Mantic Monday: Grading My Trump Predictions
I had many opinions on Donald Trump. I tried to back some of those opinions up with predictions about what would happen during his administration. Now that the dust has cleared, it's time to see how I did.
The summary: Of 48 specific predictions about Trump, I got 37 directionally right, although this is kind of meaningless. I got an average log error score of -0.48 (where getting everything right is 0 and guessing 50-50 for everything is -0.69) although this is also kind of meaningless. I quadrupled my money on prediction markets, which I think is meaningful. In terms of my more qualitative/implied predictions, got at least one important trend right before anyone else, but also made some embarrassing unforced errors.
Going through all my predictions post by post, and giving each a letter grade:
1: 10-23-2015: Trump's base is/will be surprisingly racially diverse (A-)
As far as I know, the first post I wrote about Trump was this one, where I argued against the prevailing narrative that Trump was practicing "the politics of white insecurity" or had an unusually white base of support (for a Republican). I wrote that Trump seemed to be doing pretty well (for a Republican) among blacks and Hispanics, and concluded that:
There are too few data to say anything for sure. But all of the data that exist suggest that if the Republican primary were held today and restricted to non-whites, Trump would still win. And if Trump were the Republican nominee, he could probably count on equal or greater support from minorities as Romney or McCain before him. In other words, the media narrative that Trump is doing some kind of special appeal-to-white-voters voodoo is unsupported by any polling data.
I was right. In the general election a year later, Trump did better than Romney had among non-white voters. He made large gains among blacks, Asians, and Latinos. The only ethnic group where he didn't gain at least five percentage points over Romney's numbers was whites. As I pointed out at the time, the narrative that Trump was especially appealing to white voters was bizarre and not truth-based, motivated primarily by a demand for racist Republicans on the part of increasingly woke narrative-consumers.
In the 2020 election, it became standard consensus that "people of color", though still heavily majority-Democrat, were surprisingly willing to vote Trump. But everyone acted like this was some unpredictable freak occurrence of the 2020 election; in fact the change started in 2016 or even earlier, as these graphs show. Everyone was just too stuck in their Trump = whiteness narrative to notice it.
Normally I would give myself an A+ on this prediction, but the post went on to take some digs at Bernie Sanders for actually having a disproportionately white support base. This was true in 2015 - 2016, but he picked up a lot of minority support over the next few years, and going into 2020 his base was as diverse as anybody else's. I don't think I was really making a prediction here, but insofar as I was, it was falsified. So overall I will give this post an A-.
2: 1-27-16: Trump will win the primaries but lose the general (and other predictions) (D+)
In my yearly predictions thread, I made a couple of Trump-related predictions:
21. Donald Trump will win the Republican nomination: 60%
22* Conditional on Trump winning the Republican nomination, he impresses everyone how quickly he pivots towards wider acceptability: 70%
23. Conditional on Trump winning the Republican nomination, he’ll lose the general election: 80%
24. Conditional on Trump winning the Republican nomination, he’ll lose the general election worse than either McCain or Romney: 70%
Prediction 21 - I'm proud of this one. It looks like the betting markets only gave Trump a 32% chance around the same time, so I was more willing to entertain a Trump primary victory. A.
Prediction 22 was obviously not at all true. He won the Republican nomination and continued being as bizarre and offensive as ever. I must have been really sure about this one - I had no need to express an opinion on this, but it was one of four Trump-related things I chose to predict about, so I suck. F.
Prediction 23: I can't find anyone else giving conditional odds for Trump. The betting markets gave 14% unconditional odds but I don't think I can derive conditional odds from that. Still, other people were saying 99% odds Trump would lose as late as November 2016, so I wasn't the worst person here. C.
Prediction 24: Also clearly not true and I suck. F.
Overall these numbers reveal I was blindsided by Trump's victory as much as everyone else. Still, since I know how to use probabilities and not be insanely overconfident, the worst you can say about me is that I gave 20% probability to a true event, which isn't so bad. Overall grade for this section is D+.
3: 9-28-2016: Trump will cause various bad consequences and you shouldn't vote for him (B-)
In SSC Endorses Clinton, Johnson, Or Stein I wrote about my reasons for opposing Trump. There's a lot in here and I'll try to extract relevant-prediction-shaped things - but be sure to read the post to see how honest you think I'm being in my extractions.
I started by arguing that Trump's brand of conservativism was an expression of policy-less rage against all existing institutions. I suggested people not vote Trump because destroying all existing institutions without a plan to replace them historically has gone poorly.
Although I think the “expression of rage” idea of Trump still makes sense, realistically he didn't destroy that many institutions. He just filled positions with unusually-corrupt but otherwise standard conservatives, and spent most of his time tweeting. I think my argument that you shouldn't vote for Trump because he would violently destroy useful institutions ended up kind of falling flat. D.
Next, I wrote:
Trump isn’t a known quantity. Maybe he’ll kind of dodder around and be kind of funny while not changing much. Or maybe there will be some crisis and Trump will take what could have been a quickly-defused diplomatic incident and turn it into World War III.
