A summary of John Carmack's #quakecon statementsm, from pages by Sean Martin:

* In the future he wants to go to more object oriented code so mixing &
  matching of modules will be easy.
* Quakeworld client will be out in about 2 weeks. 2 or 3 more weeks after
  that, another version will come out.
* Quakeworld will include latency reductions so that 28.8 will play more
  like IDSN, etc.
* He wants to allow more client side changes (e.g. a weapon icon on the
  status bar) in Quake 2 and to show who's running hacked stuff. There are
  bandwidth considerations for more client flexibility.
* Likely to do crouching or crawling later (Quake 2?)
* Working with the idea of exiting a level to an IP address but there is a
  problem with a server being full for the next level. Possibly a reservation
  line to help this? There's also an issue of wanting to start a level over
  again fresh because people could stay on a level forever and their large
  frag total is meaningless.
* Wants to add Clan specific support in Quakeworld. Multiskin support might
  be such that each clan can only use a specific Clan texture.
* The master servers will have separate accounts. No data sharing.
* The ranking system will be similar to chess in that you can only get a
  higher ranking by killing better people. Magnitudes of wins may be counted.
* QW originally will have the rankings will just be a matrix of frag totals.
* QW will have something like Ctrl-Esc to do a chat function that is better
  than the one currently in Quake.
* QW Servers could only allow people of certain skill levels on.
* The NT QuakeEd is currently in Alpha...maybe released in a few months.
* Quake 2 should have everything with 500 polygons because rendering will be
  over 50% faster.
* 3dfx is the best card out there for Quake and will provide the most
  significant improvement. Unfortunately it costs $300 for a card that is used
  in addition to your current 2D card. This may cause the marketshare for the
  card to be small and Quake may not have a port of it for that reason.
* All the other cards pretty much suck and will only give you smoother
  pixels and higher colors rather than higher framerates. Some card ports (S3
  ViRGe) might actually be slower than the normal Quake.
* People will be disappointed with the DirectX performance for a while.
* Even so, developers have to move to it because they can't possibly support
  all the 3D cards out there.
* Will add a Palette Shift to Quake C.
* He wants to get rid of the graphics wad in Quake.
* There may be some bandwidth choking in QW where the game may get choppy
  but it won't bog down like it does now where it gets overloaded for a few
  seconds.
* MMX doesn't work well with Quake (because it takes gets its performance by
  using floating point lines and Quake wants to use them) and generally isn't
  as good as hyped. 16-bit color at about the same speed as 8-bit without MMX.
  The sound channel part of MMX doesn't work well because there's a problem of
  getting it over the bus.
* AGP isn't so hot either. (I didn't hear this part but that's what I was
  told).
* Quake 2 should be about a 9 month project and will start in about a month.
* For the next generation, John should have the new graphics engine and
  editors done in 1 year. Development should finish about a year after that.
* Quake is the last palette based engine they will do. The next will use
  direct color instead.
* Dynamic Shadowing not likely to be in Quake 2.
* Colored lighting may be in Quake 2.
* The Nintendo 64 is "Damn good hardware" and DOOM for the N64 never drops
  below 30fps.

---------------------------------------------------------------------------