I'm not particuarly interested in an argument about what you like about video games. It's just that the component of your argument that relies on the cooperative nature of the experience as an excuse to dismiss balance concerns that's deeply flawed.
You can tell that the balance of game states both with respect to historical game states and potential future game states (i.e. how a game has, will, or could unfold) is important to single-player and cooperative experiences with a simple application of reduction to the absurd: If you modify solitaire such that you can name yourself the winner every time you lay down a card, you have made the game far less interesting to most people, and insofar as we're talking about quality as a function of appeal, worse for it.
There's no reason to believe that increasing the order of complexity would change that conclusion. People seek carefully orchestrated challenges within a set of rules that are either sufficiently complicated or sufficiently obfuscated such that they cannot guarantee greater odds at victory with one method over another regardless of whether or not the game is competitive in nature. If this is not the case, you're probably talking about a craft or profession. No game obfuscates, confounds, or challenges perfectly, but all games should strive to. This is an accepted design principle almost anywhere you go, and it's why live balanced games - including D:OS 2 for most of its lifecycle - see so much effort spent on confounding, or increasing the number of, metagame strategies. (Both approaches are intended to give rise to a greater number of metagame strategies, however, as, again, no game can challenge players perfectly and an emergent metagame is a consequence of imperfect balance.)
Then you can look at it from a content perspective, as greater balance tends to have a multiplicative effect on the amount of content a game effectively offers without conditions that are detrimental to most players. For example,
if Rogue and Wizard were to be grossly overpowered, a conditional gate would be introduced for all other classes, which would look something like "I will choose [Barbarian/Warlock/Bard/etc.] if I am looking for a greater challenge," or "I will choose [Barbarian/Warlock/Bard/etc.] if my feelings about their creative appeal outweighs my interest in clearing the game efficiently." It's slightly more complicated than this since you introduce conditional gates for the overpowered classes, and you need to know a lot about player motivation, but that's where design is supplanted by user analytics, which we don't have access to. Anyway, contrast this theoretical state of affiars to one where all the classes are balanced, and those conditional gates don't exist, effectively offering a greater number of options to a greater number of players.
These gates can have their place, say when implementing difficulty settings for example, but generally speaking they must be implemented with great intention, and it will very, very frequently produce shoddy design if they sneak their way into a game by accident.
