If you make character optimization not pay off, however, then you have an impossibly generic system, because your choices don't matter at all.
I generally agree with your posts around this issue, but not with this particular point. Choices can matter although they have no implication for mechanical effectivenss: they can matter because of the change they make to the shared fiction. A game like HeroWars/Quest relies heavily upon this dimension of meaningfulness. A game like Burning Wheel or 4e relies upon it less, but still relies upon it to an extent: because 4e, for example, assumes that the GM is adapting encounter difficulty to reflect the level and general prowess of the PCs, to some extent optimising has no effect on relative effectiveness. The reward for having a tougher PC is a "story" reward - the fictional stakes are higher (eg you're fighting Orcus) - rather than a mecanical reward - the combats are still just as mecahnically challenging, because the GM has stepped them up.
A player might also get a degree of pleasure from the
absolute effectiveness of his/her PC (eg my 20th level PC can solo Orcus!) but if that's the sole or even principal pleasure you're getting from an RPG, my feeling is you're missing out.
There is a contrast here with a game like, say, classic D&D, where improved absolute effectiveness is expected to also produce a degree of improved relative effectiveness. D&Dnext seems to be going back this way, too, with bounded accuracy.
I think this change away from the 4e approach to scaling and setting encounter difficulties may make it easier to break the game via optimisation.
Making smart, not-overtly metagame, in character choices isn't the problem.
But for some 3E players, clearly it is. They make smart, not overtly metagame choices in building clerics (nothing metagame about taking Craft Wands and making WCLW), or druids (nothing metagame about befriending a bear, summoning a bear and turning into a bear), or wizards (nothing metagame about studying and memorising strong spells like Colour Spray or Glitterdust or Evard's Black Tentacles). And the game breaks as a result.
Add into that dumping or selling any interesting or unique magic item to get the Big 6 because they contribute all the time.
Again, that's not obviously metagame. It can be a smart, ingame choice for your PC.
And if you're playing at a table where everyone else (including the DM!) is playing the same way, it shouldn't cause too many problems.
My impression is that many complaints about 3E are that, in fact, when everyone
is playing this way it
does cause problems, because of imbalances of mechanical effectivenss across the classes.
3e won't break because grease is better than magic missile in many circumstances, nor because color spray is better than burning hands in many other circumstances. It won't break because giants are more susceptible to will-save based spells than fortitude-based spells or because their ranged attacks suck compared to their hand to hand attacks.
Again, as an outsider to 3E play, my experience is that, for some, these mechanical features on their own are sufficient to break the game - in the examples you give, for example, the superiority of Grease or Glitterdust means that wizards dominate over fighters against golems, and against giants - both categories of opponent that historically are meant to favour fighters over wizards.
All of that takes a particular psychological approach to RPGs and the rules that they use
<snip>
That's an alien psychology to a lot of people who play, who don't care about having a fighter with a 20 Strength at 1st level, who are willing to invest a skill point per level in Craft: food because they think the idea of being a half-ogre barbarian fry cook is fun, who would rather be a halfling rogue fighting with a small dagger because that's the character they envision and want to play even if they'll be an average of 1 less point of damage per hit compared to a short sword (which they can also use) and a whole lot further behind a medium sized character with the same strength but wielding a long sword, and who can look at the mechanical incentives to playing a certain way and turn their back on them because that's not the style of game they want to play, nor is it necessary to play that way to be successful.
The person you describe here is clearly metagaming. I'm not saying that as a criticism. But as an observation, it seems undeniable. The person is certainly not playing their PC when they choose to be a dagger-wielding halfling or a half-ogre fry cook. They're making a metagame choice about the persona they want to adopt.
The person who focuses on pushing the mechanics hard is not metagaming to any greater extent. They have a different metagame priority.
I'd prefer it if they didn't have to design the game defensively toward that play style - nerfing certain options because they could be otherwise leveraged too much. My impression of 4e's design suggests a lot of defensive design and that alienated (unintentionally, I presume) me. I much prefer that sort of thing to be done on a per table/group basis with the rules offering insight into relatively powerful options.
