here, that I've failed to grasp?

But that would mean that the simulation is a reality of its own, independently of the "real" reality. — Ludwig V

I think is entirely dependent on S's use of the word 'reality'. The way i use, it is expressly apt to delineate between a simulated, and a non-simulated 'reality'. The former would not actually be able to come undert this label. I think the term is complete incoherent if it isn't doing this job. I realise others use it differently.

they wouldn't really be algorithms if they were simulated. — Ludwig V

I think algorithms are simulations of behavioural matrices. I can't understand the above claim, really.

but it needs to be built from and in the real world. — Ludwig V

Which is why the previous two claims seem fishy to me. It is patently obvious they are not analogous or parallel scenarios to be in, for any given S.

The above two comments seem to contradict each other. By your definition, a simulation isn't one unless it is exact, and then you give examples of simulations that are not exact. — noAxioms

That's disingenuous. A simulation of the world, as in simulation theory, must be exact.

A simulation of gravity is necessarily an approximation.

That's because the two usages of simulation are being equivocated. Which is why I say Bostrom-like simulation should be called instantiation.

It's not me using the same word two different ways. It's me trying to EXPLAIN that OTHER people are using the same word for two very different things. Leading directly to so much confusion around the subject.

You also said that consciousness is not computational, and therefore the GS cannot simulate via computation, a conscious thing. — noAxioms

Correct. I say that under the hypothesis that we are a simulation, the simulation -- more properly called an instantiation -- is exact.

But I also maintain that the hypothesis is false. So there's no contradiction.

If 2 + 2 = 5 than I am the Pope. The implication is true, and the premise is false. That's exactly what I said.

That puts you into a position to not dictate whether or not those holding a different opinion would say that exactness is required or not. — noAxioms

So I am not in a position to dictated whether or not 2 + 2 = 5 because I hold that the proposition is false? You are not making sense.

So pacman does not involve computation. Hmm....[/quoet]

Pacman ONLY involves computation. No sentience is involved.

The points you're making in this post are trivial and wrong, not up to your usual standards.
— noAxioms

the rate of computation is essential. None of your examples above are VR examples. — noAxioms

ANY phenomenon whatsoever that depends on execution speed is not (only) computational. That's not me saying that, and it is not a deep or clever point. It's merely the definition of computation.

Other examples is any other kind of real-time programming such as a self driving car. A car cannot function if its processing uses paper and pencil. A certain minimal rate of computation is required, or the task cannot be done. Computing slowly isn't enough if you take 3 months to see the stop sign. You seem to assert that what a self-driving car does therefore cannot be computation, but it very much is. — noAxioms

That shows that complexity theory is important in self-driving cars, not just computability theory.

You have already agreed with me on these definitions many posts ago, so I don't understand why you're pretending not to remember agreeing with me.

What does running an algorithm fast do that running the same algorithm slowly doesn't?
I already told you: It gets it done before the computer ceases computing. A human with a pencil lives maybe 50 years (with the pencil) and accomplishes what a computer can do in under a second, and computers tend to last longer than a second before they fail. A computer can come up with an answer while the answer is still needed. In a simulation, there are no deadlines to meet (except getting something done before the computer fails), but in any kind of real-time programming, it must be completed before the output is needed by the consumer of that output. — noAxioms

Agreed. Which does not alter the definition of computation, a point you have already agreed to.

If it whistles Dixie, it is computing something different. Both should have identical output. Euclid's algorithm isn't a real-time task. — noAxioms

You are arguing about a point you have long since already agreed with. Go take it up with the computer scientists. These are their definitions, not mine.

I take your point about real time computing, but that does not change the definition of computability.

Only to a real-time task, and none of your examples are one. — noAxioms

I take your point about real time computation, but it doesn't change the definition of computation. I'm not aware of how computer scientists patch that little issue but I'll concede it for the moment.

There's a model of physics, and any sim is only a computable approximation of that. Bostrom says that a human is a product of physics, and thus can be functionally simulated given a sufficient level of detail, which is still classical. — noAxioms

So who is the me that's being simulated? I'd like to meet him. Is that what "Prepare to meet your maker" means? Then I'm not in a hurry to meet him, so nevermind.

Same model, different supervenience, if I get my terminology straight. — noAxioms

Ok that was about the overlap between Tegmark and Bostrom. Trolling squared.

I don't know what you think it means for a real person to be simulated. — noAxioms

You (or you quoting Bostrom) say that I'm a simulation, meaning that I'm an approximation of something. I'm asking what I'm an approximation of.

Bostrom suggests a sim of ancestral history, which means that random new people get born, and these people do not in any way correspond to actual people that might have existed in the history of the GS. Much depends on what period of history they choose for their initial state. — noAxioms

So do I correspond to an actual person or not? As I go through my daily life and encounter other humanoid-appearing creatures, is there a way for me to determine which correspond to actual people and which don't? Are the non-corresponding creatures like NPCs in video games?

You know you are really out on a limb here, are you sure you want to be defending this theory of corresponding and non-corresponding people?

That would be something other than 'ancestral history'. You say take a molecular scan of a real person, create a sim model of that exact arrangement of matter, put it in a small environment, and see what it does. That's far more likely than this 'ancestral' thing, but it also would be trivial for the simulated person to realize he's not the original since he's been put in this tiny bounded space, a sort of jail, when he remembers getting into the scanning machine. — noAxioms

Farther out on the same limb. You're making a point not worth defending. I don't happen to remember being put in the scanning machine, but only because my vat programmers have erased my memory. Other than that, I'm trapped on this planet. I can't fly like the birds, I can't swim like the fish, I can't live forever, and I have to wait in line at the DMV just to be allowed to drive my car. Sounds like a tiny bounded space to me.

Now according to your stated beliefs, that simulation wouldn't work. It is computational and you say a person isn't, so the simulated thing would not be functional at any level of detail. — noAxioms

Not my stated beliefs. A premise that I reject, but can nonetheless explore the consequences of. I reject that 2 + 2 = 5, but I can still assert that if 2 + 2 = 5 then I am the Pope.

No, I did not suggest there needs to be a 2nd fishfry that is 'real'. Ancestral history simulations certainly don't produce simulated people that correspond to people in the GS world. — noAxioms

So we're all non-corresponding players now? Not just some of us?

No, not two of you. Bostrom's sim hypothesis would have all of us being in one large simulation, and no real fishfry in the GS world. I apologize if something I posted led you to conclude that I was suggesting otherwise. — noAxioms

You said the simulations are an approximation. I asked the perfectly obvious question, an approximation of what? And your answer is corresponding and non-corresponding entities. No answer at all since it raises even more questions. Like whether my next door neighbor is one of those non-corresponders, in which case I should report him at once.

What is approximated is the physics. I can simulate planetary motion by modeling Earth as a point mass. That's a super-trivial approximation of Earth that works for seeing where it is 100 years from now, but it needs more detail if you say want to see which way the planet is facing in 100 years. — noAxioms

Simulation as approximation. As opposed to simulation as instantiation. A crucial semantic error Bostrom made that has led to the exact confusion you have here. He should have long ago corrected himself. If he had called it "computational instantiation theory" it would be much more clear that all he means is God as a Turing machine.

Yes, but over time, many video games keep getting closer and closer to the sort of reality we'
re used to. Not all of them. Some are still total fiction with deliberate fiction physics, if they have physics at all. They're also video games, which makes them VR, not simulations. — noAxioms

VR? You mean Ms. Pac-Man experiences her reality but we know it's only an illusion?

Honestly I think you have gotten yourself tangled up trying to defend the indefensible.

It's low hanging fruit to debunk various videos. There is indeed whole sites dedicated to debunking relativity in all possible ways, and it is a interesting exercise to find the fallacious reasoning in every one of the arguments. — noAxioms

Not what I said. Not debunking relativity. Debunking explanations of relativity by all the popular Youtube physicists. Not the same thing at all.

The delayed choice quantum eraser isn't really an experiment having anything to do with relativity theory. — noAxioms

Makes my eyes glaze, can't hold up my end of that convo.

News to me as well. It seems to require at least some level of what would qualify as 'understanding'. — noAxioms

Yes, the chess-playing LLM is a startling datapoint.