This hedged hard, but I think looks good in retrospect. Trump mostly doddered around without changing much. But his order to assassinate Iranian general Qasem Soleimani - which by my understanding was kind of a decoy out-of-range extreme plan his generals suggested only to make their other plans sound more reasonable - was exactly the sort of potentially-WWIII-causing blunder I worried about. In this case Iran decided it wasn't worth picking a fight, but that was their good decision, not Trump's. I realize this is sort of sketchy; I'm declaring a thing which empirically went well to unobservably be a massive blunder, then counting my prediction of massive blunders right. But based on my sincere assessment of the Soleimani situation, I feel more honest doing this than not doing it. B.
Remember also that it’s more likely the House and Senate both stay Republican than that they both switch to being Democrat. So if Hillary is elected, she’ll probably spend four years smashing her head against Congress; if Trump is elected, he will probably get a lot of what he wants.
The House and Senate were both Republican for Trump's first two years, after which the House became Democratic. But Trump still managed to get surprisingly little of what he wanted, even assuming "want" is a coherent emotion to attribute to him. D.
There’s a vicious cycle where the lack of intelligent conservatives guts the system of think tanks that produce the sort of studies and analyses which convince smart people to become conservative, which in turn makes there even fewer intelligent conservatives, and so on. In the end, intellectuals won’t just vote Democrat; they’ll shift their personal views further to the left to fit in."
I think this basically happened. See eg Trump: A Setback For Trumpism. A.
One more warning for conservatives who still aren’t convinced. If the next generation is radicalized by Trump being a bad president, they’re not just going to lean left. They’re going to lean regressive, totalitarian, super-social-justice left.
I think this basically happened. A.
4: 10-01-2016: Trump will be more of a warmonger than his supporters expect (D+)
In He Kept Us Out Of War?, I argued that despite the perception of Trump as the anti-imperialism, anti-military-action president, hopes that he would wind down global empire and avoid further military action were misplaced.
Yet Trump can honestly claim to be the first president since Carter not to start a new conflict. He didn't really do much to wind down US global empire, but it seemed like maybe he was half-heartedly trying and Congress just thwarted him.
I think my basic argument here had some okay points. I predicted Trump would escalate and complete the fight against ISIS, and was right. I said Trump was a loose cannon who might start a war for stupid reasons, and I think the Soleimani assassination was almost this. But overall, I have to admit that his supporters were mostly right about this one and I was mostly wrong.
5: 11-16-2016: Trump won't be explicitly and openly white supremacist / won’t openly support the KKK etc / won't pursue especially racist policies (A)
You Are Still Crying Wolf was my most-read post ever, getting retweeted by everyone up to and including Ann Coulter. I expanded on my original argument about Trump not getting especially strong white support to argue that the racial narrative around Trump was kind of bunk. This attracted a lot of debate, so I'm going to go into more depth about it than some of the other things here.
I wrote this during a time when people were making extreme claims about Trump’s purported racial policies. Matt Yglesias (then at Vox) wrote that “My guess is in a Trump administration angry mobs will beat and murder Jews and people of color with impunity”. Jamelle Bouie (of Slate) agreed and said we were on the edge of “state-sanctioned racial violence”. Salon literally wrote an article called If Trump Wins, Say Goodbye To Your Black Friends (it was subtitled “A Modest Proposal”, but my impression is the joking part was the suggestion to build a black separatist nation in Atlanta). Vox wrote about how minority kids believed they would be forced to leave the country; therapists were called in to help Muslim kids who believed Trump was going to kill them.
The news was full of lurid stories about Trump supporters being violent and hateful. The Atlantic described how a black church was set on fire and spray painted with “Vote Trump”; there was a wave of bomb threats against Jewish community centers; a Trump supporter threatened to set on fire an Ann Arbor student who was wearing a Muslim headscarf (I was a psychiatrist in Ann Arbor at the time and had to deal with the wave of fear that swept through the community after this). These incidents fed into the general mood of panic.
The consensus take on all of this was that Trump was an “open white supremacist” in league with the KKK, who fanned and supported all of these incidents.
I wrote a post stating that the Trump administration would not feature angry mobs killing minorities with impunity, would not feature state-sanctioned racial violence, would not feature an especially large number of hate crimes, and would not feature the state supporting and refusing to prosecute hate crimes. I said Trump was not an “open” white supremacist and probably not a white supremacist at all except in whatever boring way you can call anyone who isn’t super-woke a white supremacist. I said that he was not getting especially high support from white voters, that he was doing better with voters of color than other Republicans, and that attempts to paint his base as especially white were bad statistics. I tried to make it very clear that this was trying to be a low-scope pushback against these particular claims, and not some sort of generic screen in favor of Trump. I gave my thesis as:
There is no evidence that Donald Trump is more racist than any past Republican candidate (or any other 70 year old white guy, for that matter). All this stuff about how he’s “the candidate of the KKK” and “the vanguard of a new white supremacist movement” is made up. It’s a catastrophic distraction from the dozens of other undeniable problems with Trump that could have convinced voters to abandon him.
…and tried to be as clear as possible by including disclaimers like “I thought and still think a Trump presidency will be a disaster” and “I’m not saying Trump doesn’t have some racist attitudes and policies. I am saying that talk of ‘entire campaign built around white supremacy’ and ‘the white power candidate’ is deliberate and dangerous exaggeration.”
