tecanec

When it comes to software development, the usual procedure for making a feature is to start simple and expand from there, rather than make the whole thing at once. It allows the devs to debug early so they don't end up relying on buggy code. However, there are a few things that, if relevant, must be taken into consideration from the start. Multiplayer is one such thing, and there's a good reason why: Keeping two or more simulations synchronized across multiple computers is hard. Even the slightest desync could potentially cause a butterfly effect and must either be avoided or re-synced immediately. And given that the computers can't communicate instantly, said butterfly effect may already be spreading before one of the computers were even aware of the desync, further complicating matters. Everything that could possibly get desynced must therefore be possible to re-sync. If you haven't laid the groundwork for that from the beginning, you may well have to rework so much of the game's code that you might as well remake it from scratch. But laying the groundwork for something and fully implementing it are two very different things. The multiplayer build shown to us is most likely just there for testing purposes; to make sure that they're laying the groundwork correctly. The fact that collisions are working is actually major, as physics are one of the likeliest things to get desynced because of different approximation errors. As for how much that progress is worth, getting something as fundamental as physics to work is an exponential step forward compared to something more specific like docking. (Docking would certainly be major, both technically and gameplay-wise, but even that's a small step compared to physics.) Also, keep in mind that the scene in those pictures was likely chosen based on what they were testing and how they were testing it. Even the vehicles were probably just what they had on hand or were able to make relatively quickly. Realistic gameplay isn't really the point here. There's no need to send those vehicles to Duna just to demonstrate that collisions work.

More procedual generation. Part of the goal with most space programs is exploration, but what fun is it to explore if you can read everything online? My suggestion would be to make most (if not all) terrain procedually generated, and maybe add some procedually generated moons around some planets (especially Jool).