On the other hand, the new Google AI search says to put glue on pizza in case your cheese doesn't stick. So there's still hope for us humans. @flannel jesus

https://www.businessinsider.com/google-ai-glue-pizza-i-tried-it-2024-5

..... unless you think of fishfry as an avatar. On the other hand, if I am a simulation that is not aware of the fact, I must be able to act and react in my world. In that case, I am not a simulation of anything. — Ludwig V

I am not an avatar in a video game, for the usual Cartesian reason. There's a "me" in here having subjective experiences.

I am not an avatar in a video game, for the usual Cartesian reason. There's a "me" in here having subjective experiences. — fishfry

That's exactly what they want me to believe.

If lots of civilisations are capable of and willing to make simulations then they will, and so simulated persons will greatly outnumber non-simulated persons.

Therefore, if simulated persons do not greatly outnumber non-simulated persons then most civilisations are either incapable of or unwilling to make simulations. — Michael

This ignores the possibility that there may not be "lots of civilizations". In other words, the three alternatives present no cogent assessment of likelihood.

That's exactly what they want me to believe. — Patterner

Do you think avatars in video games can believe anything?

That's exactly what they want me to believe.
— Patterner

Do you think avatars in video games can believe anything? — Janus

I didn't get a notification of this. Glitch the matrix?

, if I am a simulation that is not aware of the fact, I must be able to act and react in my world. In that case, I am not a simulation of anything. — Ludwig V

We seem to be unable to communicate. A simulated thing that was causally disconnected from its environment would be an inaccurate simulation, unless perhaps it was a simulation of dark matter, which really is unable to 'act and react' in its world in any way beyond contributing to the curvature of spacetime.

I can think of models of the weather system that are used to predict the weather. They can be called simulations. They remain quite distinct from the actual weather. There are neither storms, nor rain, nor sunshine inside the computer.

But there very much is storms and rain in the world simulated. It wouldn't be a weather simulation without such things.
Similarly, a simulation of a conscious being would not make a computer conscious, but that doesn't mean that simulated person is not conscious. Bostrom suggests that is exactly what's going on.

Yet the point of the exercise is that it remain as close as possible to what actually happens/-ed. (I can't imagine what the point of ancestral simulations would be, if not that.)

I suppose that's the point, but Bostrom has zero awareness of chaos theory if he thinks that will happen. And he doesn't suggest it. He makes no suggestion that us (the simulation) is evolving in any way the same history as in the simulating world. But yes, what's the point of running such a simulation? Not for prediction purposes, and that's almost always the motivation behind running any simulation.

Once you suppose that the simulations are conscious

I don't think anybody is supposing that. See the above. Yes, a simulated person would behave differently than 'their originals', which I put in quotes because there are no originals in the scenario in question, except as a wild guess at an initial state, giving some characters the same names and roles as historic figures.

The point of the simulations would be lost if real people capable in their own right of acting and reacting in their world.

That sentence lacks a verb, and you lost me. Real people are the ones supposedly running the simulation. The 'point of the simulation' is meaningful to those that are running it. The simulated people have no access to those running the sim, and if they detect or just suspect that they are a sim, they can only guess at the motivations behind the running of it.
Your wording in the verb-less sentence suggests that simulated people would perhaps need to exert some sort of free will over the physics of the simulation. That model isn't compatible with Bostrom's view.

It's me trying to EXPLAIN that OTHER people are using the same word for two very different things. — fishfry

Or its the other people always meaning the same thing, and thus needing only one word for it.
You seem to be the one finding two different meanings, one which sounds like how others use 'simulation', and then this other thing which for reasons not spelled out, require exactness, and is perhaps not computational. I have no idea how to simulate something non-computational, let alone doing it exactly. I don't think anybody else is suggesting any such thing.

But I also maintain that the hypothesis is false. So there's no contradiction.

Well yea, you deny the premise that physics is computational at the necessary levels of precision needed.

So I am not in a position to dictated whether or not 2 + 2 = 5 because I hold that the proposition is false?

No, you're not in the position to say what other people think follows from accepting that 2+2=5.

Pacman ONLY involves computation.

You said "Then whatever [VR] is doing is not computational.", and now you say it is nothing but.
Perhaps you don't consider pacman to be an example of VR. It's admittedly crude and not deeply emersive, but most action video games are nevertheless a form of VR.

The point seems moot. The subject of the topic is simulation theory, not VR theory. VR examples have little to no bearing on simulation hypothesis, a hypothesis you just plain deny due to your lack of belief that a human is computational.

The points you're making in this post are trivial and wrong, not up to your usual standards.
— noAxioms

I don't think I ever said that. This quote is mistakenly attributed to me. Maybe I'm wrong about that. It's a long thread.

I take your point about real time computing, but that does not change the definition of computability.

You seem to go on endlessly about me somehow disagreeing with the definition of computability. I'm not. Real-time issues don't exist in simulation hypothesis, so those are moot until one starts talking about something other than SH.

So who is the me that's being simulated?

Under the simulation hypothesis, you are yourself, which is tautologically true, SH or not. There is not a different 'more real' or 'less real' fishfry somewhere else. It is an ancestor simulation, not a simulation of a fishfry model. Your maker is still your mother, also part of the simulation.

You (or you quoting Bostrom) say that I'm a simulation

You are part of one large simulation, and yes, me quoting Bostrom. I don't buy the hypothesis for a moment.

I'm asking what I'm an approximation of.

You are not an approximation of anything. The simulation is an approximation of the physics of a system (a planet perhaps). You are part of the state of that simulation.

So do I correspond to an actual person or not?

Probably not, unless the simulation's initial state was very recent (our time) and that initial state included a real person who happened to identify as fishfry. I seriously doubt the GS people centuries in the future would know almost anything about you except your parental lineage, all of which is only relevant if the initial state was set since your birth. It has to start somewhere, and that means that the people of that time are created in thin air, with memories totally consistent with their nonexistent past. Doing that requires a full knowledge of how memory and consciousness works, not just a model of how physics works. The initial state requires far more work than does the simulation itself, which is fairly trivial if you get the state right.
Such things are easy with weather and car crashes, but a nightmare for something complex.

As I go through my daily life and encounter other humanoid-appearing creatures, is there a way for me to determine which correspond to actual people and which don't?

Probably none of them, unless they are older than the date of the initial state. Anybody conceived after simulation start has zero probability of having a corresponding real person.

Are the non-corresponding creatures like NPCs in video games?

No. They're no different, except they have real memories, not fake ones put there by the initial state. Maybe the sim only last 10 minutes and everybody is 'corresponding'. This is presuming that the people of the future know exactly who and where everybody is at some random time centuries prior. They don't.
Why do you harp on this? Of what possible importance would it be to anybody in a sim to have a corresponding person (long dead) in the GS? I do realize that I am asking this question of a person who thinks people are special in the universe can cannot be computational like everything else.

You know you are really out on a limb here

Bostrom is maybe. You forget who's pushing the hypothesis. It isn't me, but I'm a computer person and at least I understand it enough to see it for the nonsense it is.

but only because my vat programmers have erased my memory.

SH is not a BiV scenario. VR is, but Bostrom is not talking VR.

So we're all non-corresponding players now? Not just some of us?

An corresponding people from the initial state of the sim would correspond to people centuries dead in the GS world, so nobody can correspond to any living 'real' person.

Simulation as approximation. As opposed to simulation as instantiation.

Sorry, but despite your repeated use of that word, I don't know what you mean by it. You've mentioned that it needs to be 'exact', and the exact physics of even a small trivial real system cannot be exactly simulated, so there cannot be what you call an instantiation. So we're back only to simulations of the approximate physics of some chosen system.

This ignores the possibility that there may not be "lots of civilizations". — Janus

Bostrom addresses that point in his first of three possibilities listed in his abstract.

I didn't get a notification of this. Glitch the matrix? — Patterner

I occasionally get a reply that doesn't make it to the 'mentions' list. Maybe a glitch. I suspect it perhaps might be a post that was already posted, and then later gets edited to mention you, but the one in question here is short and a reply only to you, so that's a significant data point against my theory.

We seem to be unable to communicate. — noAxioms

Oh, I don't think it's as bad as that.

But there very much is storms and rain in the world simulated. It wouldn't be a weather simulation without such things. — noAxioms

No, there are only simulated storms and rain in the simulated world.

