so, i got a start on this. i'm taking what i think will probably be a more efficient aproach this time, so it should be a bit faster.

but, i just want to stop for a second to double down on what i'm projecting in the election.

first, i'll remind you that i do have a four year degree in mathematics, plus enough credits for a masters (taken while working out a degree in computer science that i stopped a credit short of). i made a hard turn out of academia, because i decided i wanted to be a musician and i was wasting my time with it. i then ended up doing a lot of political activism, and ended up as a prolific blogger, so here i am. i've also worked in polling, as a student. so, referring to me as an "amateur" is a bit of stretch; i'm as qualified to speak on this topic as anybody else that you'll meet.

and i need to get this point across very clearly: it's not that i'm too lazy to model, it's that i actually think it's a bad approach because the sampling frames suck. i'm explicitly dissenting; they're trying to ram square pegs into round holes, and more often than not end up completely wrong regarding anything that's not already obvious. in order to statistically model canadian elections with any kind of accuracy, the way that the computers accept the data has to change very radically. right now, it's largely just garbage in, garbage out.

so, this isn't an amateur approaching this from an ad hoc perspective; this is an expert that is offering a dissenting opinion around the existing methods, and offering a different approach. 

but, how do i get a liberal majority when they can barely get 30%, and might not even win the popular vote, and everybody else is suggesting a coin flip minority?

because the regional numbers actually aren't even close.

the conservatives are at 70% in alberta and saskatchewan, the liberals are at 45% in ontario and quebec is a clean split between the liberals and the bloc. when you add it all up, they're tied, nationally, sure. but, in actual fact, the liberals are going sweep the east, and the conservatives are going to sweep the west, as we've seen however many times since wwII - and i don't need a sophisticated modelling algorithm (one that the greens running at 10% in ontario, concentrated into urban ridings, is going to completely break, i might add) to tell me that.

my dissenting opinion is that we're better off fitting the data to past results than trying to interpolate it in real-time.

and, for ontario, which is going to decide this election, the numbers are most similar to the ones in 2000 - where the pcs and reform split the rural vote, handing the liberals a huge majority.

even if that doesn't happen, the conservative numbers in ontario appear to be very low, right now. the rural split with the greens might be stretching, right now. but, a liberal sweep over the 905 is a reasonable projection, and the conservatives simply can't win an election when that's the reality in front of them.

we've seen very weird election results in canada, in the past. run the 1979 data into your model and tell me what it says - i promise you that you won't get a clark minority.

these results could be awful, from a democratic perspective, but that's the norm, here.

don't let that huge conservative lead in alberta skew your analysis - they're not running well in the east, right now, and are not going to win the election.

the liberals are supposed to do better than this