Are We Just A Computer Simulation?

Great Debates
81

No. Reversible computation dissipates no energy. Yes, that’s just ‘in principle’ true, and any concrete implementation will always be imperfect, needing error correction and the like, but the energy consumption can via optimization be minimized without lower bound.

Relative to the parent universe, we might! And even in our universe, it’s not impossible to compute a sophisticated universe simulation in finite (observer) time. In fact, I’ll do you one better and simulate every simulatable universe in finite time, using only some modest physical speculation.

So, first, it’s trivial to write a short program that computes every possible program via an interleaving algorithm. Of course, this will not terminate in finite time. But that can be worked around: just have it orbit a Kerr black hole, then jump into it; if you time things right, you’ll encounter the computer after it’s completed infinitely many steps of computation. Bonus, you’ll also get to find out what happens at the singularity of a black hole! Sure, this proposal still has a few engineering kinks, such as the finite lifetime of a computer, and of a black hole. The latter might be gotten around by enclosing the black hole in an Anti-de-Sitter spacetime, where it won’t radiate; as for the former, perhaps you can encode the computation into an aperiodic time crystal, which could realize universal computation via Wang tiling. But yeah, bit of R&D might be needed.

Of course, this extreme won’t be approached in practice. Still, there’s no theoretical bound in principle to the amount of computation that can be implemented by such methods, or more conservative ones; hence, the argument that the universe can’t be a simulation because there aren’t the necessary resources is dubious at best.

82

Once again, Kurzgesagt has a video on the subject, laying out what would be necessary for our world to be a simulation,

83

I agree that PRNGs are not truly random, by definition, which is why their output is repeatable. However true randomness can be achieved by sampling chaotic physical reality. I’m not sure if this is what you meant by sampling. That it comes from the parent universe isn’t much of a problem - the simulation comes from the parent universe by definition.
I’ll evaluate the points in your next post if the half dozen speculations ever get verified. I think the point of the model is that this can be done easily in our universe as we understand it. Superficially it looks like it can, but not when you dig deeper.

84

I came in to say the exact same thing. There are fundamental limits to how much computing you can do with a unit amount of energy. Simulating an entire universe down to the quantum level isn’t possible without consuming all the energy in the host universe, assuming they are equally massive and equally complex.

This is an example of recency bias. When we learned about radio, everyone assumed aliens must communicate by radio. A few decades later we discovered that there were better methods. When Bussard came up with his electromagnetic scoop interstellar drive, everyone was suddenly sure that aliens were zipping around like this, and people started studying how to detect them so we could find what was ‘obviously’ out there. Then a couple of decades later we realized that the Bussard Ramjet couldn’t work.

We have only been a technological species for a couple of hundred years, and only developed digital computers a few decades ago. We are infants when it comes to really understanding and manipulating reality. We have close to no idea how any other civilization would behave, and a mere hundred years from now we’ll probably look back at some of today’s ideas as quaint, as we look back on the 1930’s vision of the future full of flying cars and underwater cities as quaint and hopelessly wrong.

85

Good point. I’ve been catching up on SF magazines from 20 years ago, and you’d be amazed at how many future civilizations communicate by fax.

86

Yeah, in Jumanji, when the original kids played it and the pieces moved, they said, “must be magnets”, and when the later kids played it, they said “must be microchips” or something like that. I thought it was a nice touch.

87

Remember in the 1980s, where putting an IBM PC in a pentagon summoned a demon, or caused time travel or something. It was like an updated version of thinking that cameras captured your soul.

88

For an argument in favor of us living in a simulation, we might look to the peculiarity of our current circumstances. Most intelligent lifeforms, if civilizations don’t typically exterminate themselves once they acquire the capacity to do so (i. e. if there’s no great filter that’s basically impossible to overcome), should either find themselves somewhere within a long period of equilibrium existence on their planet, having reached the maximum sustainable population, or as part of a vast interstellar community, should interstellar travel be possible. Only a vanishingly small number of individuals should experience themselves in the sort of circumstance we do, right at the cusp of maxing out population or before.

Thus, that we’re here, now, either means we’re extraordinarily lucky (for certain definitions of ‘lucky’), or it means that to the contrary, the typical experience of an intelligent individual is that of a member of a civilization just reaching its maximum (or, of course, that virtually every civilization capable of extinguishing itself indeed does so). But the sort of circumstances we’re in right now seem to be a good bet for being of singular interest to civilizations having reached the necessary technological maturity for creating realistic ancestor-simulations (should such a thing be possible, which, again, I doubt): they likely live in equilibrium, so their present isn’t much different from their past, up to the point before they reached either planetary maximum or made the jump to the stars. Furthermore, our comparative lack of technological sophistication means that there are likely few records from our time surviving; plus, this is a time of rapid and multifarious change.