The point of the simulations would be lost if real people capable in their own right of acting and reacting in their world.
That sentence lacks a verb, and you lost me. ....
Your wording in the verb-less sentence suggests that simulated people would perhaps need to exert some sort of free will over the physics of the simulation. That model isn't compatible with Bostrom's view. — noAxioms

I'm sorry. I'm afraid I can't re-construct what that sentence was supposed to be. But your version of it is what I was trying to say. I can believe that it is not compatible with Bostrom's view. The question is whether Bostrom's view is coherent.

Once you suppose that the simulations are conscious

I don't think anybody is supposing that. See the above. — noAxioms

Similarly, a simulation of a conscious being would not make a computer conscious, but that doesn't mean that simulated person is not conscious. Bostrom suggests that is exactly what's going on. — noAxioms

So Bostrom does suggest that the simulations of people "inside" the (non-conscious) computer are conscious.

I am not an avatar in a video game, for the usual Cartesian reason. There's a "me" in here having subjective experiences. — fishfry

I'm agree with fishfry here, but adding that if the "me" in here is having subjective experience, then I must be able to interact with the presented illusory environment, that is, I can cause things to happen in the environment and get appropriate feed-back from the environment. But that would make me a real person, not a simulation (though I might be a clone.)

Yes, a simulated person would behave differently than 'their originals', which I put in quotes because there are no originals in the scenario in question, — noAxioms

There's an ambiguity here. There could be simulations of people that are like fictional people. Their originals would be people in general, not people in particular (though an ancestral simulation suggests that they would need to be people in particular - if they aren't, then what makes it an "ancestral" simulation.)

Third: what type of computing power would be required to 'house' this virtual universe? Are we talking about computers that are bigger than the universe itself? — jasonm

This in in fact an argument against simulation theory...

Which is that simulation theory implies some computing power (as we understand computing and computers)
But problem is that in real world there is biology and biological things happening such as us, plants and animals, this is something which "computers" (electronic devices) don't do and therefore it's an argument why we don't live in a computer simulation.

--

Sorry if somebody already told this, I didn't read entire thread.

No, there are only simulated storms and rain in the simulated world. — Ludwig V

Nobody calls them simulated storms. I was in one last night, and we all call it a storm.
Sure, they are simulated storms in the GS (fishfry's term, which I find very useful) world since that's the world in which the simulation is running, but in our world, they are storms. In the case of a weather simulation, in the simulated world, they are storms, but in that case, the GS world is ours, so they are simulated storms relative to us. There are no people in our weather simulations to call them storms, but the point stands: What they're called depends on the point of view.

I can believe that it is not compatible with Bostrom's view. The question is whether Bostrom's view is coherent.

Bostrom proposal is consistent with the methodological naturalism under which all of modern science is based. That means that human beings are treated as just collections of matter doing what the laws of physics says that matter does. I say consistent, but then Bostrom changes the laws of physics from here to there, as does any simulation. A simulation has boundaries, and so a distant star is probably modeled (most of the time) as a simple point source of light. The people in the sim would probably notice if there were no stars in the sky but the simulation hardware is not capable of simulating stellar combustion at the molecular level for the entire visible universe.

So Bostrom does suggest that the simulations of people "inside" the (non-conscious) computer are conscious.

He proposes that we are likely in such a simulation. If you consider yourself to be conscious, then yes, the hypothesis says that you (a simulated thing) is conscious. That's different than saying that the simulation itself is conscious. The simulation and you are different things. The former is a process running in some GS world, and the latter is you, an simulated dynamic arrangement of matter in the simulated world.
His argument proceeds along probability lines, not along empirical evidence lines. This is very similar to the sort of probabilistic reasoning behind the dangers of Boltzmann Brains. No argument for or against Boltzmann Brains can proceed along empirical lines so one is left with probability.

if the "me" in here is having subjective experience, then I must be able to interact with the presented illusory environment,

Of course you interact with your environment. what kind of simulation would it be if you couldn't? Even a statue of Ludwig interacts with its environment, if only to get wet, change temperature, and exert force on the ground. Having subjective experience or not doesn't change that, but you'd probably die pretty quickly if you didn't have that subjective experience.

But that would make me a real person, not a simulation (though I might be a clone.)

You are a real person in this world, but a simulated person relative to the GS world (according to Bostrom). I am perhaps using a different definition of 'real' than you are, and this likely needs to be clarified. I consider what we can see, reach out and touch, to be real to us. You seem to be using a different definition, such as perhaps "is part of the GS", the base world. which presumes no infinite regress.

There's an ambiguity here. There could be simulations of people that are like fictional people. Their originals would be people in general, not people in particular (though an ancestral simulation suggests that they would need to be people in particular - if they aren't, then what makes it an "ancestral" simulation.)

Totally agree. There would be no particular correspondence between people or events in the sim, to people and events in (the past history of) the GS. A war in this world, or a cup being dropped and breaking, would have no particular corresponding event in the GS world. And you're exactly correct: Without this correspondence, how is it being described as an ancestral simulation justified in any way?

But problem is that in real world there is biology and biological things happening such as us, plants and animals, this is something which "computers" (electronic devices) don't do — SpaceDweller

Bostrom's hypothesis is consistent with the methodological naturalism under which all of science operates. That means that plants/animals are very much something that computers can 'do'.

Bostrom's hypothesis is consistent with the methodological naturalism under which all of science operates. That means that plants/animals are very much something that computers can 'do'. — noAxioms

I have read about it here: https://en.wikipedia.org/wiki/Simulation_hypothesis
But this is just a hypothesis based on "Let us suppose for a moment that these predictions are correct."

But computers as we understand them now don't qualify for simulation of biological phenomenon.

Forgive me - what the "GS"?

what the "GS"? — Ludwig V

It means 'Great Simulator', which is the base reality running the base simulation. So if we're 3 levels down, the GS is the first level, the only level that isn't itself a simulation.

The term was coined by fishfry in this topic many posts back, and it's easy to type.

But computers as we understand them now don't qualify for simulation of biological phenomenon. — SpaceDweller

I beg to differ. Computers as we understand them now are quite capable of the task, but at this time, perhaps 40 orders of magnitude speed and memory capacity short of the scale of simulation described by Bostrom. This presumes naturalism of course, and many here (fishfry, possibly Ludwig, possibly yourself) do not so presume.

You seem to be the one finding two different meanings, one which sounds like how others use 'simulation', and then this other thing which for reasons not spelled out, require exactness, and is perhaps not computational. I have no idea how to simulate something non-computational, let alone doing it exactly. I don't think anybody else is suggesting any such thing. — noAxioms

My point is that Bostrom and others are equivocating simulation in this manner, and this clouds their thinking and confuses everyone else. so yes, I am the one finding two different meanings. I might be wrong, but we are 100% agreed that "I am the one finding two different meanings." I not only conceded that, I insist on it. It's my thesis. Glad we're agreed.

I am allowed to have an opinion, right? I've already admitted I haven't read the paper so I could be wrong. But I do have my opinion, and you seem to understand my opinion perfectly well. In which case I'm happy. I don't require agreement. And you have motivated me a bit to go read it, though I do have other priorities and might not get to it. In which case I'll remain both ignorant and wrong. But at least I'll have my opinion :-)

BTW we do simulate noncomputational things all the time. That's why there's computation! We don't know the exact rules by which the economy works, but we run economic simulation of models of the economy all the time.

Well yea, you deny the premise that physics is computational at the necessary levels of precision needed. — noAxioms

Oh no that's not true. I deny that physics is computational, or rather I'm pretty sure it's not.

But I am certain that we simulate, approximate, and model physics extremely well. What's the famous factoid, that quantum chromodynamics predicts the magnetic moment of the electron to 13 decimal places or some such. I certainly respect and enjoy such brilliant approximating. But what the hell's an electron, and more importantly why, nobody quite knows. We're approximating. Simulating. Modeling. We are not engaging with the thing itself.

Whereas (this is my thesis and maybe not Bostrom's) simulation theory says that our very existence, as it really is, is a program in the big computer in the sky. An entirely different thing than simulation.

No, you're not in the position to say what other people think follows from accepting that 2+2=5. — noAxioms

Oh my, are we disagreeing on propositional logic? I thought that everyone (when playing the game of rationality and logic) agrees that IF 2 +2 = 5 THEN I am the Pope. One need not believe either clause to recognize the truth of the proposition.

