How Big Should Space Be?

How big do we want space to be?

It is entirely possible to simulate entire galaxies:

That does come with a couple of drawbacks though.

  • It’s far too much for a player to ever interact with
  • Those simulations entirely ignore anything not being actively rendered, which makes it unusable for the kind of game space stage is.

So if that is way too big, how big should it be? Stallirs sized? How big is big enough?

It’s somewhat annoying that there isn’t a happy medium between solar system and galaxy. They are so, so different in scale.

Im sure this’ll be though about more later, but for life on other planets, it may be best to have other worlds be uninhabitable except for the random empires you meet (each developer regardless of role could do a full playthrough up to the space stage prior to full release for these empires perhaps). That is just my idea though

I’m not really talking about life on other planets, I’m talking about the number of stars themselves, which a player should be able to visit (Even if it is visit in the “I’m orbiting around it” sense.)

Empires should 100% be procedural though.

I’m asking here, how large should the visitable galaxy map be?

1 Like

I’m sure it’ll be a few thousand stars at least, but the max size im unsure of

I think the Stellaris developers probably took a lot of time to balance the size of their galaxies, so I think we should start off with a similar size. Realistically sized galaxies has always been out of the question as at that point we’d just have a few bytes of storage that could be used per star. And the player would have no way to interact with any kind of significant fraction of stars.

2 Likes

Actually, there is usable mediums in scale: Star Clusters!

For example: Messier 67, an open cluster in the Milky Way, has only 1000 solar masses worth of stars, about 500 stars. Even better: it has over 100 stars similar to our system; it is in fact so similar that some theories about our solar system being ejected from this cluster have been presented, though it is unlikely. So having the setup for Thrive within an open cluster like this would be quite realistic.

Here is M67 in SpaceEngine: (Top-Left)

This open cluster is a fair distance from the galactic plane, so it is relatively isolated, though there are still other stars around. But we could offset our open cluster enough to be isolated to the point where we could have a hard limit of traveling outside of the cluster without it seeming unreasonable.

If we wanted more stars, then an open cluster like Berkeley 29 would work. This cluster is situated very distantly from the galaxy (While still being a part of it), so this is a good example of a well isolated cluster.

These star clusters could scale up to thousands, or even millions of stars, so we can use real scales while still being able to pick what number range works for us.

Stellaris has 200 stars for Tiny Galaxies, 600 for Medium, and 1000 for large, so this matches really well with the smaller end of open clusters.

Now, I actually think that we can do much better than this. While a 400B star galaxy like our own would require 400MB per star with a single byte per star, it only requires a single 64 bit integer to store the count itself in a single number. We would only generate the star systems that we need on-demand, and for simulation purposes we should be able to simulate distant space societies using more abstract counts, like recording that a specific empire has 10M star systems, and 25M terraformed planets, and then simulate them using that along with the details of the empire itself. Perhaps each empire’s home system would need to be generated entirely to get a good handle on the empire’s society and history, but until the vast majority of the individual systems are directly observed by the player they would not need to be generated specifically.

This is exactly the same problem that is faced in Society stage, as an Earth Scale World is far too large to simulate it to its smallest scale across the entire World. There are around 10K cities on Earth right now, and likely over 100K towns. This would also be impractical to generate each urban center at an equivalent, exhaustive depth. The problem is even larger for individual structures, one estimate gave over 2.3 billion houses alone on Earth. The vast majority of cities would never be directly observed by the player, and the simulation will only likely care about abstract details of each urban center, like how much population all of the cities in a region have, the breakdown of different sized cities (village, town, city, megacity, etc.). There are lots of values the simulator will need to know, but the exact configuration of a specific urban center would only have to be known if the player visits it, causing it to be generated to a higher level of detail, or the simulation needs a higher resolution look at the area, like if the player’s society is starting to expand in that direction.

I will have to discuss in more detail about the generation and simulation of realistic scale areas, as I think this problem repeats itself across the game whenever we want to have both local detail and large scales combined. Microbe and Multicellular stage currently avoids this problem by locking off each section of the World into separate, discontinuous areas. But, that solution will no longer work when the real-scale Worlds are reached beginning in Macroscopic, where the World is fully continuous and any global scale of detail would be incredibly sparse if not assisted by intelligently applied local levels of detail.

All that being said though, if this system of scale is unable to be figured out or for gameplay reasons we want to limit it, then we do have options for real structures with which to set the Space Stage in.