I think time has proven me right here. There was no large-scale state-sponsored violence against minorities during the Trump presidency. Trump’s race-related policies were similar to those of other recent Republican presidents. The three most-discussed Trump supporter hate crimes all turned out to be hoaxes (1, 2, 3, I swear I’m not trying to cherry-pick, these really were the most discussed incidents at the time). There was no huge spike in hate crimes during the Trump presidency, just fluctuations of about the same scale as during previous years. The much-reported-upon study showing that Trump rallies increased local hate crimes was reanalyzed and found to be wrong; adjusting for population correctly showed they did not increase hate crimes at all. Trump lost support among white voters, and gained support among voters of color at an unprecedented rate among Republicans. The KKK and Richard Spencer did not play a major role in the Trump administration, and the media stopped covering them in favor of new, less-racialized pro-Trump groups like QAnon.
I’m hammering this in because people were awful to me after I wrote that post. Every time anything bad happened involving race, I would get dozens of angry tweets saying “looks like we finally have proof that Scott Alexander is a pro-Nazi idiot” and a link back to my post. Even when things completely unrelated to race happened, people would do this. When the Capitol riots happened, with basically no links to organized white supremacy and ~15% of the rioters being people of color, after I hadn’t been blogging for six months, I still had people sending me messages about how this destroyed the last shred of my credibility or something. This was the most hate I’ve ever gotten about any post of mine and it was terrible.
I promised some of these people at the time that I would respond to them later, in depth, after the administration was over, so here goes.
I included some explicit predictions as a test of my hypothesis. Some of these predictions might seem so obvious that I shouldn't get credit for them. But again, I wrote them a week after Trump was elected, in the middle of a national panic, when people were freaking out. Trump was the KKK president! There were white supremacist militias marching on the Capitol steps! Minorities should consider fleeing the country! I was trying to come up with things to predict that corresponded to "your national panic is dumb", and here's what I came up with, plus how I’m grading them:
1. Total hate crimes incidents as measured here will be not more than 125% of their 2015 value at any year during a Trump presidency, conditional on similar reporting methodology [confidence: 80%]
Incorrect. I originally judged this one as correct, but a commenter noted an error in my math - in 2019, hate crimes reached 125.03% of their pre-Trump value. See here for an argument that reporting methodology changed in an important way that pushed this over the top, but absent proof of this I will default to saying I was wrong.
2. Total minority population of US citizens will increase throughout Trump’s presidency [confidence: 99%]
Correct, not that anyone should be surprised by this.
3. US Muslim population increases throughout Trump’s presidency [confidence: 95%]
Probably also correct - but I can't find any statistics proving this, so skip. In retrospect 95% confidence was way too low for this and I must have been at least a little caught up in the panic I was trying to argue against.
4. Trump cabinet will be at least 10% minority [confidence: 90%], at least 20% minority [confidence: 70%], at least 30% minority [30%]. Here I’m defining “minority” to include nonwhites, Latinos, and LGBT people, though not women. Note that by this definition America as a whole is about 35% minority and Congress is about 15% minority.
Incorrect. Based on this source, of 33 original Cabinet members, 3 were minorities, which is 9%. My weakest prediction - that at least a tenth would be minorities - was wrong, and obviously every stronger prediction was wrong too. I failed at a 90% prediction and am appropriately ashamed.
5. Gay marriage will remain legal throughout a Trump presidency [confidence: 95%]
Correct, not that anyone should have been surprised by this. This is another one where in retrospect I was massively underconfident.
6. Race relations as perceived by blacks, as measured by this Gallup poll, will do better under Trump than they did under Obama (ie the change in race relations 2017-2021 will be less negative/more positive than the change 2009-2016) [confidence: 70%].
Indeterminate. I originally judged this correct, but commenters pointed out that this hinged on the numbers for 2017, a year in which Gallup did not take a poll. Interpolating 2017 as halfway between 2016 and 2018 results in this prediction being correct, but commenters seemed to think this was sufficiently unfair that I will shift it to indeterminate.
7. Neither Trump nor any of his officials (Cabinet, etc) will endorse the KKK, Stormfront, or explicit neo-Nazis publicly, refuse to back down, etc, and keep their job [confidence: 99%].
8. No large demographic group (> 1 million people) get forced to sign up for a “registry” [confidence: 95%]
9 No large demographic group gets sent to internment camps [confidence: 99%]
10. Number of deportations during Trump’s four years will not be greater than Obama’s 8 [confidence: 90%]
Correct. In fact, I should have been more extreme here - Trump deported fewer people during his first four years than Obama did during his first four years!
Overall I got 6 of these predictions right, 2 wrong, and 2 indeterminate.
I also promised some of the people questioning me about this a response to more specific points - especially about Trump’s comments, and his support (or lack thereof) from white supremacists. This will be long and annoying and the rest of you should just skip to my next prediction, but here you go:
Comments: Part of my argument was that people were interpreting innocuous or stupid comments by Trump as endorsements of racism, Nazis, and the KKK. An implicit prediction of this was that, once Trump was President for a while and we had a much larger sample size of his comments, it would become clearer that he wasn't explicitly expressing support for racism.
