The Arrogance Of Certainty

How To Build Back Better In 2021, One Might Say

"We're the Natural Governing Party." - June 9th, 2016.

Less than 8 months after coming from 3rd place into Government after a campaign in which the Canadian Liberals started in third place in the polls, a Liberal staffer I had the intense displeasure of speaking to uttered that quote. I remember exactly where I was - sitting in a chair in the conference room of the Member Of Parliament who chaired the committee my boss was on. It was about 8 of us - me, the staffer I reported to, and a staffer or two for the other Liberal members of the committee.

I remember it was the 9th because it was the day of Muhammad Ali's funeral, which was on the TV in the background for a while, but I remember that day not for the events of my working day, or even for the comment itself, but for my mood as I walked from that office back to the downtown core, repulsed by the arrogance and, frankly, stunned by it. It wasn't the term itself - a widely used epitaph for the Liberal Party of Canada, given it governed Canada for all but 25 years from 1921 to 2006 - but the casual use of it, especially that soon after the Liberals faced electoral death, was sickening.

Just under three months later I was in Toronto, staying with a friend and helping the Ontario Liberals with a byelection in outer Toronto, in a part of town where white Anglo-Saxons are sparse, and non-English speakers flourish. I was just some guy, but the campaign appreciated any help, so people - including the Premier's Deputy Chief Of Staff - would just talk. It wasn't so much that I was being told any party secrets, it's just that my presence didn't deter anyone from saying much. We were told that we had 17000 "A" commits for the election three days before, and the tone was jubilant all week. The mood wasn't tense, and you could get high level party staff to talk pretty freely about the campaign, the prospects of the next election, and strategy. Nobody was tense, like they weren't expecting anything but a win.

The party didn't break 7300 votes, and the next day the same people who told me that the Premier would easily win the next election were talking about whether there'd be a leadership election.

…

In my time as a public writer, I've covered three major general elections - Canadian and British General Elections in 2019, and the US election weeks ago. In both 2019 General Elections, my coverage - informed and guided by both the polls and the brilliant modelling I could always fall back on - focused on the way that the conventional wisdom was wrong because of an over reliance on historical outcomes, and bad priors. In Canada and the UK, I was right, because the polls were right and the modelling properly calibrated what that would mean. The modelling was confident that, so long as the polls were even remotely correct, the Canadian Liberals would keep their suburban seats, even while losing the popular vote nationally, and that the UK Tories would make huge gains into Leave areas where the Tories had, in some cases, quite literally never won before. In both cases, the polling was correct, meaning the modelling was correct, and therefore the legion of articles written were correct.

There were a couple lucky breaks along the way - our following was in part made by giving people the tip to bet UK Labour in Putney at 4/1, a bet that paid off well despite Labour's horrible night, and I wrote an overconfident analysis of Matt Bevin's political demise five or six weeks out of that Governor's race in 2019, but he lost, so nobody noticed I got over my skis. Hell, a piece declaring that the 2020 Democratic Nomination would end on the convention floor, and not sooner, missed badly, but the nature of the 48 hours between South Carolina and Super Tuesday meant that people forgave and forgot about that. I had gone out on a limb more than a few times, and was never pulled back, and therefore the line of acceptable arrogance got wider. I pushed the Overton Window of my confidence further and further to the side of certainty and clarity, and conflated a proper discussion of edge cases (in a probabilistic sense) with hedging.

No, I never said Trump couldn't win, and I always included a line somewhere in every article about how it's not over until it's actually over, but it was trite, rote nonsense designed to cover my ass. I believed the polls, I dismissed the counterarguments, but most importantly there was a bravado to the writing that meant that even if I wrote the trite line somewhere in the middle of the article, it was never prominently placed.

And then, it turned out it was all just wrong.

