Hello, I wanted to aknowledge the nice answers and input there and then go on with more precisions about some points I've made in the OP and some suggestions to boot. 

It seems howewer that the thread has become about the hardcore/multiplayer vs casual/solo userbase of the game and the constraints on the game itself, I may as well address this beforehand. Sorry for another long post (and fear not, others are to come... later, need more dust)

Nor does everyone expect gameplay to serve narrative  ;) Our job is to find a decent compromise between the narrativist and gamist players.

From my concern you can infer I'm on the solo player scale. This is not the first time (nor the last) the issue is raised for such a dual game. That interesting features in the solo game would be impossible to balance was recently questioned in this RPS entry. While the author was advocating for imbalance by design, many commenters felt that balance could be achieved without symmetry. 

The comparison is not entirely fair given the TW:Warhammer has level-design and ES2 doesn't. You'd be surprised how much a hand-crafted start-location, with choke-points to defend or wide open expanses to escape across, contributes to the feel of a faction. We want ES2 to work as a competitive game, so it needs to be more or less fair. If we abandoned this rationale entirely there would be less need to balance the factions against each-other, so we'd be more free to push their asymmetry. Our player-base is not a monolith though: different people play for different reasons, and while we can't always please everyone we do need to make sure that nobody feels utterly abandoned by our design decisions.

You addressed the question of starting parameters. Whereas the tradtionnal random 4x start with anyone with a similar starting point is clearly the goal here, level design wasn't really what I had in mind, but more like unit rosters and campaign mechanics. The fact that humans/vampires and dwarfs/greenskins can't even share the same territory is too an interesting element, that doesn't really dwell on a fixed level design (think planet that may be mutually inhopitable for two different factions, thus reducing territorial friction between them)

I wouldn't consider TW:WH multiplayer balanced. Never tried it but 1 ) it's small subset of the game beside vs campaigns which may be an oddity 2 ) from most of what I've read, the factions aren't balanced. But the important point is that people playing these online battles do know this. Balance is still a workd in progress, and many units have been tweaked with that in mind, though, which means that CA hasn't given up on leveling the field.

That said, since time and resources are finite, each specific plead in this early access phase is gonna work against someone wishes. I know a subset of player consider the multiplayer as the "real" version of the game, evermore since AI has been a letdown in many 4x games (no need to evoque some recent release, ahem). But at the same time it's difficult to deny that 4X are promising a great deal of a single player experience and this is how the lion share of the players will experience it. Multiplayer is both a very time consuming ordeal and one that suggest a sensible experience of the game itself. This is utterly subjective, but I consider the balance in this context has a suspicious topic. That's not a freepass to ignore blantant issues, but the basic concept of different factions bonus and worse, diffirent starting conditions. Of course, even if "competition" outcomes can't be predicted as certain as with a real symmetric field, a tight internal balance can certainly ease things out, lest we end up with garbage or banned factions for instance.

My pet peeve however, is that the narrative and vivid life presumed to be intervowen in endless space 2 is hardly going to thrive with such an objective. Two reasons for this :

immersion may arise of kinda unique tools at each faction disposal : those can have a ton of counterintuive collateral effects on balance, making the whole refining process a nightmare

and now something often forgotten : fun can arise from the unexpected and yes, the unfairness of the game. You can either consider this kind of simulation as way to apply a plan and seek interesting results, a philosophy highly compatible with the multiplayer/competitive mode, or a "deal with the hand you're given" simulator, a sort of survival game with an empire instead of an individual. These elements of course would have a difficult time being a part of any multiplayer game.

For this second point, a good compromise stands with game options : let people enable possible unfair options 1, 2,3,4.... as they see fit. The issue however is any toggable parameter creates it's own pool of stuff to check for bugs and blatant gameplay issues. This keeping in mind that there are so many others more "mainstream" parameters values to keep in mind (galaxy size/opposing empire/resources, etc)

Now I may be wrong, but even though the goal of an interesting multiplayer and game geared toward advanced players is a noble one, this could still result in possible blandness in the long run. That's not to say stupid things ought to stay for the sake of creating a wild game. One peculiar element are the fearsome positive feedback loops inherent to 4x games that can wreak havoc once players find way to abuse them. That said, if anyone is afraid of some suggestions being detrimental to a fair competition experience, they may totally express this, but this is a mostly a topic about the feelings conveyed by playing the game, rather than the specific rules one has to master. I would even add that rules have to be judged by their consequence and intent, because for instance, many DD here have left me wondering that while they expressed the features, they didn't really paint the expected outcome (especially for combat) 

I'm kinda spoiled by wildly assymetrical games, especially fall from heaven, and an approach defined by adding fun sounding elements, even if the AI/game can't really handle them. It's not something I would advocate of course, I mean, don't we hate 4X game whose AI can't handle their own mechanics? 

