I've always appreciated Model Three, and because class dps specs are balanced so carefully these days and can be so close (not perfect, but close!) it could see a healthy return.
First point I'll want to try and make is that there will always be min/maxing - the developers can't fight it, and don't really need to venture out to try. There's a difference between end-game, hardcore, perfect group comp, four nights a week raiding and casual guild or middle-of-the-pack progression raiding, too. The hardcore raider group will always min/max their specs. This hasn't changed since BC or Vanilla. They still do it today. The middle raiders will still throw in what they can get AND their very best players, in addition to trying to make that perfect comp. Casuals who participate in raiding will still have a hodgepodge, but they'll have a model to follow. This still happens, and narrowing down the differences between how classes and roles deal damage (dots, procs, nukes, instants, and cooldowns) hasn't changed raiding trends too much, from my own lone and humble perspective. You've increased the odds that one spec might be "best" in a given situation with number balancing, but you haven't fought exclusion in hardcore raiding until the group buffs release months after the content drops, or until people get geared enough for the discrepancies in numbers aren't a big deal.
In a way, it won't matter that a Fire Mage can AOE DPS better than a Frost or Arcane Mage, because at the point where people begin to care more about their spec than their numbers, those people are already in the same group that would be using the LFR tool or grabbing the raid zone buffs like what were popular in Icecrown and Dragon Soul. The point is, cutting edge progression still min/maxes, and always will until each class/spec is exactly equal.
Second point is that "bring the class" is not always a bad thing. There's a certain amount of class pride in saying "yes, I am a Warlock and I can drain tank or range-tank that boss or add for you", or "I'm a Shaman/Mage, and I bring the buff that makes you great once every ten minutes", or "I am a Warrior, and I can wade through the heaviest blows the boss can give, even if I can't AOE tank to save my life". Class and spec identity have suffered from the design philosophies of Wrath and Cata (again, from my one, single, humble perspective) because there are very few cases where any one person can look at themselves and be proud of what they chose at the character select screen.
Side point incoming: There also aren't a lot of difficulties to overcome in current content where one can be proud of one's abilities, either, striking even more pride from performance-based awesomeness. "Bring the player, not the class" hasn't really worked as intended except to make all specs deal roughly the same damage, which is (maybe?) in part due to the maturing practice and tools used by the development team (Mastery, insight into cooldowns and spell coefficients, and how up time of spells/skills and any number of other things affect the overall dps of a spec). I'd be adventurous and say that "bring the player, not the class" hasn't functioned as intended at all, and has only worked to hurt class identity by making them more like Model 1, which, again, doesn't solve any problem. What has solved the numbers problem hasn't been pruning down the class differences. It has been careful work altering spells, skills, glyphs, scaling, and stat allocation on items. I think that this is true for both PvP and PvE, actually. Remember when not everyone had a CC or interrupt in battlegrounds, and some specs had huge burst, instead of good sustain or heals? That was probably okay. It's not healthy for arena, but there's one of my strong biases that I'll just have to admit I'm okay with.
Point three would be to consider the bigger picture (you guys already are, I'm sure). All of the Model problems are interrelated, so I can't suggest one without bleed over into anther's benefits and issues. Even if I were to suggest a return to something closer to the BC model (Model 3), it also comes with some weight bleed over from Model 4, which brings up the problem of one spec in pure DPS specs becoming alienated. With the number balancing game that the developers have now, and how good they've gotten at it over the years, for a class such as Warlock, the specs cans till be fairly close in terms of average DPS, with one being stronger in AOE and one being stronger at single target (perhaps Demo and Destro respectively), leaving the third to PvP. I'm not necessarily seeking a return to the one-spec-per-class-can-PvP ideal. I love seeing all different specs in battlegrounds, and with the new talents in MoP, I think that that's actually still quite possible. There are times where a stronger AOE spec might shine, or when the utility of high single-target burst will be better or more preferred than a high-survivability spec. Remember that some people loved high burst specs in PvP back in the day because they could guarantee a kill, even if they died shortly after.
My fourth bit here is one that's been griping on me since Wrath with DPS specs, and more on into Cata with healers and tanks: for healers and tanks, don't create roles to hold the specs, create specs that fill a role. The goal should not be that "X role can heal/tank, and Y fits in X role" but that "Y spec can perform well at X role, but that doesn't mean that they can heal/tank everything". Personally, I loved my Paladin healer. I knew she was a tank healer. I knew my Druid was a raid healer. Given very specific specs and builds, my Paladin could Flash of Light raid heal, too, even when she had only three healing spells and Beacon of Light, even before she had any AOE heals at all. It wasn't optimal, but I could do it. If we needed strong AOE heals, we turned to a Holy Priest or Druid. For strong tank heals or raid support we turned to Paladin buffs and hard heals, or Disc Priests, or a Shaman. Need an AOE tank? Paladin! In a pich, Death Knight! Strong magic mitigation? Soak it with a bear sponge or DK magic bubbles and self heals. Who's the best single-target? Damn straight, a Warrior with passive mitigation and active threat-keeping. Can a Warrior AOE tank? Ya, but it's harder than Paladin AOE tanking.
These were points of class pride, and sometimes points of frustration, sure. Number balance has taken care of these sorts of problems less with tank and healers than it has with DPS specs. It's difficult to make a good AOE tank or healer, for instance, out of a class with no AOE tank or healing abilities. But (but!) it sure did make them feel different. These days I can't hardly tell the difference between my tanks when I'm playing as a healer or DPS. When I play my Warrior I don't look at a situation and say "I'll need more help from the party here" or "I can handle this because it's my specialty!", and I miss that. I didn't appreciate the three-heal-model that Healers were given. The developers said that some people didn't enjoy Disc pre-bubbling and damage mitigation, rather than proactive healing. I'm one who would disagree. I did. I see that they're taking steps to put the bubble back in Disc in MoP. They said that pre-hotting wasn't a healthy or fun mechanic for Druids, and it skewed the meters. I'm getting more opinionated here, which isn't really productive, but I couldn't care about the meters as long as everyone is actually pulling their weight and we survive the encounter. Meters can't say everything, like the Druid is working his or her !@# off to keep hots on everyone to keep them topped off, while the Paladin is working his or her respective $%^ off to fill in the gaps that hots just won't cover. I just really want roles for healers and tanks to be broken up again. If you have to take an un-optimal healer once in a while, but you still survive the encounter, then there's no harm in it. That is to say, Paladins don't need 'dem AOE healz, dawg. Not all tanks need a taunt (Paladins did just fine without it, back in the day). Not everyone has to work the same. It's s'not fun.