…

Why was it wrong? I have no idea, but more importantly, I'm not trying to figure it out. Was I more wrong than the consensus, which was close to, but not quite, as wrong as I was? Sure, but it's not like I'm going to cry about having light blue Ohio when other people had it light red. Or, more to the point, it's not like I'd feel any better about my process if I had Trump winning Ohio by half a point, given he won it by 8%.

The arrogance was dripping off those columns like a man on a first date who only realized how much cologne he actually needed well after the fact, and it wasn't just that I'm a cocky bastard - which, of course, I am. It was the manifestation of certainty of belief, that I had figured out how to deal with the rush of data that comes out, sometimes contradictory in nature. I knew what I was looking for, and I pounded away at it until people were so bored of hearing about how much Ossoff and Biden were outrunning Stacey Abrams by with whites. I mean, at least that one came through for people.

I could easily make a case that my failures start and end with the fact that I believed the polls, and otherwise just push all the blame to the fact that the polls were fucking atrocious this cycle. I could easily say that I wasn't wrong about Florida based on the polls, and that I wasn't wrong to say that all the panic was overhyped just because it's Florida. I could easily make that case, and in some ways I have in the past. But that isn't quite true.

I was certain that I had done everything right - I had looked at the district polls, and I was trusting the data, and I was actually figuring out the polling miss, and I had figured it all out. The polls kept matching the model, and the later polls kept matching the earlier polls, and the old columns in my archive weren't being disproved by later events, and the GOP kept acting like they were losing, and at some point it just didn't seem possible that there could be a widespread polling miss in the favour of the GOP again. I let the certainty of my conviction cloud my judgement, and the result was a cologne bath's worth of narcissism that could only be matched by that of a College freshman at his first frat party.

…

In the days after the election, there was a war of words between the centre and left of the Democratic Party over why what happened happened. It was nasty, needlessly personal, and reductive - on both sides. But the biggest flaw of it was that it was lacking in the spirit of humility that any postmortem requires - because, if the answer were that easy, there wouldn't need to be one. Remember, everyone in politics thought Democrats would net seats in the House, and then they lost like a dozen of them. There was the certainty of centrists in declaring AOC and Defund The Police responsible for losses, and there was equal sanctimony from AOC et al in blaming the centre for not standing for left wing values in a sufficient manner. Thing is, both were wrong, even if they were right.

Getting the right answer matters a lot less if you get to the right place the wrong way. I've never properly articulated this, but my surprise that the Canadian Liberals could be as stupid as they've subsequently been was never as high as some people's. It was never a complete shock that they could be corrupt assholes, and I guess somewhere in the deep recesses of my mind I remembered the fact that the motherfuckers who run that party are the kinds of people who get back into office and immediately start saying shit like "we're the Natural Governing Party" unironically. Just because you get the right answer this time doesn't mean you'll get the right answer next time. The best pollster of 2018 - the New York Times/Siena - was atrocious this year. Were they lucky in 2018? I have no idea, and I'm not qualified to speculate. But past results are fairly meaningless, and testing on the past is a good way of missing the future.

…

In the next two years, we're almost certain to get a Canadian election, we're certain to get provincial elections in Canada's two biggest provinces, the US midterms will come and go, Australia will go to the polls as well as a bunch of its states, and we will have German, French, Scottish, Welsh, and North Irish elections as well, plus whatever other oddities come around. The lesson of 2020 isn't just that US polls are bad, or that pollsters have a problem getting representative samples - it's that certainty is just arrogance, and even if you get a few right in a row, it doesn't matter. This venture will offer few predictions, but much commentary, in line with a long standing view of mine. The greatest value I feel I can provide is not false precision or overstated arrogance, but honesty - a honesty lacking in much coverage of both politics and frankly life. The value of this project is in its objective - this is an open look at my mind, where it is at any given moment on any given topic.

This isn't about being right, although I hope that I am. This is about doing political analysis differently, about caring about the process, and about letting people into the parts of analysis and takery that is usually kept behind closed doors, but most of all, this is about learning the lesson of the past.

Welcome to Scrimshaw Unscripted.