In retrospect, this is a hard implicit prediction to grade. Suppose you're arguing against UFOlogists who point to grainy photos with vague splotches in the sky as evidence of aliens. You say "The future will prove me right!". Then the future comes, and a UFOlogist triumphantly shoves a new grainy photo of a sky splotch at you and says "Look! Time has only provided further proof of how many aliens there are." Of course if you disagreed about how to interpret current data, you should expect to run into the same problems about future data.
During Trump's presidency, I noticed two comments that people frequently held up as evidence for his explicit racism or pro-Nazi views.
In practice, the main post-election comment everyone holds up as Trump Explicitly Saying Nazis Are Good was his post-Charlottesville comment, where he said that there were "good people on both sides" of a protest that included alt-right figures. What actually happened: Trump said that "We condemn in the strongest possible terms this egregious display of hatred, bigotry, and violence." Later he said of the protesters that "You had some very bad people in that group, but you also had people that were very fine people, on both sides". When asked to clarify he said "I’m not talking about the neo-Nazis and the white nationalists -- because they should be condemned totally. But you had many people in that group other than neo-Nazis and white nationalists."
So Trump specifically condemned neo-Nazis and white nationalists, then said that - although he definitely was not talking about neo-Nazis and white nationalists, there were some good people in the protest. In the exact sort of UFO-splotch-pareidolia I had originally complained about, this somehow turned into a universal agreement that Trump had said neo-Nazis and white nationalists were good people.
Might there still be a case against Trump, on the grounds that all Charlottesville protesters were neo-Nazis and white nationalists and Trump surely knew this and was just dog whistling by pretending some weren't? No - in fact, there were many non-neo-Nazi protesters there - the New York Times did an interview with one of them (CTRL+F "Michelle Piercy"). This is just a total refusal to care what Trump actually said, combined with lies to make the deliberate misrepresentation sound defensible. It's exactly the sort of willful misreading that I complained about in my post.
The second comment that people read as Trump endorsing neo-Nazis was his supposed refusal to condemn white supremacy during his debate with Biden. Here's the transcript:
WALLACE: Are you willing tonight to condemn white supremacists and militia groups?
WALLACE: And to say that they need to stand down and not add to the violence in a number of these cities as we saw in Kenosha, and as we’ve seen in Portland
TRUMP: Sure, I’m prepared to do it, but I would say almost everything I see is from the left-wing not from the right-wing. I’m willing to do anything, I want to see peace.
WALLACE: Then do it, sir.
BIDEN: Do it, say it.
TRUMP: What do you want to call them? Give me a name.
WALLACE: White supremacists and right-wing militias.
BIDEN: Proud Boys.
TRUMP: Proud Boys, stand back and stand by. But I’ll tell you what, somebody’s got to do something about Antifa and the left.
When asked to condemn white supremacists and militia groups, Trump immediately said "Sure". Wallace asked if he was willing to tell these groups to stand down. Trump said yes. Wallace kept pushing him in a way suggesting he wanted a message targeted at a specific group. Trump asked who specifically who specifically Wallace was asking him to say should stand down. Wallace started to say "white supremacists", but Biden interrupted with "Proud Boys". Trump said the Proud Boys should "stand back and stand by", which as far as I can tell is equivalent to "stand down", the specific thing Wallace kept trying to make him say even though he had already condemned everybody.
I'm trying as hard as I can to be charitable - maybe people thought Trump should have followed Wallace's exact wording and said "Stand down" instead of the more temporary-sounding "Stand back and stand by"? Or maybe they got confused by the crosstalk between Wallace and Biden, and thought Trump should have said this to white supremacists in particular (a third time) instead of to the Proud Boys? But it's really hard for me to come up with a narrative where someone trying to comprehend this honestly could see it as Trump endorsing the white supremacists who he had already condemned just seconds before.
Apparently these two events were the most racist-sounding comments anyone was able to get out of Trump in four years of being President. And they're pathetic. The slightest attempt to read the transcripts of what he was actually saying makes it super clear how pathetic they are. I think my implied prediction that Trump wouldn't make especially racist comments during his presidency proved correct.
I also think events proved me right in saying that the media was going crazy in a particular way where they would read racism into anything. During the Trump administration, we got stories about how Trump watching fireworks at Mt. Rushmore meant he was "glorifying white supremacy", about how Bill de Blasio was blaming the Jews for coronavirus, and about how Pete Buttigeg was "openly racist" and riding a tide of "white supremacy". Elizabeth Warren is a fascist who supports ethnic cleansing, the Boy Scouts have effectively become a fascist youth group, JK Rowling is aligned with white supremacists, etc. And don't forget the stories about how a prison guard's tattoo secretly encoded white supremacist symbols (it ended up being his college football uniform number), that the number of words in Homeland Security documents could be interpreted as a white supremacist code, or that Melania Trump had secretly encoded the letters "KKK" into the new design for the White House Rose Garden (for a totally different conspiracy that the new Rose Garden encodes Nazi symbology, see here). At some point you just have to admit everyone went crazy for a few years and seeing started seeing Nazis in trees and rocks and grilled cheese sandwiches and Trump was an especially tempting target.
Policies: In my post I specifically exempted immigration - I thought Trump was likely to pursue unusually anti-immigrant policies and said so at the time - but said that otherwise Trump wouldn't pursue any explicitly or openly racist policies, and that in fact his racial policies would be bog-standard for a Republican.