Balance, to me, is one of the many ways to maintain immersion. I would stop to talk about balance right away if you think it's OOT, just ask :)

tl;dr : balance works toward AND against immersion at the same time, but still is related to immersion through "player's agentivity".

Player's agentivity is, in 21th century d&p roleplaying gamedesign theory (this half-crap / half-great GNS... Sorry), the set of means of control / degree of freedom / decision questioning a game provides to a player. The full experience and freedom provided by the game to a player, if you want.

It's not "the more the better", it's "the prettiest the better". Lots of agentivity is less constraint and tension, and well design agentivity mean tension (and with some tension a game is better).

Or else pick a pen and a blank paper and play pictionnary without cards & timer : Not. That. Great.

What I'm trying to tell with my poor english is if:

a- player's immersivity is a representation of a player's agentivity (Immersion can be sumurize roughtly as just the "player's agentivity in their head" or imagination or sentiments or whatever : this is the arty part of the experience)

b- unbalance nerf player's agentivity (I'll explain behond).

so c - unbalance somehow nerfs immersivity.

a+b =>c

BUT :

b' - unbalance create emergence, and emergence is one of the best fuel to a player's agentivity.

so c' - unbalance should also somehow boost immersivity.

a+b'=>c'

But since c is the opposite of c', how can this be ?

Both are real. Like twilight is day & night, beauty & ugly at the same time. Let's use an exemple to show why I say this.

Say we are playing a very unbalanced Street Fighter game (this game is interesting for a lot of reasons), called Hyper Street Fighter X: Turbo Dan Final Edition '' Nitro.

And the best character in this HSFX:TDFE''N have a very fast / unparable / easy to reproduce and fatal special move.

There is a differentiation between MP and soloplay of that game.

- In MP, most of my MP games would result in fighting this character with mine, whether I choose to play the game with him or not. Just because people likes to win.
Hence the MP would end being re-match after re-match this character vs mine, always the same, exausting the immersivity of this matchup until ultimatly I quit the MP play.

- In soloplay, I can do whatever I want since I am the one that choose my adversity, so here the immersion is untouched. Provided like you say, the AI is intelligent enough to profit of all the character's asymetries toward the one I choosed.

This is a+b => c.

In Street Fighter, though, I could use Dan.

Dan is the weakest and most terrible character of the roster of all Street Fighters he's been so far.

He's ridiculously weak.

Capcom did not made him only for comedy purposes. Capcom's decision to create Dan was pure genius.

He's here for human balance. Because even if the game would achieve, in a perfect world, the "perfect balance", even if the game would be symetric : there is both the player skills to throw in the equation.

Dan is weak because he is supposed to be picked by a very talented player against a newer player. So the newer player can sometimes win and enjoy the game.

Dan have also the purpose of "if I can beat the game with him, with one hand while listening the last first second album of Justin Bieber reverse, then I'm DA WINNER 100% platinium (self made) acheivement".

This is gameplay emergence here. Big time.

This is a+b' => c'.

This is why lots of games have a different way to produce balance : one for MP, one for solo.

While you can see all this as contradiction toward agentivity, this is just dozens of decisions Amplitude's gamedesigners have to make their mix a pretty thing. And with pretty things, there you have your "best" immersion.
Making attention to the "perfect balance", "broken balance" & plastic boobies of the uncanny valley (I may have sumurize this: sorry, this is allready tl;dr... I meant that extreme measures toward the esthetic sometimes only please 1 guy in the world).

So:

- balance, to me,  in MP does not create immersivity, but through agentivity helps to maintain it.

- unbalance, to me, in solo, can work toward gameplay emergence, and this also not creates but through agentivity helps to maintain immersivity.

Now, if the game were to be balanced, I think it should be balanced toward new / mid field players because of a lot of things.

It's not Street fighter (or LoL or CS or even Starcraft, Total War or whatever), yeah this is real, mainly because a match here is 3 hour-ish long or more. So if an hardcore fanatic of this game would clock say 1000 hours, and that would be just 333 ES2 games.

With 333 games, you're still a noob in any of those highly competitive game.

It's a lot less than the 12000 games, successfull or failed experiments, lucky match-ups, bad beats, trainings or just drunk matches of Street Fighter you can also make in 1000 hours, having also 12000 win or lose results that makes you collect 12000 feedbacks toward your strategy.

There, you would just have 333 feedbacks of your early game, 333 feedback about your lategame and 40 plays with each faction. With wastes that takes a lot of % because of a noob playing vs noobs. You game IS to be immersive or you'll quit.

333 games would be just the total feedback of a noob/early mid-level street fighter player.

That's why I think balancing the game for these dudes is the good move.

I somehow "can" imagine using ES2's would-be Dan for a couple of games for some hour-ish, but not for lots of games. So the ammount of fun this would provides is not the same when I pick Dan to fight my brother "better at Street Fighter than me" and still win by making him drink beers before the game (I also can't imagine maintaining the level of daylight robbery this trick need for 3 hours, this would be expensive in beers).