Truly not following you since I'm sure you agree with me about this. I mean, we all do believe in material implication, don't we?

You said "Then whatever [VR] is doing is not computational.", and now you say it is nothing but.
Perhaps you don't consider pacman to be an example of VR. It's admittedly crude and not deeply emersive, but most action video games are nevertheless a form of VR. — noAxioms

I'll stipulate that if I had enough quarters and enough beer I might find my experience of the game so immersive that it's practically VR. From PacMan to the new Apple headset's not that far a leap, the underlying tech is the same. So I'll call PacMan early VR, I have no problem with that.

I think the point was about execution speed. Self-driving cars and realtime systems in general are doing something that's not just computation, since execution speed is a factor. I assume computer scientists must have a technical term for that, when execution speed makes a difference in the output of a computation. It's just a semantic point, not too important I think.

The point seems moot. The subject of the topic is simulation theory, not VR theory. VR examples have little to no bearing on simulation hypothesis, a hypothesis you just plain deny due to your lack of belief that a human is computational. — noAxioms

I have opinions, I have beliefs, I don't deny them. I really doubt humans are computational in the sense of our present understanding of computation. We have some extra secret sauce, I don't know what it is.

I don't think I ever said that. This quote is mistakenly attributed to me. Maybe I'm wrong about that. It's a long thread. — noAxioms

I'll agree with you there, and the forum software confuses me sometimes.

You seem to go on endlessly about me somehow disagreeing with the definition of computability. I'm not. Real-time issues don't exist in simulation hypothesis, so those are moot until one starts talking about something other than SH. — noAxioms

Ok. Apologies for going on endlessly. I could end this. We're past the point of whatever we were talking about originally. The best thing for me to do would be to go read the Bostrom paper or remain silent, as Wittgenstein suggests.

Under the simulation hypothesis, you are yourself, which is tautologically true, SH or not. — noAxioms

A system can't simulate itself, except trivially in the sense that an apple is a perfect simulation of an apple. If a thing simulated itself the simulation would have to contain a copy of the simulation and ad infinitum.

There is not a different 'more real' or 'less real' fishfry somewhere else. It is an ancestor simulation, not a simulation of a fishfry model. Your maker is still your mother, also part of the simulation. — noAxioms

But then was is my Cartesian "I", the thing that doubts, the thing that is deceived? Is that not being simulated too? What are these thoughts and experiences I spend so much of my life having?

You are part of one large simulation, and yes, me quoting Bostrom. I don't buy the hypothesis for a moment. — noAxioms

But you'll defend it to the death against the likes of me, who hasn't even read the paper? Honestly I'm not worth the effort :-)

I understand that you're trying to just explain to me what Bostrom says, but my problem is that I have certain preconceptions about what he says, and I believe them pretty strongly. That's because even though I haven't read Bostrom, I've read a bit of simulation criticism and support. So I've formed a worldview about it.

You are not an approximation of anything. The simulation is an approximation of the physics of a system (a planet perhaps). You are part of the state of that simulation. — noAxioms

Is my consciousness part of the simulation? Is that the distinction between VR and Sim?

Probably not, unless the simulation's initial state was very recent (our time) and that initial state included a real person who happened to identify as fishfry. — noAxioms

So maybe or maybe not?

I seriously doubt the GS people centuries in the future would know almost anything about you except your parental lineage, all of which is only relevant if the initial state was set since your birth. It has to start somewhere, and that means that the people of that time are created in thin air, with memories totally consistent with their nonexistent past. Doing that requires a full knowledge of how memory and consciousness works, not just a model of how physics works. The initial state requires far more work than does the simulation itself, which is fairly trivial if you get the state right.
Such things are easy with weather and car crashes, but a nightmare for something complex. — noAxioms

Full knowledge of how memory and consciousness works. So Bostrom is assuming this problem has been solved? But that goes against the claim that "the video games are so much better now," an argument often given in support of the simulation hypothesis.

Probably none of them, unless they are older than the date of the initial state. Anybody conceived after simulation start has zero probability of having a corresponding real person. — noAxioms

So I'm not real, according to the theory. Why is simulation theory better than brain in a vat theory or Boltzmann brain theory?

No. They're no different, except they have real memories, not fake ones put there by the initial state. Maybe the sim only last 10 minutes and everybody is 'corresponding'. This is presuming that the people of the future know exactly who and where everybody is at some random time centuries prior. They don't. — noAxioms

Even if they did, they would not know what each person is going to do next. Unless you also reject free will.

Why do you harp on this? Of what possible importance would it be to anybody in a sim to have a corresponding person (long dead) in the GS? — noAxioms

If I'm a simulation, what am I a simulation of?

If I'm a simulation of something, what is that something? If I'm not a simulation of anything, then by definition I'm not a simulation. I'm an instantiation. This is the semantic point I'm insisting on, rightly or wrongly.

I do realize that I am asking this question of a person who thinks people are special in the universe can cannot be computational like everything else. — noAxioms

Well it's plainly false that "everything else" is computational. For example an oracle for the Halting problem is not computational, yet it's a standard device in computer science theory. So that's one thing that's not computational. There are many others. Chaitin's Omega is not computable but we can define it. For all you know, the world is not computational. For all I know, it is. But You are wrong to say that "everything else" is computational when I know at least two things that aren't. Maybe there are others. Like minds.

You can't go from "people aren't special in the universe," to "therefore people are computational." There's no link between those two things.

Bostrom is maybe. You forget who's pushing the hypothesis. It isn't me, but I'm a computer person and at least I understand it enough to see it for the nonsense it is. — noAxioms

You are strenuously trying to explain to me that Bostrom's idea is nonsense; but not liking my own argument as to why it's nonsense. Why are we doing this?

SH is not a BiV scenario. VR is, but Bostrom is not talking VR. — noAxioms

SH is not brain in vat? Now I'm confused again. I thought VR was like a video game, and SH is where my mind is being instantiated too.

An corresponding people from the initial state of the sim would correspond to people centuries dead in the GS world, so nobody can correspond to any living 'real' person. — noAxioms

So now I'm a simulation of a dead person. You know you keep changing your explanation of this point. I think you should consider retracting it entirely, and trying to understand why I am so insistent on my claim that the word simulation is being used wrongly here.

Sorry, but despite your repeated use of that word, I don't know what you mean by it. — noAxioms

I've explained it as well as I can.

You've mentioned that it needs to be 'exact', and the exact physics of even a small trivial real system cannot be exactly simulated, so there cannot be what you call an instantiation. So we're back only to simulations of the approximate physics of some chosen system. — noAxioms

There cannot be instantiation? What do you think the universe is? We've all been instantiated somehow. We are here. We have been instantiated. That's the point.

tl;dr: Well ... I'm pretty clear in my own mind what I mean, but if you say I'm misunderstanding Bostrom, I can't disagree, since I haven't read the paper. But I really think I'm right and everyone else is confused on this point. So if you simply want to make the point that I have an opinion and that I'm wrong. I agree. I have my opinion and I may be wrong, but the more we talk about it, the more these concepts are clear in my mind, and I think I'm right.

God instantiated the universe. You say God is a digital computer. I say that's one extra assumption and by Occam, we should just stick with God. That's what I get from Bostrom.

I am not an avatar in a video game, for the usual Cartesian reason. There's a "me" in here having subjective experiences.
— fishfry
I'm agree with fishfry here, but adding that if the "me" in here is having subjective experience, then I must be able to interact with the presented illusory environment, that is, I can cause things to happen in the environment and get appropriate feed-back from the environment. But that would make me a real person, not a simulation (though I might be a clone.) — Ludwig V

Ah the mind-body problem. I saw a video of Searle giving a lecture. He raised his right arm and said, "I think to myself, I'll raise my right arm. And my right arm goes up. How does that happen?"

I know there's no agreement regarding free will. But if we have free will, then we aren't simulations. I mean, how can you use rules and code to write something that doesn't follow rules and code?

My reply was edited since I think I finally grasped what you mean by 'instantiation', as being distinct from 'simulation'.

My point is that Bostrom and others are equivocating simulation in this manner, — fishfry

No they're not. They are using the word in a single consistent manner at all times. You admit that it is you that is finding two different meanings and trying to use two different words to distinguish them. Under naturalism, there is a physical system that is simulated using a model of physical laws. It's completely computational in all cases.

