This is a bit out of my wheelhouse, but I'm struggling over whether yesterday's UK election was a win or a loss for Duverger's Law. Basically, Duverger suggests that single member plurality district (SMPD) elections will tend to favor a two-party system. This is for two reasons. The first is that even a relatively popular third party will have a hard time winning a plurality vote in any given district. So even if the Liberal-Democratic party held a solid 30% in all districts, they still wouldn't have won a single seat. The second reason Duverger suggested is that voters are strategic -- they don't want to waste their vote on a party unlikely to win, and they don't want to help out the less desirable of the two likely winners (á la Nader voters in 2000), so they vote for the less awful of the major party candidates.
The UK 2010 election was a perfect example of Duverger's first reason. The Lib-Dems took 23 percent of the vote but will control only about nine percent of the seats. If you're a Lib-Dem supporter, this is a great argument for proportional representation. However, Duverger's second reason didn't really hold up this time. Voters did not seem to behave strategically. Yes, the Lib-Dem vote fell a few points short of what polls predicted (apparently bleeding to Labour), but support for this third party basically held.
Which suggests to me that parties and candidates can't take strategic voting for granted. Absent efforts to unite the vote around a few key candidates, voters are quite willing and able to splinter their votes.