Once again we are discussing the issue of cookie-cutter specs created by theorycrafters and the high-end, and those specs being enforced by the playerbase at large.. An article on BlizzardWatch started the current issue, and Talarian offers his viewpoint.
I think that most of the comments are approaching this from the wrong angle. Everyone seems to assume that the players who insist on cookie-cutter specs are being irrational and "mean", denying other players the option to be creative. But what if the players insisting on cookie-cutter specs are entirely rational, and entirely correct in their reasoning?
So let's start there. Why do players insist on cookie-cutter specs?
Well, let's say you're a raid leader and you have two applicants: Chris and Sam. Chris' spec is drastically different from the accepted theorycraft, as are his secondary stats. Sam, on the other hand, is texbook cookie-cutter. Which of the two is more likely to be the better player?
You may not like to hear this, but 99% of the time Sam will be the better player. And it won't even be close. Sam will probably do twice the damage that Chris does. Adherence to the cookie-cutter spec is usually a sign the player has done outside research, who at least has read a guide and knows the best rotation.
Essentially what the raid leader is looking for is a "Proof of Competence". A cookie-cutter spec is one such proof. Yes, this is unfair to good players who want to experiment, but from the raid leader's point of view, she cannot infer that. The information she has is limited, and it's best to follow the Proof of Competence, even if she occasionally turns down a good player.
So how can we encourage a wider variety of specs? The answer to that revolves entirely around the Proof of Competence.
First, you could substitute something else as the Proof of Competence. Examples here are logs showing good performance, or achievements. These are often harder to obtain, though. Requiring a cookie-cutter spec is better than requiring people to have already beaten the fight you are working on.
Second, you can not allow the opportunity to view the Proof of Competence. For example, you could not allow inspection of specs. But that doesn't stop people from just asking questions. Another example is LFR, where the group is automatically put together. It's much easier to take a variant spec into LFR.
Third, you could make the spec matter a lot less. Throughput talents almost always have a "right" answer. If all talents were utility talents, most people would not care so much. For example, SWTOR talents are almost entirely utility talents, and no one cares what talents you take. However, the downside is that they require something different as a Proof of Concept. In SWTOR, pick-up groups usually require that you have already beaten the instance previously (by linking the achievement), which makes life hard for newer players.
Fourth, you could make content easier. Arguably LFR and Normal Mode in WoW are like this. People are less likely to insist on a Proof of Competence when success is likely.
Fifth, you can encourage extended groups such as guilds, and diminish the viability of transient pick-up groups. The thing about a Proof of Competence is that you only have to demonstrate it once, at the start of the relationship. Once the other players are confident in you, you have a lot more freedom. If the great mage in your group wants to experiment with a different spec tonight, the rest of the group is often happy to let her try. People in established raid groups have far more leeway to experiment with spec choices than people who run with pick-up groups.
In conclusion, players are being entirely rational when they insist on cookie-cutter specs. If you want to allow your players the freedom to choose their talents, you have to address the need to prove competence. From my point of view, cookie-cutter specs are actually among the least restrictive Proofs of Competence. Pretty much every other option is worse.