I am allowed to have an opinion, right?

I acknowledged your opinion. It isn't wrong, merely inconsistent with Bostrom's naturalism opinion.

My opinion is that the economy isn't an example of something noncomputational.

I deny that physics is computational, or rather I'm pretty sure it's not.

With that I completely agree, which is why any computation of our physics is necessarily an approximation.

Whereas (this is my thesis and maybe not Bostrom's) simulation theory says that our very existence, as it really is, is a program in the big computer in the sky. An entirely different thing than simulation.

I'm unclear of the distinction between that and simulation. Bostrom says that it is humans (or 'post-humans') running the big computer. Simulation theory in general doesn't require that detail.

Oh my, are we disagreeing on propositional logic?

Not at all. I am balking at your equating a premise that science in general would find false (2+2=5) with one that science in general accepts as true (naturalism).

So I'll call PacMan early VR, I have no problem with that.

Good. Best they could do at the time. Even today, few non-headset games even have a first person perspective. Minecraft and Portal come to mind. I'm sure there are others, but still a small percentage. Earliest one I can think of is Battlezone. Remember that one? It pre-dates pacman I think. Ground breaking stuff it was.

I assume computer scientists must have a technical term for that, when execution speed makes a difference in the output of a computation

Yes. "Real time". But technically, all computation has this requirement, which is one reason nobody makes real Turing machines. Imagine if you had a 4-banger calculator that took 40 years to add 2+2. Would you use it? Does that make adding 2+2 something more than computational?

I have opinions, I have beliefs, I don't deny them.

A good stance, and I worded it as 'belief' instead of 'opinion', which may have been too hash. The simulation hypothesis can only be considered under the naturalism it presumes, whether or not naturalism is part of one's opinion.

We have some extra secret sauce, I don't know what it is.

Your opinion then is that we have the secret sauce, and that whatever it is, it isn't computational, although I don't know how you can infer it being noncomputational if you don't have any idea what it is. So probably also another opinion.

But then was is my Cartesian "I", the thing that doubts, the thing that is deceived?

There isn't a separate Cartesan "I" thing under naturalism.

But you'll defend it to the death against the likes of me, who hasn't even read the paper?

Explaining it and defending it are two different things. The abstract is accurate, meaning I find it reasonably valid and sound, although it seems that it has been updated since wiki lists 5 options now instead of the original 3, but the new ones seem to overlap with the old ones.

That's because even though I haven't read Bostrom, I've read a bit of simulation criticism and support.

Much (the majority?) of criticism and support seem to be from people without a reasonable understanding of what it says. You can include me on that list. Don't trust what I say, but I have read the actual paper at least, and I know the difference between it, other sim proposals, and with a VR proposal. Many of the articles discussing it seem not to know the differences.

Is my consciousness part of the simulation?

So says Bostrom, yes. Naturalism says it is if the simulation is run at a sufficiently detailed level, which is still classical, not necessarily down to the quantum level.

Is that the distinction between VR and Sim?

A VR does not produce a second consciousness for the avatar. A sufficiently detailed VR might for an NPC, but nothing like that exists in any current VR system. The current VR immersion (with the 3D headset and all) is barely better than the one for Pacman. With a good one, there'd be no controller in your hand. You would not have access to say your real body being touched.

So maybe or maybe not?

Very likely not.

Full knowledge of how memory and consciousness works.

No, that isn't needed, but it is needed if the sim is gleaning intent from the physics it is simulating, and Bostrom very much does propose that it is interpreting human intent. Also, that understanding is needed for any human that is not born, but is part of the initial state. So bottom line, yea, it is needed.
A pure simulation of a human from the human's initial state has no need for knowledge of how memory and consciousness works, for the exact same reason that physics doesn't need to know the details of the workings of the things that result from the physics.

So Bostrom is assuming this problem has been solved?

Centuries hence, it seems so. Without it, there can be no plausible initial state, unless you go back 3 billion years where the initial states were less complicated

But that goes against the claim that "the video games are so much better now," an argument often given in support of the simulation hypothesis.

No video game claims any understanding of what is referred to as the hard problem. If somebody references a game as an illustration of Bostrom's hypothesis, then they don't understand the difference between a sim and a VR. But they're probably just using games as one way to demonstrate Moore's law, which Bostrom presumes to continue for centuries.

So I'm not real, according to the theory.

If all this is a simulation, I am still very much real according to my stated definition of 'real' and you've not given yours. SH is very different than BiV and Boltzmann brains.

Even if they did, they would not know what each person is going to do next. Unless you also reject free will.

I don't think there is the sort of free will you're thinking if our world is a simulation. A simulation like that doesn't have causality from outside the system. If it did, it would probably be a VR. I say this, but I've done chip simulations that get driven from external state. The signals fed to the chip are artificial, not from other simulated circuits since it's only the one chip being tested. Such a chip simulation is hard to classify as a VR.

If I'm a simulation, what am I a simulation of?

You are part of the physical evolution of the chosen initial state. That answer pretty much applies to any simulation, including all the ones I've seen done. You want to call it an instantiation and I think I see how you're using that word. A simulation is the execution (instantiation) of a mathematical model, that model itself being an approximation of some hypothetical corresponding reality. Since it is the execution of a model, it is presumably exact, except the model might include randomness, in which case the exactness is wrong since multiple instantiations of the same model will evolve differently. Bostrom does propose some randomness in his model, so not sure how 'exact' it would be. Said randomness need only be apparent, so it can be driven by a pseudo-random mechanism, which restores the deterministic nature of the simulation.

You can't go from "people aren't special in the universe," to "therefore people are computational."

I don't think any physical thing (people or otherwise) is computational. But an approximation can be, and people are no exception to that according to science.

You are strenuously trying to explain to me that Bostrom's idea is nonsense; but not liking my own argument as to why it's nonsense. Why are we doing this?

You're not taking down Bostrom's argument. You presume his premisies to be false. I presume them to be true, and I think his conclusion doesn't follow from them.

SH is not brain in vat?I thought VR was like a video game, and SH is where my mind is being instantiated too.

That's right. BiV is like the video game: an artificial (virtual) experience stream to the real (not simulated) experiencer, effectively a video game for the B in the Vat, whatever its nature.

So now I'm a simulation of a dead person.

Very unlikely for the reason's I've stated. Only if you're part of the initial state, and then only if that initial state had some kind of access to the molecular state of everybody on Earth many centuries prior, which they don't because there's no tech today that can do that.

There cannot be instantiation? What do you think the universe is?

Under Bostrom's view, the universe is a simulation, or at least something that can be seen from the simulation since most of it is just phenomenal.

Yes, our universe is what it is, and that's an intantiation in your wording. But the wording give no clue as to the nature of how it comes to be, since any story fits. Bostrom gives one possible way that it is instantiated. A deity is another. Both fail to solve the problem of 'why there's something and not nothing', but Bostrom isn't positing a solution to that problem. The deity answer often is such an attempt, and a failed one since it explains a complicated thing by positing an even more complicated thing.

We've all been instantiated somehow. We are here. We have been instantiated. That's the point.

I think I understand your usage of that word, and I don't in any way presume that I am instantiated. But that's me, being far more skeptical than most. Being instantiated doesn't solve any problems. I personally suspect that the sum of 2 and 2 is 4 even in the absence of anything actually performing that calculation (absence of it being instantiated). Apparently I am in the minority in this opinion.

if you simply want to make the point that I have an opinion and that I'm wrong. I agree.

I never said your opinion is wrong. It's just a different one than somebody else's. Different premises.

I have my opinion and I may be wrong, but the more we talk about it, the more these concepts are clear in my mind, and I think I'm right.

I think I'm in the minority of being somebody who has opinions X and Y and such, and I also think I'm mostly wrong about them. Some are probably right, but I realize that the odds of me getting most of them right is stupidly low.

God instantiated the universe. You say God is a digital computer.

I say that?

I say that's one extra assumption and by Occam, we should just stick with God. That's what I get from Bostrom.

'God' sound like the extra assumption in that statement. Occam says it's better to ditch both the deity and the simulation layers

But if we have free will, then we aren't simulations. — Patterner

Totally agree. Some take that as evidence against the argument, but only because 'free will sounds like a good thing, therefore I must have it". To me it sounds like a bad thing, but I don't hold a presumption that the entities in the simulation will be held responsible for their choices, by entities not in the simulation.

