A number of posters have pointed out ways in which spell casters can contribute to the game otherwise than by dealing damage. That is true - in fact, I would have thought it is quite uncontroversial (even if someone thinks damage is
better, still it's obvious that charming a bandit king contributes to the game).
But that doesn't respond at all to the OP's concerns that fighters and other damage-dealer types get crowded out in a non-feat game by sorcerers, and in a feat game get channelled into a handful of optimal builds (optimal because of the feat support).
a feat like Great Weapon Master may be seen by some as a clearly superior option, and a player who doesn't select it is not being "optimal". But that's only when combat efficacy is the only goal.
For those who aren't as concerned with combat efficacy, or at least for whom it's not the sole concern, such balance is far less important, and certainly not a necessity.
This seems to miss the points that, at least for me, were the heart of the OP.
You are correct that, if a player doesn't care about damage, then the fact that a two-handed sword is better than knife-fighting is neither here nor there. But the player who
does care about damage can't easily realise that goal playing a knife-fighter. This is the OP's "Problem #2".
The "further flaw" that the OP identifies is that, even if we put feats to one side, the knife fighter is in serious danger of being eclipsed by the warlock or sorcerer, who can do the same sort of ranged damage as a knife fighter, or even better, while also having the flexibility benefits of being a D&D spell caster.
It seems to me that there are two main responses to these points (other than agreeing, as eg [MENTION=205]TwoSix[/MENTION] seems to). One is to dispute their mechanical foundations eg to show that, in fact, the rules support a knife fighter whose mechanical effectiveness is comparable to a greatsword wielder, a warlock, a sorcerer, etc. I don't think anyone in this thread has tried this in a serious fashion. (I feel that showing that the sorcerer has enough spell points for "only" 5 or 6 powered-up encounters is proving the OP's point rather than refuting it!)
The other is to show that
it shouldn't matter to the player of the knife fighter, who
wants to do meaningful amounts of damage, that his/her typical expected damage is less than that of other fairly standard builds.
I don't think this second response is hopeless, but I think it needs to be tackled head-on. Simply telling that player that s/he's wrong to want to do meaningful amounts of damage doesn't count. [MENTION=996]Tony Vargas[/MENTION] is coming closest, I think (with his empasis on "GM empowerment").
As someone who is mostly a spectator in this thread, I'm finding the failure to fully engage with the OP's claims a bit frustrating, as the second response in particular has the prospect of being quite interesting in bringing out some deep considerations in RPG play and RPG design.
EDIT: I wrote the above before reading [MENTION=23716]Gadget[/MENTION]'s post. I read that post as making a point at least somewhat similar to mine.