i'm going to tell you what i'm surprised about.

the range was 34-43, so i wasn't arguing that the polling was wrong so much as i was arguing that the analysis was inaccurate. if the conservatives ended up with 36, that wouldn't mean the polling was wrong, as it was correctly in the range. and, at 39-40%, that doesn't mean the polling is more right: it's a range. i was just pointing out that the conservatives usually poll at the bottom of the range. they did federally in 2015, 2011, 2008 & 2006 - as well as in every provincial election i can remember.

if they end the night in the middle of the range, that is unusual. it's not more right, it's just not consistent with the general trend. & that's what i'm surprised about. this election is consequently an outlier in terms of how results line up with polling measurements.

i haven't seen turnout numbers yet, but what it suggests is that turnout was probably quite low. this is what the models suggested, and that's fine. but, we're used to low turnout, and it's never affected the analysis before. this must have been really low turnout - so low you might want to wonder about it.

so, that's the first thing i'm surprised about - that the conservatives polled in the middle of their range rather than at the bottom of it, like they usually do. and, that would have to be because turnout was low.

the second thing i'm surprised about is that the liberal numbers are as low as they are, and that's the other side of the equation. they could still end up a tad over 20%. but, i was expecting that this is where the error would wash out, and they'd get to 22 or 23.

the end result is the outcome i feared: liberals just didn't vote at all. mostly in the gta. so, you're going to hear people talk about the conservatives swinging liberal votes, but if you look at absolute numbers, that will come out as false: conservative numbers likely stayed flat in the gta, while liberal numbers crashed. so, the conservatives ended up winning seats as a consequence of liberals staying home, the result of voter suppression through media assassination.

my analysis was rooted in rejecting the idea of equally distributing undecideds, which were clearly disproportionately historical liberal voters, and correcting the distribution of liberal voters to account for their ideological leanings. i continually pointed out that if these voters don't vote at all, the pcs would split the vote - but i didn't think that would actually happen.

had all of those liberal voters voted...

alas.

i got my huge swings in downtown toronto, but as soon as you get out of the core, those liberal voters stay home rather than vote liberal or ndp.

so, give the election to the media: it succeeded in suppressing the vote.

...if it was the media & not the machines.