Would that not then mean that a discussion of this type IS worth having?
Did anyone say that it was not a conversation worth having?
My remark about the Catch 22 was not that people shouldn't talk about the theory behind architecting a distributed MMORPG. It was an observation that theorizing at some point has to lead to action if the theories are ever going to be realized.
My contention was not that nobody should talk about global storage of character data. It was that a game has to already exist for storage to matter.
It doesn't even actually matter how one would develop a truly distributed game. "We" are not developing this distributed game. We are all dependent on Codewalker and Friends to either give us tools for content manipulation or to reveal that they actually have been working on an emulator in secret.
Let me put it this way - if Codewalker, Leandro, and Guy Perfect, et al, were all killed in a freak meteorite shower, where would we be? Could we make some sort of game out of Paragon Chat? (After a proper mourning period, of course.)
If the answer is yes then what's stopping anyone from doing that other than it being different from some other vision of a "CoH" game?
We're discussing distributed gane management but we weren't really discussing a distributed game. We were discussing a number of affiliated games interchanging player information between them. That doesn't require that the interchange exists from day one. It only requires that the interfaces to the storage are abstract enough that any number of storage or data interchange methods could be plugged into the game.
Now, if we're talking a truly distributed game, where there's no concept of "my server" and "your server" but only "the cloud" then yes, in that case I'd agree that the data storage is innately tied to the design of the game itself.
Otherwise, a game server, especially one that you want to optionally run "offline", can be decoupled from its permanent storage mechanism and would even necessarily have multiple such mechanisms available to handle situations of "full network", "partial network", and "no network".