I usually conduct my post-mortem of what I got right and wrong on an annual basis, but I’ve decided on this occasion to do a quarterly update.
Why? Well, the most important reason is to sharpen my analysis. However, when I conducted my 2019 post-mortem, I realised that I had done something very bad: I had adjusted some of my thinking and methods mid-year to account for my mistakes without committing them to paper or informing readers that I had done so. As it happened, the changes I made held up very well – but I don’t think it’s good intellectual hygiene to make changes to your methodology without committing them to the record.
Two events in 2020 have caused me to rethink some of my priors, and I’m keen to get them down on paper, not least because my thinking on these three mistakes is likely to shape my analysis of the rest of the year – and who knows, might well contribute to my next set of big mistakes. Two involve coronavirus, but the first involves the Labour leadership.
I completely failed to predict Keir Starmer
I don’t mean in 2020, of course – by the time Starmer was officially announced as Labour’s next leader, we could say with as close to 100 per cent certainty that he was going to emerge triumphant from the race.
I remain of the view that internal party contests are generally pretty easy to predict once they get going, particularly in the case of Labour ones, where there is so much useful data available. The issues afflicting opinion polls (propensity to turn out, too many politically engaged respondents) simply don’t matter in polling party members. Constituency Labour Party nominations have reliably predicted the winner of every contest among party members for which we have information.
But the Friday after the general election I was confidently predicting, as I had done for most of the past three years, that Emily Thornberry would become Labour leader. My reasoning had been unchanged since September 2015, when I tweeted: “if Corbynism does go down in flames in 2020, leadership probably goes to the most moderate politician to stay loyal to Jeremy Corbyn”.
My underlying view throughout the Corbyn years was that when it was over, the rightmost politician to have remained within the tent would become Labour leader. I wasn’t then in the habit of putting percentage figures on predictions – while Philip Tetlock recommends it in his book Superforecasters, I had been reluctant to do so because, while I think there are strong benefits to doing so to the person making the predictions, I am not convinced there is much value to the reader, as people are in general pretty bad at understanding risk and percentages.
But fortunately, I was in the habit of betting politically, which allows us to recreate without hindsight what my actual sense of the Labour leadership race was throughout the Corbyn era. Of the bets I placed between September 2015 and December 2019, 85 per cent of them were on candidates whom you might at one point or another have broadly described as “Corbynsceptics within the Corbyn tent”. In order of frequency, I bet on the following candidates: Emily Thornberry, Angela Rayner, Jon Ashworth, Lisa Nandy, Heidi Alexander, Andy Burnham, Keir Starmer and Owen Smith (that last, I hasten to add, in September 2015). The remainder of my bets were on Clive Lewis and Rebecca Long-Bailey.
Some of that reflected opportunism rather than a particularly strong belief in any of the particular merits of the candidate. Because “the most loyal Corbynsceptic” is, by definition, a moveable feast, I knew I was looking for a type of politician, but I didn’t know exactly who of that type would emerge. I made only one bet on Starmer, in the immediate aftermath of Jeremy Corbyn’s second leadership win – largely because I took the view that someone a lot like him would become Labour’s next leader, and his odds (20-1) were good. I can’t plausibly claim to have thought that Starmer would win.
To take a different type of political bet, when I put £10 on Justine Greening at 200/1 to win the 2020 London mayoral election the day after Sadiq Khan was elected, I didn’t expect to get the money back: I just took the view that Greening was, for my money, far and away the most impressive London Conservative, with a great backstory and a strong record of delivering in liberal-friendly departments, but because it was only a tenner and the perfect storm of “Conservatives nominate their best candidate + Sadiq Khan fails to deliver + there’s a generally favourable political backdrop for the Tory party” felt sufficiently probable than I might as well take a chance. I viewed the one-in-ten chance that I would make £2000 as more than worth the 90 per cent chance that I would lose a tenner.
If anything, I thought it was less likely that I would get the £200 profit off my ten pounds on Starmer (I only ever bet small amounts) than I would make £2,000 off Greening. I just thought that it was worth a punt.
Yet when I look back at my own reporting, Labour members were telling me that they liked the look of Keir Starmer not only in 2016 but in 2015! I wrote repeatedly that the evidence of Labour selections – when the process was free and fair – was that the party’s centre of gravity hadn’t moved all that much under Corbyn. These are all indicators that should have had me chucking tenners on Starmer whenever his odds lengthened.
I made two mistakes. The first was that I thought that the idea that it was “time for a woman” to lead Labour would have a greater sway among Labour party activists than it did. I over-indexed from the 2015 contest, in which a non-trivial number of Labour Party members opted to back Yvette Cooper over Andy Burnham for that reason. The problem is that there was very little ideological difference between Cooper and Burnham in terms of their left-right positioning. (I think there’s an argument to be had about whether the gap between the two is actually bigger than we might think, but I’m referring solely to their perceived position among Labour members in 2015.)
