Hacker News Reader

Comment by tlb | Hacker News Reader

An argument against simulation is that our universe seems gratuitously large. Why have billions of whole other galaxies in it? Given a finite computing budget, you'd expect whoever's running the simulation to want to simulate more planets with civilization and fewer giant dust clouds.

If it is a simulation, it must be doing some very clever lazy evaluation to avoid thinking about every quark in every atom in every star in every galaxy, while still producing results that astronomers can't distinguish from physics on earth.

View on HN

I suppose that really depends on what the subject of the simulation is. If it’s you personally, it only needs to simulate other scientists telling you about quarks, the solar system and the latest JWST images. If it’s humanity in general then other civilisations are not required. Of course it might be that this is a giant quark simulation and things like gravity, planets, life and humanity are an interesting quirk that’s arisen recently.

The behavior of distant galaxies actually doesn’t match physics on earth. It’s a huge open problem. The leading hypothesis is dark matter, but that resists experimental confirmation despite extended effort, and has plenty of doubters.

Besides that, I think we’re assuming too much about their motives and capabilities. Who’s to say simulating dust clouds isn’t the point, and we’re just an accident? Or that the point is a survey of the evolution of astronomical systems, or an attempt to predict the fate of the universe? And, since any technology that could simulate even this planet is completely beyond human comprehension, why should we expect them to scale up only about as efficiently as humans do, rather than, say, orchestrating some reaction that results in a computer that grows exponentially and boundlessly? Or that the purpose or mechanism are as beyond our comprehension as a machine that could simulate Earth? Or, if they do have a point of limiting cost, why should we have any idea where that is, and where the cost is so low that resources are spent like they’re free?

> The behavior of distant galaxies actually doesn’t match physics on earth.

How so?

https://en.m.wikipedia.org/wiki/Galaxy_rotation_curve

It doesn't say anything about the rules of physics being different that I can see.

>> An argument against simulation is that our universe seems gratuitously large. Why have billions of whole other galaxies in it? Given a finite computing budget, you'd expect whoever's running the simulation to want to simulate more planets with civilization and fewer giant dust clouds.

What if the simulation is done in a way like we render video game graphics. We never render stuff that is not "visible", well mostly to spare GPU and CPU cycles. What stops a super advanced simulation to simulate only stuff we can "see" or stuff we observe. If no one is looking at the Andromeda galaxy, no simulation needs to "render" it, saving processing cycles. Or maybe, the nature of the probabilistic aspects of quantum physics is an indication that that the simulators are trying to simulate in a "lossy" way, kind of like MP3 vs FLAC files.

Render distance is one pretty good explanation for the Fermi paradox

> gratuitously large

That's very relative. That never stopped us either when overcompensating with hardware. Now imagine some very advanced quantum computer (or something 10 generations after it) that can scale infinitely after you just run the big_bang.sh. Who knows, they maybe even make it more scalable an stable as it progresses, by some self-scaling optimizations, which always play catch-up with chaos.

You would still need some giant factory sized next-gen quantum computer but maybe that's just worth their budget, after extracting discoveries from civilization x or y.

I think the hardest part, if this ever happens, is to fast-forward just enough an cherry pick from gazillions of civilizations, all while hoping that they to not evolve quick enough to escape your quantum VM. This and maybe the part with shutting it down and becoming a destroyer of worlds.

> Now imagine some very advanced quantum computer (or something 10 generations after it) that can scale infinitely...

Yeah, um... about that assumption...

This solves the "too much detail" problem by assuming that they have magic hardware. That's absolutely not realistic in any way.

I mean, this whole thing feels like a substitute for God for those who don't believe in God. It's Someone who existed before the universe, and created the whole universe out of "nothing" (nothing pre-existing in this universe, anyway). Now you're giving Him infinite resources (or miraculous ones that can scale to infinity, which amounts to the same thing).

Then again, "infinitely" might not even be necessary (got carried away there).

Maybe our full universe fits predictably in some pre-allocated space of such an advanced machine/project and 1 million years on our part is a days worth of disk space on their side. An then there are retention periods too...

Regarding the "substitute for God" part... good point, but what if we really come up ourselves with such simulations? It doesn't mean we are gods (literally), it doesn't even mean we are ourselves in a simulation. Just that we found clever ways to simulate the process into replicating something that we judge as correct (ironically). Like a game nowadays but a trillion times more complex. Looking at how much AI improved recently, what is to say that 100 years from now we cannot simulate this and then cheat our future with data extracted from other civilizations, rinse and repeat, to evolve exponentially.

Sure, but it's still limited thanks to the speed of light. Perhaps the universe we're being rendered by is 20 orders of magnitude larger. We'd have no way of knowing. We used to think the world was gratuitously large, then the solar system, then the Milky Way, now the observable universe.

> you'd expect whoever's running the simulation to want to simulate more planets with civilization and fewer giant dust clouds.

How so? What if the goal of whoever is running said simulation is something else besides simulating life and civilizations or whatnot, and we are here because an unexpected result?

If it were me, as soon as I found a civilization in my simulation, I'd shut down all the dust clouds and run many versions of the civilization. The big limitation of studying our civilization from the inside is that we only see one path, so it's hard to tell what parts of history were inevitable vs. accidental. In a sim, you could fork 100 copies of earth starting from 1900 and watch all kinds of different futures unfold.

We can't know for sure what would interest simulation runners, but they must come from a technological civilization and would presumably find other civilizations more interesting than dust clouds.

Who's to say that they don't have plenty of other simulations for simulating civilizations, and we're not just an unintended side effect in a dust cloud research sim?

“Why does all this life keep showing up when I’m trying to simulate dark matter clouds. I’m this close to restarting with the parameters tweaked.”

You NPCs are quite amusing! There's a texture map for when I go outside and look at the sky at night, so I think there are stars. And then there's some cute large language model to generate books, magazines, TV shows and web sites asserting the existence of "astronomers" who have "observed" galaxies and whatnot.

It is kind of impressive when I go into the city and see all the NPC AI running in real time. But the whole Trump plot arc is too stupid to be real.

Reply on news.ycombinator.com