As far as I can tell this was basically true. He seems to have done relatively well for himself with eg prison reform. He responded to the George Floyd killing by condemning it, signing an executive order pushing police reform, and expressing support for ending chokeholds by police. I'm not aware of any non-immigration race-related policies he pushed that were especially controversial.
Racist supporters: Around the 2016 election, there were a lot of stories about the KKK marching for Trump, about how arch-alt-rightist Richard Spencer was leading white nationalists in Nazi salutes for Trump, et cetera. We were promised an age of renewed white supremacist activity, supported by or maybe actively merging with the government.
Needless to say, none of that happened. White supremacist activity in the United States never went beyond normal levels. After Charlottesville, the media tried to convince everyone that this was the promised spike in white supremacist activity, but it wasn't: there had been a bunch of white supremacist rallies larger than the Charlottesville protest throughout the Clinton, Bush, and Obama presidencies - the media just hadn't signal-boosted them as hard. 500 Nazis rallied at Lake Tahoe in 1996; 125 at Yorktown in 2005; 2000 in Budapest in 2009. Charlottesville's couple hundred were kind of par for the course - and all of these numbers are around the same number of people who eg attend Satanist rallies. After Charlottesville there were no further large white supremacist rallies, with an attempted Charlottesville II attracting about 20-30 people.
Further, white supremacists generally stopped being interested in Donald Trump, and have mostly abandoned him. In 2018, Southern Poverty Law Center wrote Are White Nationalists Turning On Trump? - by now, the answer is clearly yes. From VOA News: Once Ardent Trump Supporters, White Nationalists Splinter Ahead Of Presidential Vote. Richard Spencer, who invented the term "alt-right" and whose Trump salute helped spark the panic, disavowed Trump in early 2020 and announced he was supporting Biden in the election. White supremacists are irrelevant and you should not care what they think, but if for some reason you insist on caring what they think, they are no longer particularly pro-Trump.
At the same time, Trump's support among people of color has grown significantly. Although (as mentioned before) the racial gap in partisan support was already starting to close in 2016, this accelerated throughout Trump's presidency, with 2020 Trump getting the most black and Latino votes of any modern Republican candidate. Trump also significantly lost ground with white voters.
I think this is probably the clearest indicator I can come up with for whether or not my prediction has proven true. Trump lost the support of white supremacists, lost the support of whites in general, and gained support from minorities. This picture is consistent with people of all races not viewing him as conducting an especially pro-white or anti-minority Presidency.
Overall I give myself an A on this prediction, and am delighted to have outlasted all the people who hated that post and kept trying to convince me it "wasn't aging well" or whatever. (Ann Coulter still sucks though).
I will still never predict anything this controversial again - it’s not worth the cost to my peace of mind.
6: 01-04-2017: Trump will strong-arm people into giving him good photo ops, which will make him look successful (F)
In Trump And The Batman Effect, I updated way too hard on Trump convincing a couple of companies to keep their jobs in the US during his first month after winning the election, and decided that Trump's overall plan for his presidency was to strong-arm specific people into doing a few good things and bask in the credit, while everything generally got worse all around him.
In retrospect this wasn't a particularly coherent idea, and Trump didn't really do this. It was barely a month after Trump got elected and I think I was desperate to have some kind of Trump-related take that hadn't already been done to death. This was the best I could do, and in retrospect it was stupid. F.
7: 1-2017 through 1-2020: Various predictions about the Trump administration over the next year (B)
This is my dump for all my other yearly prediction threads. Ones that came true are left intact, ones that proved false are starred, ones that I wasn't able to judge are italicized:
From 1/2017, about 1/2018:
41. Donald Trump remains President at the end of 2017: 90%
42. No serious impeachment proceedings are active against Trump: 80%
***43. Construction on Mexican border wall (beyond existing barriers) begins: 80%
44. Trump administration does not initiate extra prosecution of Hillary Clinton: 90%
***45. US GDP growth lower than in 2016: 60%
***46. US unemployment to be higher at end of year than beginning: 60%
47. US does not withdraw from large trade org like WTO or NAFTA: 90%
48. US does not publicly and explicitly disavow One China policy: 95%
49. No race riot killing > 5 people: 95%
***50. US lifts at least half of existing sanctions on Russia: 70%
51. Donald Trump’s approval rating at the end of 2017 is lower than fifty percent: 80%
52. ...lower than forty percent: 60%
From 1/2018, about 1/2019:
1. Donald Trump remains president at end of year: 95%
2. Democrats take control of the House in midterms: 80%
***3. Democrats take control of the Senate in midterms: 50%
***4. Mueller’s investigation gets cancelled (eg Trump fires him): 50%
5. Mueller does not indict Trump: 70%
***6. PredictIt shows Bernie Sanders having highest chance to be Dem nominee at end of year: 60%
7. PredictIt shows Donald Trump having highest chance to be GOP nominee at end of year: 95%
8. [This was missing in original]
***9. Some sort of major immigration reform legislation gets passed: 70%
10. No major health-care reform legislation gets passed: 95%
11. No large-scale deportation of Dreamers: 90%
12. US government shuts down again sometime in 2018: 50%
13. Trump’s approval rating lower than 50% at end of year: 90%
***14. …lower than 40%: 50%
***15. GLAAD poll suggesting that LGBQ acceptance is down will mostly not be borne out by further research: 80%
From 1/2019, about 1/2020:
1. Donald Trump remains President: 90%
2. Donald Trump is impeached by the House: 40%
***3. Kamala Harris leads the Democratic field: 20%
***4. Bernie Sanders leads the Democratic field: 20%
5. Joe Biden leads the Democratic field: 20%
***6. Beto O’Rourke leads the Democratic field: 20%
7. Trump is still leading in prediction markets to be Republican nominee: 70%
8. Polls show more people support the leading Democrat than the leading Republican: 80%
9. Trump’s approval rating below 50: 90%
***10. Trump’s approval rating below 40: 50%</s>
***11. Current government shutdown ends before Feb 1: 40%
12. Current government shutdown ends before Mar 1: 80%
13. Current government shutdown ends before Apr 1: 95%
***14. Trump gets at least half the wall funding he wants from current shutdown: 20%
15. Ginsberg still alive: 50%
Are there any consistent patterns here?
