I've chosen two ridings, Saguenay and 'Quebec' on the map. While the Greens do run in Quebec, they tend to do poorly in most areas. In fact, the Green vote in Quebec is concentrated in the english parts of Montreal, where, in some radical cases, they can be the 2nd largest party. As such, I will ignore them in this examination, which is in francophone areas. The reason I've chosen these two ridings is that these are two areas where all the 4 major parties in Quebec are competitive. While the Tories have not run ridings in 'Quebec' in quite a while, they do perform well in these areas. Well enough that in a 4-seat riding it's quite possible, if not plausible, that they could walk away with 1 of those seats. In Saguenay the 4 party balance is even closer, and the reason I've chosen this riding is to show how the number of seats can be crucial. Saguenay's population and it's geographic distribution are very friendly to a single multi-member STV riding. The population is pooled far enough away from anywhere else that expanding this riding is unpalatable. The entire saguenay area is an area on to itself, and as such, 3 members just makes sense. This puts pressure on the 4 parties, as if each party only runs 1 candidate (and they would) whichever candidate finishes 4th, simply does not win a seat. Should a party risk running 2 candidates, they risk seeing the other 3 elected while their candidates split the vote.
This is the strategy that you will see within STV. How many candidates to run, how to divide them within the riding to properly manage the vote, and even to some degree, what sort of 'deals' to make with other parties. What if, for example, the NDP, knowing they might find a difficult race in Saguenay, approached the Greens. What if the NDP said "hey, we know you are having some problems winning in Vancouver. You always come close, but you just don't quite clench it. If you'll withdraw from Saguenay, we'll only run 1 Candidate in Vancouver, and urge our supporters to rank you #2." To the public the NDP can simply say that they only think they can win 1 seat in Vancouver, and that if the Vancouver NDP thinks the Greens should be preferenced #2 their's their business.
These are the sorts of things we may see if we go down the route of STV (which is looking more likely)
Lastly, I want to look at the "Ontario" riding. You'll notice it's a very strange shape. Guelph is within it, but a small portion of another riding is between it and the remainder of the riding. Despite that, the riding actually makes sense given commuting patterns. Guelph is 'closer' to Kitchener and Waterloo than it is to Burlington and Oakville, while the riding that surrounds it is centered around Georgetown, which most certainly is closer to Burlington and Oakville than it is to Kitchener and Waterloo.
That aside, the reason I've chosen this riding is the Greens. In the last election, Gord Miller ran in Guelph, and there's no question that a "star" candidate like him would have been the only candidate in a STV riding.
The important thing is we are talking about a 6 seat riding. Arguments like "I can't vote for you because you can't win" become moot. As such, you can expect to see the Green vote increase significantly in areas like this, where they, in fact, can win. The simple existence of ridings like this is reason to expect the Greens can, and will, win more seats. Had this riding existed in the last election, I have no doubt that Gord Miller would have been elected as an MP. With STV and the transfer of votes, it becomes likely that NDPers, in an effort to stop either Liberals or Conservatives from winning, would preference the Greens well.
Due to the existence of 'dead' votes, it's likely the race for the final seat in many ridings will see three candidates, all below the threshold. In these cases, whoever is last, becomes the kingmaker.
Regardless, I hope that this has been a helpful look at STV and how future elections might be run. It's gone on far longer than I'd thought it would. In the future I do plan on examining specific races in Ireland to show as an example for how 'non obvious' things might happen. Until then, I'll leave you with this thought; every system has complex things, including FPTP. The only difference is we are exchanging one set of complexities for another.