I don't think that's a fair diagnosis of 4e. It is not designed defensively towards a "push the mechanics hard" style. It is designed to welcome and support a "push the mechanics hard" style. 4e's design utterly takes for granted that the shared fiction will be shaped by the mutual application of the mechanics by all at the table, without holding back.
It has no aspirations to free-forming as an ideal. That is quite different from (at least some approaches) to classic D&D, 2nd ed AD&D and (also, perhaps) 3E.
I think pemerton has a point about point-buy systems tending to be up front with advice (they really do use advice rather than mechanisms) to keep PCs under control and under GM supervision with character generation. 3e, being the D&D edition most responsive to player choices when it comes to build options, could use a bit more of that
Here are some interesting comments by Ron Edwards on design, and being upfront about it (from
here and
here).
Character generation text and methods are extremely diverse within each GNS mode, which is one of the reasons I favor group communication during this phase of pre-play. For instance, some Gamist-ish games utilize point-allocation systems, which looks similar to the widespread method in Simulationist-ish games. However, for Gamist purposes, this method is all about strategizing tradeoffs, rather than establishing a fixed internal-cause to "justify" the character. Similarly, Gamist character creation utilizing Fortune methods isn't the same as the few Simulationist randomized methods - in the former, it's a lot like gambling, whereas in the latter, it's about a character maturing through Fortune's vagaries represented by in-game effects like culture, weather, disease, and so forth (e.g. Harnmaster). . .
As far as I can tell, Simulationist game design runs into a lot of potential trouble when it includes secondary hybridization with the other modes of play. Gamist or Narrativist features as supportive elements introduce the thin end of the metagame-agenda wedge. The usual result is to defend against the "creeping Gamism" with rules-bloat, or to encourage negatively-extreme deception or authority in the GM in order to preserve an intended set of plot events, which is to say, railroading. In other words, a baseline Simulationist focus is easily subverted, leading to incoherence. . .
Another common problem is rules-bloat, which usually creeps into Simulationist game text as a form of anti-Gamist defense. I suggest that adding more layers to character creation is a poor idea, as it only introduces more potential points of broken Currency. . .
My recommendation is to know and value the virtues of Simulationist play . . . and to drive toward them with gusto. Don't spin your wheels defending your design against some other form of play.
Powergaming
This technique is all about ramping a system's Currency, Effectiveness, and reward system into an exponential spiral. As a behavior, it can be applied to any system, but most forms of D&D offer an excellent inroad for it . . .
Powergaming doesn't necessarily destroy the enjoyment of play . . However, it's fair to say that Powergaming is only functional if everyone is committed to it, and it carries dangers of leading to Breaking (see below).
To prevent Powergaming, many game designers identify the GM as the ultimate and final rules-interpreter. It's no solution at all, though: (1) there's no way to enforce the enforcement, and (2), even if the group does buy into the "GM is always right" decree, the GM is now empowered to Powergame over everyone else. . .
Breaking the game is defined as rendering others' ability to play ineffective in terms of any metric that happens to be important in that group. Theoretically, any and all games are breakable: one can always sweep the pieces off the board. But I'm talking about doing so in the context of identifying internal inconsistencies or vulnerable points in the design, breaking the game by playing it and rendering the Exploration nonsensical.
Here's the key giveaway in terms of system design: it is Broken (i.e. Breaking consistently works) if repetitive, unchanging behavior garners benefit. The player hits no self-correcting parameters and is never forced to readjust his or her strategy. . .
Breaking the Game isn't quite the same thing as Powergaming, because once a game is Broken, the group rarely continues to play. However, the latter often leads to the former, because Powergaming reveals vulnerable points in game design that are then Broken. Trying to prevent this one-two combination of behavior has led many game designers mistakenly to provide endless patch rules, full of exceptions to cover the exceptions, none of which accomplishes anything except to open up even more points of vulnerability.
The challenge for D&D design is that it is trying to be all things to all people (or, at least, it wants to avoid a tight design focus). At least to date, D&Dnext seems to be looking to the "GM as arbiter" rather than rules bloat as the "solution". It will be interesting to see how that works out!