.

I don't hold a presumption that the entities in the simulation will be held responsible for their choices, by entities not in the simulation. — noAxioms

If entities create a simulation that includes other entities that do not have free will, the creators would be ... what's there right word ... idiots if they held the creations responsible for their choices. i'm not sure it would be worse to hold characters in a story you write responsible for their choices.

My reply was edited since I think I finally grasped what you mean by 'instantiation', as being distinct from 'simulation'. — noAxioms

Yay!

But may I say that earlier, you noted that these threads are getting long. I too could live with a much more focussed conversation. Actually I'm near the end of my interest in this topic. I've already stipulated that I haven't read the Bostrom paper, so for all I know, I'm misunderstanding his argument. Can we wrap this up soon? It's feeding time in my digital vat.

No they're not. They are using the word in a single consistent manner at all times. You admit that it is you that is finding two different meanings and trying to use two different words to distinguish them. Under naturalism, there is a physical system that is simulated using a model of physical laws. It's completely computational in all cases. — noAxioms

Naturalism is computationalism? I genuinely doubt that, but I'm no expert.

I acknowledged your opinion. It isn't wrong, merely inconsistent with Bostrom's naturalism opinion. — noAxioms

I concede that I have not read Bostrom all the way through. Every time I've taken a run at his paper in the past, my eyes glaze over, my mind says, "This is bullpucky," and I click on something else.

My opinion is that the economy isn't an example of something noncomputational. — noAxioms

Really? The economy is the deterministic output of a computer program? Your Nobel in economics awaits if you can prove that, even if the economics Nobel is not really a Nobel.

I deny that physics is computational, or rather I'm pretty sure it's not.
With that I completely agree, which is why any computation of our physics is necessarily an approximation.[/quoe]

We are 100% in agreement on this point.
— noAxioms

I'm unclear of the distinction between that and simulation. Bostrom says that it is humans (or 'post-humans') running the big computer. Simulation theory in general doesn't require that detail. — noAxioms

I was going to respond, but we've been over this. If my reality is nothing but a "simulation," then I'm not real. There is only the simulation. Meaning that I'm not a simulation, I'm an instantiation. Ok I responded anyway. Wish I hadn't.

Not at all. I am balking at your equating a premise that science in general would find false (2+2=5) with one that science in general accepts as true (naturalism). — noAxioms

I said that "If 2 + 2 = 5 then I am the Pope" is a true proposition of sentential logic. Nobody said anything about naturalism. How did you interpolate that here? You entirely changed the subject.

Good. Best they could do at the time. Even today, few non-headset games even have a first person perspective. Minecraft and Portal come to mind. I'm sure there are others, but still a small percentage. Earliest one I can think of is Battlezone. Remember that one? It pre-dates pacman I think. Ground breaking stuff it was. — noAxioms

Never much of a game aficionado I'm afraid. I remember Lunar Lander and Asteroids.

But I do object to the "best they could do at the time" argument that video games are getting better, therefore in the future they'll be indistinguishable from reality. Since we have made zero progress on instantiation (there's that word again) consciousness, we have made NO progress in that area, and the argument fails.

Yes. "Real time". But technically, all computation has this requirement, which is one reason nobody makes real Turing machines. Imagine if you had a 4-banger calculator that took 40 years to add 2+2. Would you use it? Does that make adding 2+2 something more than computational? — noAxioms

I am not the person who makes the distinction between computability and complexity theory. That's the computer scientists. You already agree with me. I agree that I don't know the official terminology for how this affects real time systems.

A good stance, and I worded it as 'belief' instead of 'opinion', which may have been too hash. The simulation hypothesis can only be considered under the naturalism it presumes, whether or not naturalism is part of one's opinion. — noAxioms

Maybe you could explain what you mean by naturalism. I don't know what the word means.

Nevermind, I looked it up. It's "the philosophical belief that everything arises from natural properties and causes, and supernatural or spiritual explanations are excluded or discounted."

Fine. I can stipulate to that. But naturalism does not imply computationalism! I'm sure I've already made this point at length.

Your opinion then is that we have the secret sauce, and that whatever it is, it isn't computational, although I don't know how you can infer it being noncomputational if you don't have any idea what it is. So probably also another opinion. — noAxioms

It's clear to me that there are profound limits to what can be computed. It was clear to Turing as well, since he was the first person to document an easily described problem that can not possibly be solved by a computation.

There isn't a separate Cartesan "I" thing under naturalism. — noAxioms

So the simulator implements my consciousness. And exactly how does it do that? And exactly what is it that makes a program conscious? Is my web browser conscious? Who believes such nonsense?

Explaining it and defending it are two different things. The abstract is accurate, meaning I find it reasonably valid and sound, although it seems that it has been updated since wiki lists 5 options now instead of the original 3, but the new ones seem to overlap with the old ones. — noAxioms

I made a recent attempt to read the paper, by far not my first time, and as usual my eyes glazed and I clicked on something else. Perhaps someday I'll get it, but its charms and logic are lost on me so far.

Much (the majority?) of criticism and support seem to be from people without a reasonable understanding of what it says. You can include me on that list. Don't trust what I say, but I have read the actual paper at least, and I know the difference between it, other sim proposals, and with a VR proposal. Many of the articles discussing it seem not to know the differences. — noAxioms

I have conceded many times over the superiority of your knowledge in this area. I have not conceded the superiority of your opinions about what it all means. In fact you yourself have said, more than once, that you don't agree with the conclusions of the paper. So why bother with the likes of me? Leave me in my ignorance, please. I don't care much about simulation theory because it's such obvious pretentious garbage.

Is my consciousness part of the simulation?
So says Bostrom, yes. — noAxioms

But how is that done? What lines of code must I run to imbue my program with consciousness? I'd be very interested to hear your answer to this question. Is it just a matter of lines of code? Microsoft Windows has some fifty million lines of code. Is it conscious? Is it just a matter of running it faster? I can't believe you are defending such an indefensible proposition, that a computer program can be conscious, without having any inkling of how it's done.

Naturalism says it is if the simulation is run at a sufficiently detailed level, which is still classical, not necessarily down to the quantum level. — noAxioms

Naturalism does not say a computer program can become conscious. It says we REJECT the supernatural, we don't embrace it!

A VR does not produce a second consciousness for the avatar. A sufficiently detailed VR might for an NPC, but nothing like that exists in any current VR system. The current VR immersion (with the 3D headset and all) is barely better than the one for Pacman. — noAxioms

But in terms of implementing consciousness, it is no better than Pong. That is the "video game improvement" argument I objected to earlier. We have made NO progress in implementing consciousness. No progress at all.

With a good one, there'd be no controller in your hand. You would not have access to say your real body being touched. — noAxioms

I would still be the one having the experience. The "I" having the experience. My experience is not induced by a headset. I really don't think you are thinking these matters through.

No, that isn't needed, but it is needed if the sim is gleaning intent from the physics it is simulating, and Bostrom very much does propose that it is interpreting human intent. Also, that understanding is needed for any human that is not born, but is part of the initial state. So bottom line, yea, it is needed. — noAxioms

Can't comment on what I haven't read. If Bostrom thinks a computer can instantiate consciousness, the burden is on him to say how, since nobody has the slightest idea how.

A pure simulation of a human from the human's initial state has no need for knowledge of how memory and consciousness works, for the exact same reason that physics doesn't need to know the details of the workings of the things that result from the physics. — noAxioms

Where is your evidence that computer programs are conscious? Why are you arguing this nonsensical line that you can't possibly have any evidence for?

Centuries hence, it seems so. Without it, there can be no plausible initial state, unless you go back 3 billion years where the initial states were less complicated — noAxioms

So in the future there will be a breakthrough. Well who can argue with that? Will that be before or after pigs fly? You know, I can't believe Bostrom is making such a weak and nonsensical argument.

But that goes against the claim that "the video games are so much better now," an argument often given in support of the simulation hypothesis.
No video game claims any understanding of what is referred to as the hard problem. If somebody references a game as an illustration of Bostrom's hypothesis, then they don't understand the difference between a sim and a VR. But they're probably just using games as one way to demonstrate Moore's law, which Bostrom presumes to continue for centuries. — noAxioms

Quite possibly I've seen other people make the video game progress argument. If it's not Bostrom's, I have terribly maligned him. But my simulator made me do it, honest. I had no choice.

