Is that something that people should simply not try to wrap their head around, or is there a better way to think about it than comparing it to an explosion (... which has things moving away from a center)?

relative magnitide: farther objects are smaller and dimmer.

What's funny is that these figures are so totally unintuitive when you look at the sky. For example, from looking at the night sky, you'd say the moon's projected size is about the size of a hazelnut, not the size of London (except when close to the horizon where it seems to be rather the size of an orange). Obviously neither case is true, but it's what you see when you look up.

Venus appears just about big enough so you can tell it's probably something in the solar system and not something else. Pretty much everything else you can see with the naked eye is just a "tiny dot". You would never associate any of them with something like "ping pong table" or even "soccer field", would you?

Guess where I have a problem with that is the amount of data that is needed to simulate a whole universe and how much data the simulation would have to store somehow. Will it actually ever be possible to simulate several universes like the one we can observe (and, more importantly, interact with)?

You don't store all of it.

Procedural generation.

Have a look at some of the modern terrain engines, hell yes it's possible. Actually I think it's inevitable. Some hacker in 3 or 4 hundred years will do it for fun.

It depends on what scale you simulate, and since you use words like "galaxy" and "universe", it likely matters very little.

We must not take ourselves too important. Most of your interactions don't matter an awful lot to the galaxy or to the universe. Not even to our planet.

Even though mankind has been working very hard to destroy the planet during our entire existence, we still haven't been successful. The entire existence of mankind is just a fly speck on earth's windscreen. It only hasn't switched on the wipers so far, because we're not annoying enough just yet. Nature on this planet will prosper perfectly fine ten million years after we have died out. Without us, and without all our great achievements. A couple of million years isn't a lot to a galaxy.

But even if we manage to completely destroy this planet (as in, doomsday bomb), the universe will hardly notice. Earth is a sandcorn in an unimportant solar system inside a not-very-special galaxy.

So if you simulate "galaxy scale" feel free to go procedurally and screw interactions, because your interactions don't matter anyway. Simulate interactions only on a scale where they are noticeable, e.g. on a space station or in a city on an inhabitable planet.

Have a look at some of the modern terrain engines, hell yes it's possible. Actually I think it's inevitable. Some hacker in 3 or 4 hundred years will do it for fun.

It'll start to become possible as soon as we're capable of easily working with 128-bit coordinate systems (I'm not kidding by the way).

It'll start to become possible as soon as we're capable of easily working with 128-bit coordinate systems (I'm not kidding by the way).

400 nanometers resolution is probably good enough for everybody

Sure, but the maximum range of a quad-precision float is around 1.1897 × 104932 whereas the size of the universe is estimated to be 4.35184307 × 1026 metres across.

You wouldn't even need a local set of coordinate systems within that :)