My problem is this
Citing an example for your argument is well and good and part of debate. Telling people to stop trying to discuss the matter because you have stated your opinion isn't. After your Oblivion example, you should be open to people raising points about how accurate that is or isn't a basis for determining all/average disk requirements for next-gen games, not telling people to shut up discussing the matter. Having seen Oblivion I think that's not the be all and end all of what next-gen games will be looking like or needing. A load of generic caverns and medieval villages doesn't show the same level of requirements as a city-based game for example. As for procedural synthesis, saying it's going to fix all our problems is lovely, but how's about presenting at least some intelligent points how that can be done without heavily impacting the rest of the system? In a GTA/Getaway type game with up to say 30 people on screen at a time, driving down a street so new characters are appearing all the time, how much of an overhead is generating firgures on the fly going to be compared to loading pre-made characters off disc? Want to argue cost of producing a thousand different characters on that disc in the first place? Fine. How's about pre-generating those character procedurally offline and save them onto disk. Want to argue that compression will save the day and these thousand characters of 2MB models and textures can be compressed to useable sizes on a DVD? Okay, go do it. Give examples of current mesh compression schemes that are showing 100:1 compression or what have you. Don't cite a single freak procedurally generated FPS created for a competition as evidence that all games can be compressed and procedurally created to fit any size medium. Point of fact the world record for petrol consumption is in excess of 10,000 miles per gallon, but no-one in their right mind is going to reference that competition as what to expect from a normal-use automobile, Want to debate the relevance of a 2MB per character figure and suggest that in reality the most a game will need is 50 components that with randomized texture colouring and transformation at load time can provide all the people you could ever want with negligable processing overhead? Please do.
But don't say everyone should stop discussing this topic because you deem it closed and obvious, or everyone who disagrees with you is just plain wrong.