The voting system puts multiple seats in a single riding and has voters rank their candidates in order of preference, with portions of each vote getting transferred between candidates. It was soundly defeated for the second time by B.C. voters in the 2009 election.
One of the arguments proponents of STV used was that it eliminates any incentive for strategic voting -- that is, there is no reason for people to vote for a candidate other than the one they truly want to win. But game theory shows people can still get better outcomes by voting strategically with STV.
In the University of Montreal paper, researchers found that in 9% of cases, people could get better outcomes by voting strategically with STV. However, they noted that strategic voting would be extremely complicated, and in order to pull it off effectively, you'd have to develop some belief about how other voters will rank candidates on their ballots.
It's not surprising that when the researchers ran an experiment with human participants, they found people don't vote strategically under STV, even though they vote strategically under other voting systems:
"In the case of Approval voting, strategic voting is simple and produces no paradoxical recommendations; we observe that our subjects vote strategically under this system. On the contrary, voting strategically under STV is a mathematical puzzle, and we observe that voters vote sincerely."
I always thought that STV being too complicated was a downside of the system, but this study shows its complexities actually have some upside.