Do I have choice, by the way? Does Bostrom deny free will? My web browser can't be a word processor, no matter how hard it tries. Programs do exactly what they're programmed to do, a matter of great frustration to programmers.

If all this is a simulation, I am still very much real according to my stated definition of 'real' and you've not given yours. SH is very different than BiV and Boltzmann brains. — noAxioms

I don't see how. If I'm a simulation running in a computer, I might as well be a brain in a vat or the only conscious thing in the universe. What distinction among these ideas do you see?

I don't think there is the sort of free will you're thinking if our world is a simulation. — noAxioms

No free will. Ok. So sim theory is ultimately nihilistic. I murdered all those people but it wasn't my fault, Your Honor, my simulator made me do it.

A simulation like that doesn't have causality from outside the system. If it did, it would probably be a VR. I say this, but I've done chip simulations that get driven from external state. The signals fed to the chip are artificial, not from other simulated circuits since it's only the one chip being tested. Such a chip simulation is hard to classify as a VR. — noAxioms

External inputs don't matter, since programs are coded to do one thing or another thing depending on the input. Programs don't have free will by virtue of getting external inputs.

You are part of the physical evolution of the chosen initial state. That answer pretty much applies to any simulation, including all the ones I've seen done. You want to call it an instantiation and I think I see how you're using that word. A simulation is the execution (instantiation) of a mathematical model, that model itself being an approximation of some hypothetical corresponding reality. Since it is the execution of a model, it is presumably exact, except the model might include randomness, in which case the exactness is wrong since multiple instantiations of the same model will evolve differently. Bostrom does propose some randomness in his model, so not sure how 'exact' it would be. Said randomness need only be apparent, so it can be driven by a pseudo-random mechanism, which restores the deterministic nature of the simulation. — noAxioms

You contradicted yourself at least three times getting from the beginning to the end of that para. No free will but there might be if there's randomness, but it might only be pseudo-randomness, in which case it's not random after all.

Anyway I don't think I have the heart to unpack all the (in my opinion) confusion in that paragraph.

You can't go from "people aren't special in the universe," to "therefore people are computational."
I don't think any physical thing (people or otherwise) is computational. But an approximation can be, and people are no exception to that according to science. — noAxioms

According to science? What serious scientist parrots a single word of this nonsense? Name and shame please. I'm not talking about Neil deG Tyson or the deluded George Smoot in a TED talk. I mean a scientist who wrote a peer-reviewed paper that says "people are no exception" ... wait, what? I went back to try to parse what you wrote and you agreed -- YOU AGREED! -- that no physical thing is computational.

Then we're done. You have accepted my point.

Yet you think I'm an approximate computation?

I urge you to think about what you are saying.

You're not taking down Bostrom's argument. You presume his premisies to be false. I presume them to be true, and I think his conclusion doesn't follow from them. — noAxioms

Ok. That's interesting. Why do you think his conclusion doesn't follow from his premises? That might be interesting.

That's right. BiV is like the video game: an artificial (virtual) experience stream to the real (not simulated) experiencer, effectively a video game for the B in the Vat, whatever its nature. — noAxioms

So brain in vat IS is like simulation after all?

Very unlikely for the reason's I've stated. Only if you're part of the initial state, and then only if that initial state had some kind of access to the molecular state of everybody on Earth many centuries prior, which they don't because there's no tech today that can do that. — noAxioms

Where are all these rules and justifications written down?

Under Bostrom's view, the universe is a simulation, or at least something that can be seen from the simulation since most of it is just phenomenal. — noAxioms

A simulation of what? And here we go in circles again.

Yes, our universe is what it is, and that's an intantiation in your wording. But the wording give no clue as to the nature of how it comes to be, since any story fits. — noAxioms

The theists say God did it. The sim theorists say God did it and God is a Turing machine. Why am I required to spend any time at all caring about this argument?

Bostrom gives one possible way that it is instantiated. A deity is another. Both fail to solve the problem of 'why there's something and not nothing', but Bostrom isn't positing a solution to that problem. The deity answer often is such an attempt, and a failed one since it explains a complicated thing by positing an even more complicated thing. — noAxioms

Oh no, that's my point. Sim theory is God as a program, constrained by the laws of computation. That's one extra hypothesis. Unconstrained God is MORE LIKELY than constrained God. Why is God constrained to computability? If there is a God, I imagine God can solve the Halting problem (simply by looking at all the programs and seeing which ones halt) and therefore, God is not computational. That's a pretty good argument IMO.

I think I understand your usage of that word, and I don't in any way presume that I am instantiated. — noAxioms

Really. You're not here at all? You're imagining all this? But the thing that's imagining exists. So that was instantiated.

You were instantiated and you are instantiated at every moment of your existence.

But that's me, being far more skeptical than most. — noAxioms

That you don't exist? That takes skepticism a bit too far.

Being instantiated doesn't solve any problems. — noAxioms

Nor does denying you exist.

I personally suspect that the sum of 2 and 2 is 4 even in the absence of anything actually performing that calculation (absence of it being instantiated). Apparently I am in the minority in this opinion. — noAxioms

That's a different subject entirely. Best if I don't respond with my opinion.

I never said your opinion is wrong. It's just a different one than somebody else's. Different premises. — noAxioms

Well these days even that's heresy.

I think I'm in the minority of being somebody who has opinions X and Y and such, and I also think I'm mostly wrong about them. Some are probably right, but I realize that the odds of me getting most of them right is stupidly low. — noAxioms

This conversation is flaming out entirely.

God instantiated the universe. You say God is a digital computer.
I say that? — noAxioms

You say that Bostrom says that, but then you say you disagree with Bostrom. But here I meant you as someone in general who believe in simulation theory.

'God' sound like the extra assumption in that statement. Occam says it's better to ditch both the deity and the simulation layers — noAxioms

So you reject simulation theory. Good. We're in agreement. Let's stop arguing with each other and join forces and fight the evil simulation theorists.

Ah the mind-body problem. I saw a video of Searle giving a lecture. He raised his right arm and said, "I think to myself, I'll raise my right arm. And my right arm goes up. How does that happen?" — fishfry

How could the mind-body problem not be relevant if people are positing that sims might be people (and sometimes asserting that at least some people are sims?)
Yes. Sometimes I find his tendency to present dualism as common sense ridiculous and sometimes annoying. It reminds me of Bishop Berkeley and his wilful refusal to recognize that he is contradicting common sense.
But the rhetoric of that sentence is genius. A mystery created from a commonplace.

But I do object to the "best they could do at the time" argument that video games are getting better, therefore in the future they'll be indistinguishable from reality. — fishfry

I do so agree. That argument is pure hand-waving. Completely acceptable in a (conventional) fiction, where we aren't expected to ask questions.

Nevermind, I looked it up. It's "the philosophical belief that everything arises from natural properties and causes, and supernatural or spiritual explanations are excluded or discounted." — fishfry

"Naturalism" is used much more widely than that. I've been classified as a naturalist because I reject dualism.

If my reality is nothing but a "simulation," then I'm not real. There is only the simulation. Meaning that I'm not a simulation, I'm an instantiation. — fishfry

Better put than I managed.

How could the mind-body problem not be relevant if people are positing that sims might be people (and sometimes asserting that at least some people are sims?)
Yes. Sometimes I find his tendency to present dualism as common sense ridiculous and sometimes annoying. It reminds me of Bishop Berkeley and his wilful refusal to recognize that he is contradicting common sense.
But the rhetoric of that sentence is genius. A mystery created from a commonplace. — Ludwig V

I believe in that same lecture (or perhaps a different one) he did NOT advocate dualism. He advocated what I call "secret sauce," my phrase, not Searle's. That is, consciousness is physical, but not computational. That's the point I've been making to @noAxioms.

"Naturalism" is used much more widely than that. I've been classified as a naturalist because I reject dualism. — Ludwig V

I thought naturalism (as I understand it, based on a five second lookup) is the exact opposite of dualism. Is naturalism = physicalism? Or is there a further distinction?

My point is that naturalism is not necessarily computationalism. Mind could be physical but not computatational. This is something I believe, though of course proof is lacking.

