The main premise is that many archetypes aren't effective because they don't do enough damage.
And that's not a claim that DPR is king. It's a claim about the failure of those archetypes to contribute to play in their main dimension of (possible) contribution.
If someone (you? some other posters?) think that the main contribution a fighter makes to the game is
not damage, or some other combat function (like damage soaking or avoidance) that is at least commensurable with damage, then run the argument! I think it could be interesting.
But that potentially interesting argument isn't advanced by trying to rebut a claim (purely hypothetical in this thread) that DPR is more important than anything else.
(Katharine Kerr tackled this issue in an article in Dragon #95, whcn she tried to address the role of fighters in an extension to the XP system designed to reward non-dungeon-exploratory play. But I'm not sure here approach in the 1st ed AD&D context would still work in the 5e environment. Eg she assumed that a fighter would be the best in the party at intimidation. In 5e there's a good chance that's not true.)
I think it's because the definition of "viable" is dependent on whether you're the type of player who builds their character towards a function, or towards a concept.
I see the heyday of the "concept:" approach as 2nd ed AD&D, although it certainly seems to be on he upswing at present, and I'm guessing quite a bit of PF play must focus on it also (hence a good part of the enthusiasm for the range of options (= concepts, in this context) that PF supports).
I know three main ways of approaching "concept" play.
One is the "indie" style, which, of D&D editions, only 4e embraced. This style gives the player resources (generic ones, or ones that are particular to his/her PC build) that enable the player to make mechanical choices that will realise the concept. Come and Get It is the best known example from 4e. The "through death's eyes" move in Dungeon World is another example ("Name an NPC who will die. The GM will make your vision come true.")
If the resources are very generic, this can become a free descriptor approach (like eg HeroQuest revised; and Cortex+ Heroic comes pretty close) where the descriptor is important for fictional positioning and narration of consequences, but has no bearing on prospects of success. (So eg being a knife fighter compared to an archer is relevant to permissible action declarations given context, but does not effect your likelihood of killing someone if an attack is declared.)
5e (like AD&D 2nd ed before it) eschews this approach for non-magic-users.
The second approach is how 2nd ed AD&D seemed to do it, and is what [MENTION=996]Tony Vargas[/MENTION] seems to have in mind when he talks about "GM empowerment". The GM manipulates framing, and possible outcomes, to ensure that the concept is realised. At the limits of this approach mechanics don't matter much, and the PC build is mostly a signal to the GM as to the desired manipulations.
This is the exact antithesis of the "indie" approach.
The third approach is the one that isn't working for the OP: you build your character according to concept, the GM applies the mechanics more-or-less at face value, and we find out whether or not your concept is mechanically viable. I think this can work in systems with a relatively high degree of "sim" in their DNA (eg RQ, RM, Burning Wheel) because those systems will try to reflect eg the fact that, in the real world, you can kill someone by stabbing them with a knife. In D&D, though, I think there is more scope for some concepts to be crowded out by eg the hp and damage dealing rules.
This is another way in which D&D continues to support wargaming - because (outside of 4e) when GMed in a more referee-like fashion it pushes towards "build for function" rather than "build for concept", and that is a very wargame-ish thing to do!