anyway.

This makes Chris go "Sure..."
Try reading different versions of the same epic- from different locations or times. The reason there's so many variations of King Arthur's tales, the Mahabharata, the Ramayana, 1001 Arabian Nights, or The Journey to the West is that while the larger plot arc appeals to people in general, there's often revisions to the smaller episodes, or even completely new content added according to the local culture who is doing the retelling. The general trend is to take the overall plot arc, then build in the details and parts within that- not start from the beginning and work to the end. For example, the Ramayana originated in India, but is popular throughout Southeast Asia. While everywhere you go, the general plot stays the same, you can see places where secondary characters become primary characters (Hanuman), or even new characters are added that are more in line with the local culture. The problem for gaming is assuming that a serial set of events will spontaneously organize into an epic structure. When you actually look at any epic- it's not a collection of history- at best its historical themes carefully reworked for maximum presentation and entertainment- but definitely not the same as serially creating fiction without structure to guide it.

This makes CS go "intriguingly similar to metaplot"
That description of epic is intriguingly similar to meta-plot idea (as in the White Wolf or L5R thingies), that I understand is generally hated. Question of execution rather than underlying principle? Could published meta-plot be done well?

This makes TC go "I'm not sure..."
I'm not sure he's refering to a published meta-plot as he is speaking about something created by the players during play.

This makes WMW go "Impossible Thing?"
Taking metaplot as a canonical limit rather than a template for elaboration and remixing will smash you into meta-Impossible Thing Before Breakfast territory in a big hurry.

This makes CS go "canonical limit problems"
(the mathematical pun is ther for no reason) Canonical limits might at the outside be able to do interesting things to a game, but yes, they are definitly going ot cause serious problems most time. That may be the main problem with meta-plot as it has been conceived in published material so far. If it were recognized more as the starting materail for a remix, then it might be more useful. Don't like that major character? Tweak them into something you do like and play them. Don't like that subplot turns out, take its start as your starting point and run a game about how it turns out differently. That might be closer to the way cultures handle epics.

This makes CS go "Outside"
By 'outside' I mean I can conceive of a game in which having to fit your actions within the constrictions of canon would be fun. There are online journal-based HP games that focus on canonical limits as the source of fun (I can maybe hunt up the names if anyone is interested). The players specify that their character is going to try to do such and such, which would violate canon, and everyone else tries to think of ways to either thwart their characters actions or to make their actions' effects not actually violate canon. It is not the mainstream source of pleasure in games, so I was thinking of it as the fringes of gaming, thus 'outside'. Just to be clear.

This makes Chris go "Uh, guys, take 5 steps back..."
I'm talking about sagas and epics as historically passed down from oral to written history- not gaming narratives. But if you wanted examples in gaming- take a look at Polaris or My Life with Master- both lay out that a certain event will occur (the fall of the civlization, the master overthrown), but within that structure, you're free to make whatever content works. Both include endgame mechanics to help the group apply pressure and make focused and tight stories within that structure.

This makes Chris go "Oh, and..."
Those games provide the outline, the structure, the "outside" and you fill inside of those boudaries. That's building outside-in.