I underestimated how well the economy would do the past four years. I think this was less about underestimating Trump's economic leadership, and more about guessing the economy had nothing to do with the president but assuming it would cool down after a longish recovery under Obama. In fact it kept getting better and better. I'm a little nervous about this because the economy is one of the few gauges of a president's performance that has really objective indicators, and it did well under Trump, which is a little bit of evidence he might be more competent than I thought? But it's not very much evidence and for now I'll stick to my guess that it doesn't have much to do with the president and he got lucky.
I overestimated the likelihood of Trump building the border wall, passing immigration reform, or otherwise delivering on his campaign promise to fight immigration. I think this was partly me overestimating his honesty and competence, and partly me partly falling victim to the panic about how Trump was about to deport all immigrants and only the #Resistance could stop him. Although I was able to partly resist that panic when writing my Crying Wolf post, I was less able to resist the immigrant-related piece of it and notice that Trump wasn't really competent / dedicated enough to accomplish anything here.
8: 2-15-2018: Predictions on the next five years, especially the 2020 elections (C)
As part of SSC's five year anniversary, I tried to predict the following five years. Three of those years have happened now. The Trump-relevant predictions were:
The crisis of the Republican Party will turn out to have been overblown. Trump’s policies have been so standard-Republican that there will be no problem integrating him into the standard Republican pantheon, plus or minus some concerns about his personality which will disappear once he personally leaves the stage. Some competent demagogue (maybe Ted Cruz or Mike Pence) will use some phrase equivalent to “compassionate Trumpism”, everyone will agree it is a good idea, and in practice it will be exactly the same as what Republicans have been doing forever. The party might move slightly to the right on immigration, but this will be made easy by a fall in corporate demand for underpriced Mexican farm labor, and might be trivial if there’s a border wall and they can declare mission accomplished. If the post-Trump standard-bearer has the slightest amount of personal continence, he should end up with a more-or-less united party who view Trump as a flawed but ultimately positive figure, like how they view GW Bush. Also, I predict we see a lot more of Ted Cruz than people are expecting.
1. Trump wins 2020: 20%
2. Republicans win Presidency in 2020: 40%
Prediction 1 was directionally correct. Was it overconfident? It feels like I was overconfident to say only a 20% chance for Trump to win way back in 2018, when in fact Trump came kind of close. But my prediction on Election Night was still only about 20%, 538's model for most of the 2020 predicted about 20% or less, so I think this might have been just right.
But Prediction 2 suggests I thought there was only a 50% chance Trump would win the Republican nomination. In retrospect that was way too low. I think I genuinely should feel embarrassed about this one - I put myself in the Republican Party's shoes and imagined I would reject Trump, ignoring all the evidence that actual Republicans liked him quite a lot and would probably continue to do so.
The long paragraph sort of flips in and out of making that mistake. I think I was right that the Republican Party wouldn't have some crisis about whether or not to accept Trump. But I think I was predicting that would be because (the public image of) Trump would shift towards the Republican consensus. But in fact the Republican consensus moved towards Trump. I think I got the non-crisis right, but the way it happened wrong. The Republican Party has fundamentally changed to be a paranoid victimization-narrative-based party. Trump got there first, everyone else followed, and while the details can change I don't think they're going to flip back.
I don't think we've seen any more of Ted Cruz than anyone expected.
Overall I give this section a C.
9: ~Autumn 2020 Semi-Prediction: QAnon was a moral panic and you should ignore it (D+)
I'm listing this as a semi-prediction because I never actually made it; I took the blog down before I could say anything about this formally. You will just have to trust me that I believed this. Still, I believed it very hard, and it's worth doing a postmortem on.
My argument was something like: QAnon are, for real, the sort of people that Democrats like to imagine all Republicans are. They're crazy and have no valid concerns. So Democrats had a very strong incentive to exaggerate their role within the Republican party and convince people that QAnon was the driving force behind Trump support. Nobody ever gave an estimate for how many people were in QAnon, and every report on it was consistent with "a tiny handful". There was also a big attempt to make QAnon look violent by everyone giving long lists of "Violent Acts Caused By QAnon", when you examined them closer, literally did not include a single violent act caused by QAnon (it was all either nonviolent acts, plots that never materialized, or criminals who were incidentally QAnon supporters).
