do you? i think he's just not familiar with the terminology (like most users). ofc early optimization is bad but i think he means 'built better from the start'. technically speaking, better design and architectural choices. some of his suggestions are quite good, point to fundamental flaws in the current system and thus totally fall into this category (i.e., can't be 'optimized away'), e.g. networking and seamless instances.
i actually agree with him, i don't think current elite is a good foundation for the next. practice makes perfect.
You should really ask dm06 this question, since it wasn't me who said that.
To be honest, Salt is never fun to play. The prime example of this issue is guild wars 1 to guild wars 2. That occurred because guild wars engine couldn't handle what they wanted to do, so they re-wrote it. From what I've seen of the Cobra Engine, it doesn't have those kind of limitations. One...
forums.frontier.co.uk
On seamless instances, I don't think that's an architectural or engine problem so much as a game design one: the game does already have seamless instance transitions in normal space as people get closer together and further away. The problem with supercruise is likely that at the great speeds, it can't quietly have already done the preparation behind the scenes when you hit 'drop' because at 60 FPS one frame distance at maximum drop speed is 16km. Likewise, as you enter supercruise you have to be detached from the instance instantly, whereas if you leave the vicinity in normal space it doesn't really matter exactly where it unmerges the instances as you'll be well out of sight either way.
So to solve that, a hypothetical game design needs a fast travel mechanism which:
- works in multiplayer so isn't time acceleration, with return-trip lag of a substantial fraction of a second
- allows the game to predict a player's destination reliably at least a few seconds in advance (local asset preloading is needed here as well, not just networking, or it looks non-seamless for a different reason)
- works with realistic scales including planetary ones (or takes the major "this isn't as good as Dangerous" hit of abandoning those)
(Fast travel is arguably the major impossible problem of space game design, even before you get on to questions of technical feasibility, and both multiplayer and planetary landings add substantial extra constraints)
On networking, they'd have had to do something very wrong indeed for the network code to be so tightly integrated into the rest of the game for "rewrite the network code" to be harder than "rewrite the entire game".