The second day of the CSM summit kicked off with a 2.5 hour meeting on CCP's lag-fighting efforts. CCP Veritas keynoted an extensive presentation on not only what CCP is doing, but the procedures they have put into place to improve the efficiency of their efforts.
I will not go into great detail here, because I don't want to steal their thunder -- there was enough for several devblogs, which CSM encouraged CCP to produce, of course. However, I will hit some of the highlights so you have some idea of what to expect.
In the short term, CCP is working hard on issues like Session Changes, Module Cycling, performance improvements on heavily loaded nodes (such as Jita), and general performance improvements that provide more headroom (which means a server node can handle more people). These changes have both direct and indirect effects on fleet fight performance, often in ways you might not expect.
In the longer term, CCP wants to ensure that heavily loaded nodes exhibit graceful and predictable performance degradation (as opposed to suddenly falling off a cliff and dying horribly), increase performance limits, and transition to running systems on multiple cores.
None of this is really new, but what made a particular impression on me is the obvious seriousness of the effort being made to address lag; it was serious before, but now it is deadly-serious. The lag-fighters are no longer just an ad-hoc task force, but a full dedicated team.
With respect to the oft-expressed opinion of the player-base that Dominion caused a huge increase in lag, CCP Veritas presented a very interesting analysis of what actually happened that I think everyone will find fascinating when it appears (hint, hint) in devblog form. Was there more lag? Yes. Did it happen for the reasons most people think? Nope. Ah, EVE, "Unexpected Consequences Online" -- gotta love this game.