Later some people did a terrible mismanaged poll which purported to show that 35% of Republicans were QAnon supporters, and then someone did a better poll which showed 1-2% of Republicans were QAnon supporters. Everyone reported on the former, nobody reported on the latter, and this left me pretty annoyed.
In ~November 2020, I did a survey of Twitter in which I searched the keyword "QAnon", and the most recent 100 results were all Democrats attacking QAnon; no matter how far I scrolled I could not find any actual QAnon supporters. But I don't know if this was because vocal QAnon opponents actually outnumbered QAnoners by that much, or because Twitter is doing a really good job censoring them (is it even possible to censor supporters using a keyword this successfully without opponents getting caught in the crossfire?) Either way, this added to my evidence that this was a pointless moral panic.
But I've since come to count this as a mostly false prediction. The main thing was talking to a few patients and SSC commenters who said their family members strongly supported QAnon in ways that put serious strain on their families; since these are a semi-random sample of the population, that suggests it's pretty big (not necessarily bigger than the 1% to 2% of people in polls, but 1% to 2% of people being strong active supporters of a conspiracy theory is actually a relatively big deal.
The Capitol riot didn't particularly make me update further on this, both because I had already updated, and because according to researchers only 8% of Capitol rioters were QAnon believers (although this was definitely the 8% with the most interesting fashion sense). This confirms to me that QAnon is a small minority, but not an irrelevantly small minority, of extreme Trump supporters. I also feel like it could have gotten worse/more important in a way that made it worth watching, although luckily in this timeline it seems to have at least partly imploded after Trump lost the presidency.
I think it's plausible that QAnon was 1-8% of Trump supporters, which is much larger than I believed at the start, although still not as large as I felt like people were trying to make me think.
Overall I judge this one a minor-to-medium failure, and give it a D+.
10: ~Autumn 2020 Semi-Prediction: There will not be a Trump coup (B)
Another one I never got the chance to say publicly, and you'll just have to trust that I was thinking it. My argument was basically the same as Maciej Ceglowski's argument here .
I graded this one a B. A lot of people who did predict a coup took the Capitol riot as proof that they were right. I didn't see it this way, because, well, it wasn't a coup. When rioters got in the Capitol, Trump tweeted saying that they needed to stay peaceful, then released a video saying that this wasn't what he wanted and they should leave. I've read Edward Luttwak's guide to holding a coup, and telling your supporters "stop and don't hold a coup" is not in it. Actual coups involve interfacing with the military, friendly media, local police forces, etc and coordinating some kind of meaningful strike. Trump did none of that and was clearly not capable of it. Trump's plan, as far as it's worth dignifying with that term, was to challenge the election results a lot, and get a bunch of angry supporters to chant outside the Capitol in a scary way while senators were voting on it. This is evil, anti-democratic, and terrifying, but not a coup. When his plan for angry supporters chanting turned into angry supporters rioting and getting into the Capitol, he was clearly against this and tried to stop it. While he is an idiot for not anticipating this possibility and deserves blame for it, I don't think it qualifies as a coup, and I think everyone who gave arguments for why an actual coup was unlikely were basically right.
I did make a bet with someone who thought there would be a successful coup (eg Trump would remain president after January 20th despite losing the election), and won $100.
Overall I give this one a B.
11: Bonus - How much money did I make on prediction markets? (B+)
During the 2016 - 2020 period, I quadrupled my money on PredictIt.
I started using PredictIt during the 2016 election (I bet $20 on Hillary and obviously lost everything). After that I decided it would be fun to start with some seed amount and see what happened to it over time, so I put in another $200 and was a lot more careful with it. Over the course of a few dozen bets, I was able to turn my initial $220 into about $900 by the end of the 2020 election.
I bet on a lot of things, but the ones where I put a lot of money in and really won big were:
- Predicting that Marine Le Pen would not be elected President of France. A lot of people who overupdated on Trump and Brexit decided that polls didn't work anymore and the far-right would win no matter what the polls said, and they bought her up to a price much higher than her numbers justified. I decided Le Pen's polls were much worse than Trump or Brexit's and I was comfortable assuming the polls at least kind of reflected reality, so I bet against her and made about $100.
- Predicting that the 1/2019 government shut down would last a long time. I figured there was more partisanship than ever before, neither Trump nor Congressional Democrats seemed like the sort of people who would back down, and so it was fair to take an Inside View and assume this shutdown would last longer than previous ones. I made another $150.
- Predicting that Biden would win the Democratic nomination. I bought up a thousand shares of Biden for 16 cents a share in early 2019, which was an amazing move - then proceeded to screw it up in as many ways as possible. I wondered if maybe Sanders or Warren or O'Rourke would win, so I bought a bunch of them too. Then the one week when Warren was doing super-well, I panicked that I had missed out on the real winner and bought a lot of Warren, then sold her for really low again once she crashed. None of these mistakes were able to completely wipe out my gain from buying so much Biden so early, and I made another $200 or so.
