Anyway, then we get his definition of metagaming. Using any knowledge the player has instead of knowledge that the character has available.
I have no quibbles with this definition. However, it makes metagaming impossible to avoid. Unless the DM and players have sat down and discussed for hours in minute detail every experience the character has had, every story they've ever heard, etc. how can we really know what the character knows aside from the limited information given by the DM when setting scenes?
Yes, there are ways to roll the dice to see what a character knows. But is the player or the DM tracking the results of each of these rolls? Some may. Most don't, in my experience. So it will be inevitable that a player will need to use some knowledge that they possess that their character doesn't from time to time.
Around the 3:19 mark, he starts talking about Perception checks to find a secret door. If the player rolls it, and rolls low, the player knows there could still be a secret door there. Asking another character to check is a form of metagaming, because if you had rolled high and failed, you'd be confident that you don't need another PC to check as well. [Relevant to the yet unfinished discussion on secret or open die rolls.]
At 6:40, he begins his discussion of whether metagaming is good or bad. First he gives the extreme views: any metagaming at all completely ruins the game, or meh, metagame away.
After saying metagaming everything is fine if the DM/group is good with that, it violates the concept of role playing. Here, I'll disagree. From what I've read, Gygax and Arneson didn't really care one way or the other how "in character" the players were in their original Blackmoor and Greyhawk campaigns. And clever thinking by the player was something to be rewarded. I could be interpreting what I read wrong, but the amateur thespianism that Luke seems to believe is the heart and soul of role playing was not part of the hobby in the beginning. So when he claims that metagaming is not the way the game was intended to be played, I think he's off. A certain level of metagaming is expected.
Now, Luke goes on to say that he does think some metagaming is acceptable (around the 8:45 mark). And funnily enough, I think he's got it backwards here, too. He says that players knowing that encounters are balanced for them is a good thing, because otherwise they'll run in fear of unknowns. My West Marches group has been a lot more cautious since they met a wight that killed one PC and drained another before they took it down. And in my opinion, this has enhanced the game for them. They need to approach encounters carefully, see what they can learn, and flee if necessary. And they're not completely afraid of everything. Recently, groups have charged in to an intellect devourer lair in one session, and stuck around to defeat an aboleth after they learned it wasn't just a trio of nixies in the river. It hasn't made them afraid, it's made them cautious, which is a good thing.
At 9:25 we get his next acceptable form of metagaming, which is letting PCs adventure together when they probably shouldn't. Like the paladin and assassin in the same group. Now, AD&D didn't allow this to happen. By the book, the paladin would refuse to join the group unless the assassin was left behind. Modern games ease up on the restrictions, meaning this form of metagaming is only necessary in these editions. I'll actually agree with Luke on this point, though. I never did like the overly restrictive AD&D alignment interaction rules. If an assassin's talents are useful, and a paladin's talents are useful, why not have them team up? Their interactions about how to approach the adventure will hopefully liven things up rather than be a drag.
Next point -- why not form a large party? Why not hire hirelings and retainers to help increase the party size? And all I think is, that's smart play, and not at all metagaming. The fact that there is strength in numbers is something any character in any sort of world should realize. And in old school play, it's just what's expected.
From around the 10:50 point, he gives his solution to the metagaming problem. First, pick your battles. Solid advice. Even if we disagree about what is good metagaming and what is bad metagaming, knowing when to stop it and when to let it slide is good advice. Because, as I said above, it's nearly impossible to avoid metagaming by the strictest definition because it's impossible for us to know everything that our character knows.
We also agree that we need to remember that this is a game. And while he thinks "having fun" is paramount, I think a big part of the fun of D&D is figuring out a challenge presented in an encounter. And often that involves a clever idea which is a form of metagaming. This could be assessing a tactical situation in combat, or finding a non-standard use of a spell or magic item, or whatever. It's highly likely that the player is considering the situation as a whole in these instances, not through the lens of their character's in-game knowledge and intelligence.
Finally, I like his proposed solution to the metagaming problem. No matter where you fall on the "metagaming is bad" spectrum, having a conversation with the players and letting them try to justify the metagaming is a good idea. And since it's just a game, letting the player have the final decision about whether to metagame or not is probably a good thing, too.