The biggest changes have been new sub-sites. The most recent of these is FanFare, for discussing TV and movies (and possibly expanding to other media in the future): http://fanfare.metafilter.com/

There have also been subtle changes, like removing the ability to post images in comment threads, or adding a five-minute window where you can edit comments for typos. I appreciate the amount of thought that goes into both implementation and policy changes and how they are likely to affect discourse on the site. It's the people and the conversation that make MetaFilter, not any flashy technical features.

Whoa! I totally didn't know about this subsite. This is great.

Launched in response to TWOP's shuttering.

It's brand new, and they're still working the kinks out. But Matt has said he wants to expand it to include discussions about movies, and possibly books as well.

It's added a lot of small features, like dynamic comment updating and edit windows, but most of the new development has gone into new sub-sites. They're pretty conservative about changing the core commenting dynamic, especially given their informal policy against undertaking technical solutions to social problems, which I think has been a beneficial policy for the quality of discussion there.

They have (or have had) developers on staff to assist Matt. It was originally written in ColdFusion (to be fair, the site has been around since 1999), though I imagine it's moved onto greener pastures since then. The look of MeFi is iconic, much like Slashdot, so any radical changes to its design would likely cause more problems than it might solve.

Sure, I don't necessarily mean aesthetic changes, but - particularly when you've developed a strong community - there's often a resistance to modifying a core offering. That benefits the community but also often begins to erode your ability to attract new users.

It's a difficult balance, no question.

Sure. For what it's worth, though, we've continued to see steady signups for years and years; the concern over the ossification of a conservative site design is a reasonable one and one we chew on plenty, trust me, but as far as that goes we haven't all else aside seen active new-user interest go away.

Any other efficiencies to be gained on the technology side?

If the userbase is steadily gaining, it seems like monetizing that traffic should be inherent and profitable unless the infrastructure is inordinately expensive.