You're mistaking a cooperative "game" for a cooperative "activity" here.* The incentives to play well remains regardless of whether you're playing in a cooperative or a competitive environment, that goes with "game" not with "competitive." A meaningful choice requires it result in different board states, and that players be able to discriminate between board states they want and board states they don't in pursuit of their goals.
Sorry, not my intent. The point was exactly that there's an incentive to make the best possible choices whether the game is cooperative or competitive, the difference is that bad choices, in a competitive game, will lead to you losing, which is to bad for you, but fine for the player that wins - depending on the player, it might be annoying due to the lack of challenge, or not, if it's just all about winning for them. Avoiding bad choices is part of the challenge, and making better choices than the next player is part of what winning is measuring.
In a cooperative game, bad choices hurt all the players, and making a better choice than the next guy is not necessarily going to help you win, while everyone making the best choices will help you win.
In an RPG, there's a whole extra layer beyond winning/losing - there's genre emulation, character concepts, storytelling, &c - that needs to coexist productively with the game elements. Balancing the game elements facilitates that.
Now I think it's a lot less clear that initial character creation (particularly at the level of something with such long spanning consequences as class selection) is the correct place to start putting choices in with that much impact, but I think it's completely reasonable to prefer a game where a player making choices during ability selection and level up will reliably outperform a random number generator doing the same thing, or even a simple algorithm.
Obvious examples abound in D&D, you don't go dumping both STR and DEX with a fighter, for instance, which an entirely randomized chargen might do. Dumping INT might be less destructive to a wizard guild, if you avoid spells that require attacks or saves, but a randomized chargen wouldn't avoid them.

Designing a balanced game certainly doesn't mean removing complexity, and optimization is always going to be possible... balance aspires to keeping choices viable, if done well, optimization ekes out a modest advantage that doesn't obviate the contributions of others.
(Optimization isn't the enemy of balance, it just leverages imbalance - in a very imbalanced game, optimization can just stop being engaging because it's too easy to pull too far ahead of both other players and the challenges the game presents. )
To avoid that, you'd either need to not print synergies into your abilities or remove choice altogether from the class chassis.
Removing choice in the name of balance is, like, really missing the point. You can cope with imbalance by formally or informally setting aside the bad choices. But it's not really an improvement, just an acknowledgment.
there's a whole class of competitive play wherein trying to win is a means, and the gameplay itself is the end. You'll get players who will happily offer advice to their opponents or essentially pause play to analyze the board collectively.
I have certainly had experiences like that. It's interesting that there's a genre of games that emphasize it.