The second and bigger one: I wildly underestimated Starmer’s political abilities. I placed three £10 bets on Lisa Nandy from 2015-16, and just one on Starmer 2016-19, and if you had asked me which one of the two would fight a leadership campaign that was ruthlessly focused on winning the contest, that recruited incredibly widely from across the party’s talent pool (I can’t think of a contest ever in any party, and certainly none I have covered, in which someone has had a team quite so broad across that party or with such a deep bench, talent-wise), I would have said “Lisa Nandy” every time. (And indeed, in a very real financial sense, I did.)
In essence I thought that Starmer was of a type that you see semi-frequently in parliament: hugely impressive people in terms of their previous job, and hugely impressive when they are given the right ministerial portfolio, or are on the right select committee when they can draw on their previous experience. These people are, as I’ve written before, the backbone of their parliamentary parties. But what they tend to lack is what you might call a “small p” political sense: they have a great grasp of the politics of educational disadvantage or the barriers to small businesses becoming large ones, but they tend to struggle at the meat and drink of climbing the greasy pole. This was a pretty big mistake, though, because I had already reported way back in 2014 on the effectiveness of Starmer’s campaign to be selected. I feel that if I had ever forced myself to write down the case for why I was so long on Starmer’s odds, I would have rapidly realised that I was underrating his chances – and that my assumptions couldn’t be reconciled with my knowledge.
I think the big problem here is that one of the best ways to improve the quality of your forecasts is to publish and update your assumptions and analyses, which I’ve hitherto been reluctant to do, because a) I’m lazy but also b) I’ve always been concerned that it interferes with the rest of my job: my fear is always that whenever you write “so and so will probably lose”, you are inundated by two questions from them or their staff: queries about why you dislike that person and attempts to change your mind. The latter is fine and helpful, but the former is wearying (and self-fulfilling).
So I’ve decided to start publishing and updating my assumptions about politics every quarter, ie at the end of this month, then again in August, and then finally in December, when I will also do my usual post-mortem.
I didn’t think that we would all end up being confined to our homes by coronavirus
I am not that worried about this one. I think Scott Alexander of SlateStarCodex and Tom Chivers over at Unherd are both right: the test here is not “did you predict this catastrophic event?” but “were you aware that something like this could happen?”
I didn’t publish confidence estimates that the novel coronavirus would be a serious issue, but as with my betting slips, fortunately we have another heuristic available: the small items in Morning Call, my daily email about British and global politics. What I am aiming to do with the news-in-brief item is to keep readers well informed both about big stories and about things that might become big stories. (And also occasionally to find something lighter for the final story to keep people’s spirits up.)
At the start of the year, there were two stories internationally that I thought could develop into major stories, but were unlikely to: the first were the questions about party donations to New Zealand First, the junior partner in Jacinda Ardern’s coalition, which I thought could complicate her path to re-election in December 2020, and the second was the new respiratory illness emerging in China. I considered them important enough to cover in Morning Call, but not so important as for me to lose sleep over the prospect of either: I thought the odds of either were about as plausible as I did of me getting that £10 back when I bet on Justine Greening: they were one-in-ten chances.
Where I feel I fell short is that I should have been blogging and asking about what the British government’s preparation for a potential coronavirus outbreak was. Just as I’d expect the Foreign Office and the Department of International Trade to be maintaining and sharpening its diplomatic relations with National, the main opposition party in New Zealand, and Act, the right-liberal party on whom it is likely to depend on if it enters government in December, given the questions around New Zealand First’s funding, I’d expect the government to be taking similar contingencies around the passage of coronavirus.
A good example of this can be found in the Conservative governments of David Cameron and Theresa May: yes, diplomatic officials were just as prone as the rest of the diplomatic circuit to make jokes about how they were looking forward to working with the next American president, “whoever she may be”, but they were also working to make sure they had connections in place should the unexpected happen. Many were grumbling about how much time they were wasting doing so, but they were still doing it.
I think I should take much more time to blog about looming international events that I think might happen and whether the British government and Whitehall are prepared for them.
I thought the British government’s Plan A to tackle coronavirus was about right
The four British governments started this crisis with a virus-fighting approach that put them, along with Sweden, the Netherlands, Japan and a handful of other countries in a minority group among developed economies: to fight Covid-19 not via lockdown measures but through a series of lower-key interventions to delay the peak of the outbreak rather than to suppress it entirely. The government believed that the measures taken elsewhere that had seen the virus suppressed could not be maintained for long enough to be a viable approach.
I thought this strategy made intuitive sense, and also wrote that, while scientists were divided, the important political question was whether the government was providing sufficient economic support for people self-isolating, as the scientific question was inherently uncertain and would answer itself in time one way or another, while the question of “can this person afford to self-isolate?” was up for grabs.
I don’t actually think either of these were wrong: we still don’t know for sure that you can maintain prolonged consent for these measures, and the debate about the UK’s approach did resolve itself. But I think what I visibly failed to communicate was something I knew, which was that Britain was taking a less cautious approach than many other countries.
I don’t think the government or the bits of the media that had absorbed this were sufficiently clear about it. I was far too quick to move to giving an opinion about the government’s strategy rather than an explanation of the government’s strategy.