Firstly I'd like to say: each to their own as far as play style goes. The following is informed by my play style.
Finding the right middle ground between player abilities and character abilities is hard. Resisting the temptation to use metagame knowledge is hard.
I think the best thing is swing between all points of the compass
as the situation requires. There is no one hard and fast rule. And it will vary with different groups of course. I myself am lousy at puzzles. Can't solve 'em, can't write 'em. I NEED to use my character abilities when they come up. But I'm a good bullshi... ,er, talker. If I'm playing a charismatic character I can just let myself go. (Bards are my second favourite class to play.)
In both cases the GM is free to play from whatever angle they like. I would be quite happy if the GM were to give me puzzle solving hints based on a character skill check or modified my Diplomacy skill check based on my speech. And I think that's fair, after all I am playing a make believe person in a make believe situation, there should be some sort of frisson between me and the character. Exactly how much depends on all sorts of things. Group dynamics, the player's particular skills/limitations, the GM's particular skills/limitations, the system (different games place different emphasis on the value of this sort of thing and this tends to colour the play style), the adventure requirements. And more I haven't thought of I'm sure.
This is where the skill of the players (and I include the GM as a player here) is important.
You need know what's expected of the game at any particular point. This is the meta game. And I think it informs player skill at all levels. It tells a player what stats they will need to focus on when they design a character. It lets player know an obvious trap or the likely outcome from a scenario. Which monster to target in a fight. It tells a GM what are likely to be sticking points for plot development and what sort of challenges are appropriate for the group. And it can completely short circuit a game.
All the participants need to know how much meta game knowledge they can use in the game. And of course this is different for all groups. My group are pretty good at not letting meta game knowledge interfere with the game. So much so that I can let them know about future events without having them use it to short circuit said situation. This has the advantage that I can drop hints about the likely turn of future events and this in turn builds up the tension and suspense. Good player skills leading to increased enjoyment for all. Hooray!
Another example of meta game knowledge comes to mind: I was running the guys through a small dungeon crawl (frost giant ice mummies in a tomb carved from a glacier.) The guys reached a door that had a bunch of powerful wards and sigils on it. A quick KS: Arcana check later and I told them that the wards were made to keep something in rather than to keep people out. They wisely turned around and walked away. If they had of gone into the room they would almost certainly have been killed by the beasty within. Now my players know that I will put in stuff like this. And they've released bad
from mystical prisons before (haven't we all?) They know that what ever is in there is going to be bad ass, otherwise why the mystical protections? I consider this an example of good use of meta game knowledge. The meta game coincided with the in-game just nicely.
It occurs to me I should probably be hyphenating 'meta game.'
It occurs to me it is 1.30am and my mind is starting to wander.
Night all.
