The theoretical computer science community is off to another round of discussions regarding its conference system. Lance Fortnow, ex-officio as SIGACT chair, put forward a suggestion of reforming STOC, a drastic change that may alter the whole conferences-as-top-publications culture of the field (as Lance has previously suggested doing). He set up a blog with the proposal itself as well as many responses — mostly negative ones. Other recent related posts are here, here, here, and here (added 21/6: and here and here.)
While I like the CS conference system, I still think that a change is needed in the pan-TCS conferences as the field is getting bigger and deeper. Like most others who commented on the proposal, I doubt that the suggested new format has enough value to attract attendees. As many suggested, I would start by co-locating with other related conferences, as was done this year (with CCC and EC).
Turing's Invisible Hand 
I actually didn’t like the “co-location” of EC with STOC this year. The two locations were about 3 miles apart making it very difficult to go back and forth and forcing you to pick one over the other.
TCS Conferences again?