If my reality is nothing but a "simulation," then I'm not real. There is only the simulation. Meaning that I'm not a simulation, I'm an instantiation.
— fishfry
Better put than I managed. — Ludwig V

Thanks, but noAxioms doesn't believe in the word.

@noAxioms, I was motivated to click on the Bostrom paper and I actually found the line that stops me in my tracks every time.

Suppose that these simulated people are conscious (as
they would be if the simulations were sufficiently fine‐grained and if a certain
quite widely accepted position in the philosophy of mind is correct). — Bostrom

As you know, I am not one of those who "widely accept" this utterly unprovable and extremely unlikely claim. I can't see reading further. Bostrom assumes that consciousness can be implemented on a computer. I say no. It can't.

I found something else that stops me from reading the rest of the paper. It's the very first sentence.

Many works of science fiction as well as some forecasts by serious technologists
and futurologists predict that enormous amounts of computing power will be
available in the future. — Bostrom

Yes. But no matter how enormous the amount of computing power we have, it can not and does not increase the space of problems we can solve. That's the essence of computability. More power makes the computation run faster. It can't compute anything that wasn't already computable before.

So now I see why I can never get into the Bostrom paper. Right in the first paragraph, he loses all credibility with me. It's a core assumption that computers can create consciousness. There's no evidence for that. Why read further? I'll concede that IF consciousness is computational, then I'm already a "simulation," ie an executing computer program. And who's the simulator? God or a future civilization? What difference does it make if I'm a program either way?

So even if he's right, there's no reason to read the rest of it. Computational consciousness is one of his core assumptions. If it's false, the paper's worthless. If it's true, the paper's trivial.

And if the simulators are a future civilization, who created them? In the end it's either "God did it," or "We don't know." So this paper adds nothing. He slipped in the computational consciousness assumption, in which case there's actually nothing else to say. We're all executing programs.

Oh, and instead of justifying and supporting his computational consciousness claim, he blithely says it's "widely accepted." By whom?

Once again, I'm dismayed at how many otherwise clever people take this paper seriously. And once again, I could not get past the introduction and the first paragraph.

If entities create a simulation that includes other entities that do not have free will, the creators would be ... what's there right word ... idiots if they held the creations responsible for their choices. — Patterner

Quite right, but they still can be held responsible for their choices in the simulation itself. If you make a bad choice (cross street without looking), it's your fault if you get hurt/killed. No point in having a better brain if it isn't useful to make good choices. Not having free will does not mean you have no choice.

I'm not sure it would be worse to hold characters in a story you write responsible for their choices.

Characters in a story have no will at all. Their will is at best that of the author, and perhaps the author is responsible for their actions.

Naturalism is computationalism? I genuinely doubt that, but I'm no expert. — fishfry

Naturalism is not-dualism. No secret sauce.
Physics is not computational, but an approximation of it is, sufficient to simulate consciousness.

The economy is the deterministic output of a computer program?

Strawman. I never said that.

Meaning that I'm not a simulation, I'm an instantiation.

The way you seem to define instantiation, you are one whether or not Bostrom's hypothesis is true.

Since we have made zero progress on instantiation (there's that word again) consciousness

Your assertion. I disagree. I do agree that video games are not where this progress is being made since no video game to date has need of it.

So the simulator implements my consciousness.

Thee simulator implements physics. Physics implements your consciousness, regardless of whether the physics is simulated or not. Under supernaturalism, this isn't true.

And exactly what is it that makes a program conscious?

The program has no need of being conscious, just like atoms are not conscious. You are conscious, not the program, not the physics that underpins how your consciousness works.
Technically, a simulation of some system is far simpler than say microsoft code, but it is also much larger since it needs far more data than the capacity of say some desktop.

I would still be the one having the experience. The "I" having the experience.

That's right, which is why a video game is not a model of the simulation argument. Sim is not VR. Video games are VR. VR is dualism. Sim is physicalism.

If Bostrom thinks a computer can instantiate consciousness, the burden is on him to say how, since nobody has the slightest idea how.

It's not on him to say how. It's on those GS guys 10 centuries from now. Part of being 'posthuman' is apparently that they've figured it all out, at least far enough to glean focus and intent from watching raw physics happen, because the algorithm he suggests depends on these things.

Where is your evidence that computer programs are conscious?

Strawman. I never said they were. If this world is a sim, it isn't any program that is conscious, it is just us. I don't think this world is a sim.

So in the future there will be a breakthrough.

A lot of them, yes. Far more than I can accept.

But my simulator made me do it, honest. I had no choice.

Patterner above makes a good reply to this. Determinism made me do it. I'm not responsible. Doesn't work that way.

Do I have choice, by the way? Does Bostrom deny free will?

Nicely illustrating the mistake of equivocating choice and free will. Don't need the latter to have the former, as evidenced by our having evolved expensive brains to make better choices. Free will does not add any survival benefit.

Programs don't have free will by virtue of getting external inputs.

Unless the external input IS the will, as it is in any VR.

You contradicted yourself at least three times getting from the beginning to the end of that para. No free will but there might be if there's randomness, but it might only be pseudo-randomness, in which case it's not random after all.

You've identified no contradictions. Randomness is not free will. I did not mention free will in the paragraph quoted. There is no free will in Bostrom's proposal.

You asked what Bostrom's sim is a simulation of. I answered that.

According to science?

Per the methodological naturalism under which science operates. If one presumes otherwise, it isn't science.

Yet you think I'm an approximate computation?

I never said any such thing. You do like putting crazy words in my mouth.

I urge you to think about what you are saying.

I urge you to read what I'm saying.

So brain in vat IS is like simulation after all?

I urge you to read what I'm saying.

The sim theorists say God did it and God is a Turing machine.

I urge you to read what sim theorists are saying, because it certainly isn't that, and it isn't anything I've said.

Why do you think his conclusion doesn't follow from his premises? That might be interesting.

First option: We never get 'posthuman'. His description of the requirement for this posthuman state is so high that the probability of option 1 being the case is 1 to an awful lot of digits. His argument requires that probability to be close to zero. I could go on, but that's enough.

Really. You're not here at all?

I am quite here, no problem. But I'm not a realist, and 'instantiation' seems to be synonymous with 'to be made real in some way', or more exactly, to set the property of being real to true. I define being real as a relation, not a property like realism does, so an instantiator ceases to be a necessity.
You asked. I don't expect you to accept it, and you'll no doubt bend it to something I didn't say.

That you don't exist? That takes skepticism a bit too far.

No, I just have a different definition of 'to exist', a relation, not a property. And yes, this very much solves a problem that plagued me for years, one that comes up in this forum frequently since the typical answers don't work.

And if the simulators are a future civilization, who created them? In the end it's either "God did it," or "We don't know."

And you said that my (minority) view didn't solve any problems, yet here is one that isn't solved by the more mainstream stances.

How could the mind-body problem not be relevant if people are positing that sims might be people (and sometimes asserting that at least some people are sims?) — Ludwig V

Mind-body problem is only relevant to dualism, and sim theory isn't dualism, so the there's no problem. I think the term is 'interactionism', how the dual aspects interact with each other.
It's very relevant to a VR. How does my decision to point a gun at the baddie cause Lara Croft to raise her arm? There has to be a causal connection between my decision and her arm, and there is. But under sim theory, there isn't two separate things that need to interact, so the problem doesn't arise. If Bostrom is wrong about his philosophy of mind, then his hypothesis falls flat.

I believe in that same lecture (or perhaps a different one) he [Bostrom?] did NOT advocate dualism. ... That is, consciousness is physical, but not computational. — fishfry

Wait, Bostrom said that mind is not computational, and yet pushes a view that our consciousness is the result of a computation? That seems to be a direct denial of his own paper. Got a link to where this is said?

I can't see reading further. Bostrom assumes that consciousness can be implemented on a computer. — fishfry

It's really hard to critique the paper if you cannot set your personal beliefs aside for a moment and take a non-dualist perspecitve for a moment. The inability to do so renders yours objections invalid, as evidenced by all the strawman statements you make above.
Nobody is asking you to accept his conclusion or believe his premises.

Oh, and instead of justifying and supporting his computational consciousness claim, he blithely says it's "widely accepted." By whom?

The science of neural biology for one. There's possibly an exception to that, but I've never seen it: Somebody presuming your stance and implementing the scientific method to actually investigate it. Amazing that nobody tries such an obvious empirical thing.