I've been following the discussion over there, though I haven't gotten involved (though it seems to have died down by now). A lot of the resistance to SMW is part of a longstanding dislike, or at least lack of understanding, of SMW among core MediaWiki developers; though as Niklas points out, it does seem to have decreased this time around. Though the fact that it's taken basically seven years for that shift to have occurred is somewhat depressing. :) Actually, I think a lot of the shift can be attributed to the efforts of just a few people: Ryan Lane at WMF operations, and Niklas and Siebrand on the Translatewiki team.

My only other comment is about performance: someone on the wikitech-l thread noted that Wikia has had performance problems with SMW on some of their sites, which is true; but I've talked to the Wikia people, and I think just about all of these issues can be traced to their use of Semantic Drilldown. SD (an extension I wrote) is useful but can overload the database if it's used indiscriminately on larger sites. I've given them tips on how to reduce the performance impact of SD - I don't know if they've taken any of them. But given that, I don't think the Wikia experience, at least, applies to Wikitech or mediawiki.org.

-Yaron