Total Annihilation and its ilk balance that by having the commander unit, which is game-over if lost. There are a lot of really underhanded tactics one can employ to sneak a win here. (My favorite is picking up the enemy commander with a transport and then scuttling the transport...) Plus the large army size means the player likely has exploitable gaps in their defenses as their attention is split so many more ways.

That said, BAR is trying to balance the game to reduce spam currently. Having an endless stream of trash units charging ahead is great for scouting, even if they don't do much damage.

I guess because economy still requires territory control, and a bigger territory is harder to defend ?

Also it's probably not a coincidence that these games have a lot of units dealing area damage, and most importantly, units cannot shoot through each other, and when they try it can often end in friendly fire, which makes snowballing weaker ?

Anyway, desperate situations where the underdog eventually manages to turn the tide are not uncommon. I even recently seen it happen in an AI vs AI match !

Zero-K solves the problems solved by popcap in other ways. Income tracks map control, and controlling more of the map is hard, so players that are ahead don't have the freedom to sit back and build up an overwhelming army. Armies are fighting or jockeying for position throughout the game and army size and composition keeps escalating. Defenders advantage shifts to attackers advantage as things escalate, so people can't just sit in their base. The overall effect is that, while there is a pop cap at the technical level, it is 10x higher than the unit counts that even the largest battles tend to equilibrate to.

Talking about the lack of popcap is a bit backwards in a way. There are just so many subtle things that mean the question of popcap never has to come up. Maps are pretty open, units don't fire through each other, AoE exists... the list goes on. A better question might be why games with a popcap couldn't be designed away from the failure mode of people sitting in their base making larger and larger armies. Having to put a hard cap on army sizes, that people regularly reach, is sort of a drastic measure.

On supply, I'd say what you said about supply is pretty important. But the aspect of it you bring up isn't limited to popcap. Eg C&C games have power plants that act as construction supply. Rebuilding your base is easier the second time around, as long as you keep your power plants. There is something similar in Zero-K with energy. Metal comes from points around the map while energy can be built in your base, and both resources are used for construction. When you take more metal spots you also need more energy, and the energy is nowhere near as vulnerable. That is the main supply-like comeback mechanic, which works since part of your economy is always vulnerable. Reclaiming wrecks for metal is another comeback mechanic.

Comebacks in high-level SC2 games don't usually happen as you say. More often, the losing party relies on their opponent overlooking some peculiar tactic like using DTs (for those who are not familiar with SC: invisible units that can only be detected with specific means). Otherwise a bad engagement very often means defeat in short/mid/long term.

Finally, ZK does have very long range static weapons that can hardly be countered (unless you destroy them). If the opponent "camps" in their base, you just start to build one that will either vaporize their army or buildings, or slowly wear them down. But really, camping is usually newbie-level strategy.

The first indicator of that popped up, when valve ended greenlight, and just greenlit all the things. Which propped Evolution RTS upon the steam front page for half a day, flooding the original lobby system and showing how inadequat the whole ecosystem was for that. The original Evo dev Forb learned from the whole mess after he returned from his day job.

Zero-k then learned the lessons, developed a ingui lobby, started the whole matchmaking and better server deployment, detached from the whole "one central server" thing of the spring eco system.

BAR did some graphical overhaul, with Floris, Beherith, Sprung and the whole original crew supported by new faces like Teifion and on and on. They optimized the Spring engine into a new version- more tailored to BARs needs and reworked alot of the stuff.

Its gpl open source, so the project order and who invented what is pretty flowy.. everybody copies from everybody, one progress is everyones progress..

Presuming that the main code contributions also come from the Swedish scene I'm not entirely surprised that this happened. It sort of fits my impression of the gamer/programmer culture here (although I can't quite express why)