So, if there are ancestor simulations, simulating the progress of civilization up to about the state we’re at now would likely be highly attractive to a technologically mature civilization; hence, the number of intelligent beings finding themselves within such a simulation might greatly outnumber those that find themselves within a state of technological maturity and population maximum. Thus, then, our experience right now would be the typical experience for an intelligent being.

Again, stated that way, that’s just not right. Reversible computing can be done with (theoretically) no energy dissipation, and in practice, there is no fundamental lower bound to the energy needed, although of course you’re going to run into practical difficulties eventually, like running out of space and needing to delete data (which does cost energy).

There is a connection between energy and the speed a computation can be performed, which has its source in the energy-time uncertainty principle, but this energy isn’t necessarily dissipated. Rather, you can use an energy storage, such as a battery, take the needed energy to perform an elementary operation (flip a bit, say) in time t, and return the energy to the storage.

89

Why? We’re living in a relatively boring time. 1500 - 1700 with the global expansion of Europeans and the clash of cultures involved seems a lot more interesting. So does the time when we finally expand to the planets and moons, if we ever do. For the past 50 years we’ve been sitting on our collective asses and playing video games.
And there are a lot more of us than there used to be, so the probability that we personally live now is greater than for someone back in 1550.
There is also an implication here that the people running the simulator wanted to get to us - but that assumes they knew the simulation would come up with someone like us. They would have saved a lot of effort if they diverted the asteroid and let the dinosaurs develop intelligence.

90

Because this is the time we not only have game changing technologies that are literally changing society at fundamental levels, we have a struggle between totalitarian authoritarianism and liberal democracies, but we are also on the cusp of myriad technologies that will, again, transform our world and society…at the same time, we are also on the cusp of potentially world-ending forces that are in motion. 1500-1700 have zippo in comparison to today. The real issue is that people so wrapped up in what seems commonplace don’t realize we are living in truly interesting times. :slight_smile: 100 years from now when someone is searching through the archives and sees your and similar posts about how boring today is they are going to be laughing their asses off…or rolling their eyes and not understanding how we could be so dense…kind of like we do today when we look at the stuff our ancestors believed and though was so solid and true.

Yet, today, someone like you or me or the average person in China can have their voice heard by 1000’s, 10’s of thousands, or even millions…something that even the elite then couldn’t do. Frankly, this is the perfect time for a simulation, as it’s as transformative a time in history as there has ever been.

Plus, when I die and go back to the real world I’ll just boot up the Roman simulation and go kill some Gauls or something.

Only if you presume that the dinosaurs are the real ancestors and they are running an alternative history when hairless apes (can you imagine??) actually become dominant. :stuck_out_tongue: Seriously, if they are running an ancestor simulation this would actually be one of the best times, as there is so much data available to them to use…enough that they could actually simulate you…or me…or the other posters in this thread as good AI versions. They would have all your posting history, plus whatever else is available about you online. Information about your job, maybe social media info and posts, tax records, tons of stuff about you and your personality. It’s a good time to be an AI…much better than trying to create someone in 16th century Spain from basically a few lines in an old book that is probably completely made up horseshit about conquering the wild natives in Peru or central Mexico. Most people in the past lived and died in complete and total obscurity, while today I’d say that in most countries at least some data is available about a large percentage of their citizens…much more about you and me and everyone in this thread than even the elite of the time period you underscored.

91

Well, uh, I suppose that’s subjective to a degree, but I think the only time in history where we’ve (so far) faced the threat of self-annihilation and made the first tentative steps beyond our home planet has some interesting features to recommend it. Let’s just say that I’d hope that there are still people thirty to forty years from now that complain about their boring times, but I fear such might be utopian.

But that’s really besides the point. The events of the European expansion are unlikely to be generic (i. e. a feature of most civilizations during their ascent to either planetary equilibrium or to conquering the stars), but facing the threat of self-annihilation and the social changes brought on by the mass availability of information technology are. You have to think about this from the perspective of a civilization that might’ve lived for thousands, or millions, of years in relative stability, either carefully shepherding their native planet’s/star system’s resources, or expanding to the stars; the question that’s gonna be of most interest to them is, how did we get here, and what were things like then?

Besides, by ‘now’ I don’t just mean the last 50 or so years; there’s no real computational shortcuts, so I should think that something like the 10,000 years from the emergence of civilization to its maturity would be the simulation’s target. It’s just that, in that period, we’re likely living right around the time of maximum population, making the generic experience of an intelligent being something close to ours.

