Couldn't you make a parameter that, when a user joined a multiplayer game, would check your config and turn on shadows or reject your connection?
That would be a bad idea. The fact is it's the lighting that's slowing everything down.
Look at the characters:
Other than the texture detail and bump/spec mapping, the polly count isn't much different from Q3.
The detail in the levels are questionable as I've only played through 3 levels and I doubt those will make it (as is) into Doom 3 anyway. But they didn't have huge poly counts anyway. More than Q3 certainly, but not huge poly counts.
Based on the Alpha (I know "it wasn't fully optimised") most people with average PCs will have to sacrifice the lighting to get the frame rate. So would you like to be kicked off servers, because 640x480 wasn't an acceptable res and to get better you switched off the lighting?
The Doom3 engine is all about the lighting. What other innovation is there in the engine?
I've read articles posted in Bluesnews where the designers where asked why Doom 3 seems to be quite a different game from the original Doom series (re: slower paced and less monsters)
Their comment was it was partly due to the engine being unable to handle huge hoards of monsters at once.
Designing the levels and so forth where not so much about making the engine look good, but not to let the engine look bad.
After reading those comments (I'll see if I can find the link again) I realised the Doom3 engine was surprisingly restrictive for an engine developed by Carmack. To the extent that that the game had to be developed around it's restrictions.
The source engine appears to have more useful features (lip sinc etc) that would make designers life easier. It isn't a one trick engine: Unlike the Doom 3 engine.
Check this link:http://hatekill.yojutsu.com/?viewarticle=preview_doom3
It's very anti Doom3 but tries to justify why