Those battles all take place under .25 or lower time dilation so that the servers can handle the load. Star Citizen plans on having a massive player count per shard under full normal time.

RSI's "plans" are less reliable than a politician's campaign promises.

O(n^2) is a pain. Just updating everyone with what happened in a tick requires 4,000,000 data points a tick at 2,000 players.

At 10 Hz that is 40 million updates a second.

There are caches and prediction that can help but I ignore all gameplay in that number, it is just flying around.

How do they plan on doing that though. There are like fundamental computer science problems to be solved, and it's not just like a problem you can throw a faster computer at.

"Trust us, we picked cryengine, a game engine with near zero multiplayer tech, for a game that is literally trying to break new ground in computer science. We definitely know what we are doing, and Chris Roberts is totally a great project manager that hasn't come close to killing nearly every project he has been in charge of, including several space games."

That's how.

In large EVE battles you don't really see ships, you see lots of little boxes on the HUD and a list of hostile targets on the overview window.

Combat is all about range, and if you're close enough too actually see the enemy ship models, you're probably about to explode.

I'll be amazed if Star Citizen ever comes close to the scale of EVE battles, though. There's so many technical challenges in making an MMO scale, especially when a game is so focused on graphical fidelity.

(Then again, EVE is a 20-year-old game at this point, and was designed for the limited PCs and bandwidth of the time. But I'm assuming that they're using modern top-end hardware for the servers, and they still struggle with the biggest battles)