92

We’ve had debates here about whether technological change from 1900 - 1950 was more impactful than change from 1950 - 2000. I’m not convinced of either position but I know that being born in 1951 gave me a much better chance of surviving (and my mother too) than being born in 1901. And the fight against totalitarians was bit more of a thing 80 years ago. We’re pretty much always at a cusp. We just finished watching an interesting series on PBS call “How the Victorians Built Britain” and the change during Victoria’s reign was as least as great as the change recently. And who knows what it will be.
I grew up with 6 channels of TV (black and white, and only so many because I lived in NY) long distance being a major deal, long air flights being a major deal, and of course no computers. 60 years ago I read Arthur C. Clarke’s pretty accurate predictions on the future of telecom - except for the one, in a story, where the plot point was that free access to pornography would make Western Civilization collapse.

I’m not getting why this would be a big deal to simulator developers. So, I could send “foo” to someone in China and the censor could send “bar” back. I’d also speculate that our current everyone is an influencer for 15 minutes culture makes each of us less influential than when Walter Cronkite meant something.

I think you’re being a bit pre-Copernican. The simulator writers would look nothing like us or dinosaurs. My point was that letting the dinosaurs be intelligent would cut down a lot on simulation time. If the universe is a simulation, it definitely does not revolve around us. In fact odds are no one has ever paused it to look at us. There are countless other cultures out there, no doubt earlier and more interesting, to look at. Maybe they set the next breakpoint at + 1 billion of our years and are only monitoring significant events, like blowing up stars or something.
I know the big meme in SF these days about building AI to simulate people based on their on line experiences, but anyone doing one for me is not even going to be close. I’ve been on message boards since 1975 and I have learned about online personas. I’m not even on Facebook. I bet the AI would think I really have a friend named George - a character I built for my Star Trek previews who got revived for my column.
I can buy an AI replica when you load memories into a computer. Online presence, nah. Or you’d need something like in the John Scalzi series where the Emperorx gets fitted with something that records everything they do.
Anyhow, if they are simulating us from scratch, what does it matter what our online presence is? They’d have access to a lot more than that. If you mean people from the future trying to accurately reproduce our world, no chance. Not everyone puts every fucking meal on Instagram.

93

I get the impression that you are backing off from a universal simulation to one just of our planet. Why? But I’ve never played Sims.
If we stay on this planet, I’ll buy that we are close to max population. If we settle other planets and even other stars, slower than light, then we are nowhere near max population. And I’d think the cultural divergence of colonies separated by long speed of light communications lags would be far more interesting than what we’ve got now.
We’re not all that special.

94

I’m not, no.

Maybe, but that would also be their everyday reality. The point is, that no matter whether a civilization stays on its planet, or moves to the stars, there’s a moment of phase change, and almost every intelligent lifeform will experience the ‘after’, and (sans ancestor simulation) only a vanishingly small fraction experiences the ‘before’; hence, to those in the ‘after’, the ‘before’ is of singular interest, and would be a good candidate for an ancestor simulation, should such a thing be possible, which I of course don’t think it is.

Another argument against the simulation hypothesis I haven’t seen advanced much is an ethical one: any civilization reaching the technological level to produce such simulations probably has at least some degree of eusocial ethics, and thus, ought to find it abhorrent to subject sentient entities to avoidable suffering; but it would be well within their means to either abstain from simulation, or make the simulations uniformly pleasant, hence, our suffering would be avoidable. So I don’t think it’s likely that ancestor simulations would be permissible to civilizations having the means to implement them.

That’s of course if what we see is all there is to the simulation. Perhaps things would be permissible if all simulated sentient entities got to experience subjective bliss in some form of life after death, or something.

But another aspect that seems to call the utility of such simulations in question is the fact that if the simulation argument is cogent, then each simulated civilization ought to assume they are in a simulation, and hence, behave accordingly—perhaps go on collective hunger strike to lobby the simulation creators to enhance their living conditions, or just fall into despondency because they’re ‘not real’. In any case, a simulated civilization couldn’t be reliably assumed to behave like a real one, in fact, should be expected to differ in behavior; consequently, such a simulation would be pointless.

95

I think you are way overestimating the uniqueness of our current situation, but that’s just my opinion.
But I like your point about ethics.
I doubt a hunger strike would work, since any culture would be an insignificant part of the complete simulation. Would they play the suffering forward. I’d hope that many, perhaps most, wouldn’t, which hurts the probability argument.

96

The point is, that if the goal of the simulation is an ancestor simulation (which is what most versions of the simulation argument assume), then if the simulation argument is cogent, each simulated civilization should wise up to the fact that it is simulated, and orient its behavior around that fact—whatever that may mean, it would defeat the purpose of simulating a civilization that believes itself to be real.

