>Two things had become constants at id: the lack of interesting games, and the boundary-pushing tech. Lets be honest, the only thing at id that kept it notable was Carmack. And I say that with a crushed, broken heart, as one who's run a TF server, mastered the trick jumps, and played thousands of rounds well after Quake was out of its prime.
Indeed. What was remarkable about Quake and Quakeworld was not the single player game (though lord knows I've played it through enough times by myself and in co-op) or the story, but the graphics technology, the client-server architecture (which *still* hasn't been beaten today, IMO, - no other modern game lets you move as fast as QW), and the ease of modability. QuakeC is a terrible hack, which is why they dropped it in Quake 2, but it had several important advantages: almost anyone could pick up the source code and mod it (leading to Team Fortress and then CustomTF), and since it was all run within a sandbox, you could download executable code from the internet and run it on your server without risking compromising your server. Quake 2, with its DLLs, didn't have that protection, which is one of the reasons why I stayed with Quakeworld.
Because QW was sandboxed, it was theoretically easier to debug, but the aforementioned hackishness of it meant that in reality debugging the thing was a nightmare for several important classes of bugs. I remember spending hours looking at where my code would crash, putting in sanity checks everywhere, and then having the problem turn out to be we were exceeding some internal limit in QuakeC. That the compiler would just silently ignore. Or entity overruns. Or the netcode limit on sending updates. Or the hardcoded limit on entity speed that had a soft limit that it would silently ignore. That sort of thing.
It was very impressive technology for something Carmack just hacked together in (IIRC) a couple days. I spent two quarters in my compilers class building my own language, and we didn't even have to write a VM to interpret the emitted code. It was brilliant, but hackish.
I'm sad as well, Phrosty... Carmack leaving id is the end of an era for me. I still have my emails I traded with him back in the day on implicit parallelization of Q2 code on the Tera Supercomputer...