Then again, every real civilization, if the simulation argument is cogent, ought to mistakenly believe itself to be simulated—so perhaps, that’s a reason to abstain from creating simulations even if they should be possible: because it’s the only way any real civilization could believe themselves to be real, and base their actions around this view (again, if the simulation argument were otherwise cogent).

97

A post was merged into an existing topic: RangerLoops troll posts

98

I think you are missing the point of an ‘ancestor simulation’ unless you are using ‘ancestor’ to simply mean any organism on a tree of life. I’m using it a bit differently…i.e. some variant of homo-sapiens or a follow-on species that is wanting to simulate some period in our history.

Certainly, that doesn’t have to be the case, and we simulate worlds with orcs and elves and stuff now, but such a world doesn’t need the level of detail we see in our reality, while an ancestor simulation would. I was tongue in cheek btw about booting back up in the Roman period like it’s a game, btw, but if you were a sufficiently powerful civilization you could spend your whole life in a simulation, ‘playing’ as a real character, especially if you could compress time in the game.

Why build from scratch or whole clothe when you don’t have to? Do you not see that an ancestor simulation using actual data of everyday people would be a lot more interesting, let alone accurate than one that is just pure fantasy, or built on a few lines from a very few people?? I don’t care how smart they are, or how powerful their technology, there simply isn’t a lot of data about large numbers of people prior to our current period…so, anything you write is going to be pretty much made up or extrapolated using very small data sets. That is not the case today. But if you can’t see why that is compelling, far be it from me to keep beating this dead horse. :slight_smile:

99

The original simulation idea, from a long time ago, was simulating the universe. That’s where the probability argument comes from - with many simulations throughout the universe, and many levels of simulation, I can see that one might think it more probable that we are a simulation than not.
None of that works for ancestor simulation, so I did miss it. The massive effort it would take to run such a simulation means that there won’t be that many on one planet. If you are testing social dynamics, why such a big simulation? Starting with us is basically last Thursdayism. You are not going to be accurate, since you don’t have the information to faithfully reproduce ancestors, so why bother trying to be. Thus running the simulation forward from primitive man, whose bank balance you don’t have to know, seems easier.
Since the rest of the universe is basically a backdrop, with interesting things like supernovas and gravity wave generating events thrown in, it is really pre-Copernican.
The hypothesis is obviously unfalsifiable, but it isn’t even very plausible. It doesn’t use all the energy in the universe at least.

100

I don’t see why that would be the case. Pretend you want to draw an arbitrary section of a Koch snowflake or other fractal pattern, approximating it to an arbitrary scale. We’ll define computations with two atomic operations: line segments drawn and line segments erased, where the segment length is the smallest possible at the given scale. I postulate it would take less computation to draw only a necessary subset of the curve than to draw the entire curve.

Here’s an animation. I’ve chosen quadrant II as the section to be drawn, to a scale of 3 iterations. The number of cumulative operations is given at the top left. Unsurprisingly, drawing the entire fractal takes more operations (63 operations) than merely a necessary subset (18 operations).

I would like you to imagine a book of fiction with a branching narrative and six different endings. Please examine the model structure below.

A. Chapters I-III
      ││
      │└───────────────────────────────────────────────────┐
      │                                                    │
B. Chapters IV-V                                           │
      ││                                                   │
      │└──────────────────┐                                │
      │                   │                                │
      │             D. Chapter VII                         │
      │                   ││                               │
      │                   │└──────────────────┐            │
      │                   │                   │            │
C. Chapters VI      E. Chapter VIII     F. Chapter IX      │
                                                           │
      ┌────────────────────────────────────────────────────┘
      │
G. Chapters X-XIII
      ││
      │└──────────────────┐
      │                   │
      │             I. Chapters XV-XVI
      │                   ││
      │                   │└──────────────────┐
      │                   │                   │
H. Chapter XIV      J. Chapter XVII     K. Chapter XVIII

Reading a story is simulating it in your mind; this must be the case because you can predict future plot elements based on your mental constructions. There is hardly any calculation involved because most of the information is pre-arranged in an accessible format. The only conscious calculations you must make are the decisions at each branch point. And so, for this exercise, the atomic operation during simulation is the decision to choose between branches.

An example readthrough might be chapters I-III, X-XIII, XV-XVI, and finally XVII. But for convenience we can use the letter labels, that particular plotline would be “AGIJ”. Each label in a plotline, beyond the first, requires one decision: the plotline “AGIJ” therefore requires three operations. The ‘computation’ involved is based on the operations, and the operations that matter in this case are not plot elements that change over time, but decisions the reader must consciously make.

There is a way to reduce the computations at run time to an effective zero - simply by printing a book with only one plotline. All simulations can be cached.

~Max