- Predicting that Biden (and the Democrats) would win the election. I kept buying this throughout 2020, on the basis of the spread between PredictIt (which gave Trump about 45% chance) and 538 and other top pundits (who gave Trump about 20% chance). This never stopped seeming like a good deal, so I just kept buying more and more of it as my other bets closed and more and more funds became available. I think I won about $200 on this, though it's complicated for me to say exactly how much because of the next point.
- On Election Night 2020, after Trump started outperforming his polls, shares in Trump winning briefly reached 70 - 80%. Running the exact degree to which Trump was outperforming his polls through 538's model, I determined there was still only about a 20 - 30% chance he would win. I had originally planned to see how much I could make off my original investment alone, but this seemed so much like free money that I took out my credit card and put a few thousand more dollars on Biden. I won about $2000 off this. I can't count it in my official score because it wasn't part of the original investment, but I would have been really angry at myself if I hadn't done this.
Along with these good bets, I lost a bit of money on some other bets, including Warren winning the nomination, a lot of speculating on Trump's exact number of tweets and various candidates' exact polling averages, and a disastrous assumption that surely they couldn't elect Binyamin Netanyahu as Prime Minister of Israel yet another time, could they?
Looking over my history, I find that I did well when I was speculating on very high-level important events I'd been following closely, like who would win the election, and poorly on low-level things I tried to do math on, like how many times Trump would tweet in a given week. This is the opposite of what I originally expected - I'd thought more money and analysis would go into election predictions, and the small niche markets would be where you could eke out a gain. I think what probably happened is that the election markets attracted a lot of dumb money that I could beat, and the tweet markets were mostly semi-professional investors who were much better at mathematizing it than I was.
Overall I approximately quadrupled my original $220 investment into about $900. I plan to take out all the money I won except about $1000 and see how much I end up with at the end of the first Biden administration.
If I average out the grade I (unprincipledly) gave myself on each of the sections, I got a C. I tried to assign grades so that doing as well as an average pundit = a C, so I think I did about average.
According to my own judgment, I usually did better on predictions about race, and worse on other things. An optimistic take on this is that race has become so emotionally charged that most people have kind of crazy beliefs about it, which makes them easier to beat. A pessimistic take is that race has become so emotionally charged that everyone including me has crazy beliefs, which makes me a more biased judge and lets me award myself points more shamelessly than I would do anywhere else. The few cases where we have objective numbers we can use to judge (eg the changing racial distribution of voters) pushes me more towards the optimistic view on this.
I also seem to have done better on the prediction markets than in my letter grades and scored predictions. I think this is because the dumb money I bet against on prediction markets was dumber than the hypothetical "average pundit" I was grading myself against. Real pundits generally gave Trump 20% odds compared to prediction markets' 40% odds, so it was easier to beat the latter.
Did my liberal bias lead me to underestimate Trump? I'm not sure. He did better than I expected on the economy and on not starting wars. But he did worse than I expected on getting any of his policies enacted. Maybe this is what I should expect if I was suffering from liberal bias - maybe the liberal narrative was "Trump is an evil supervillain who will successfully complete all kinds of terrible things", which made me underestimate how well Trump would do on things I liked, but overestimate how well he would do on things I hated. But I don't know if that was actually the liberal narrative - there was at least as much "Trump is a completely incompetent buffoon." Overall I think the real Trump administration was in the 60th percentile of my pre-administration range of estimates of how bad it could be (where 0th is the best I could possibly imagine, and 100th is the worst I could possibly imagine) - but this is almost entirely because of the botched coronavirus response. In terms of non-coronavirus stuff it was probably the 20th percentile. I don't want to say this means "I overestimated its badness" because the botched coronavirus response was a real thing that happened, and part of the way I expected the Trump administration to be bad was to botch any crises it ended up in. Overall I don't think I found I had significant liberal bias.
Did my less-liberal-than-a-lot-of-the-media bias lead me to overestimate Trump? I think no - I had a low enough opinion of him that it would have been hard for it to be an overestimate - but I might have overestimated his supporters. I somewhat misjudged the severity of QAnon, and I didn't expect not-originally-Trumpist Republicans to go over to him in the numbers they did.
I think the biggest mistake I made was to expect Trump to be more competent (at achieving his own goals) than he was. Trump's winning the election was weird enough that I think everyone was flirting with Scott Adams' thesis that he was some kind of bizarre "clown genius" or "drunken master" type whose apparent bumbling was just him playing so many levels above the rest of us that we couldn't understand it. I think something like this might be true for his ability to speak to the Republican base, but it obviously didn't translate into politics, policy, or having any idea what to do with the Presidency. Possibly my underlying error here was overestimating the correlation between different types of intelligence - just because someone has figured out a weird drunken-master strategy for winning the hearts of conservatives doesn't mean he has any special talent at anything else. I think it took me a year or two to fix this misapprehension, aided by things like his split with the potentially-competent-Machiavellians in his administration like Steve Bannon or Peter Thiel, and the constant churn of corrupt and ineffective appointees.
I think the second biggest mistake I made was to expect the Republican Party to eventually see through him and stop liking/respecting him. Trump's approval rating the month after he took office was 40%. Last month it was 39%. The guy wasn't exactly popular, but he held onto the popularity he had for dear life, and that 40% of people never abandoned him.
Normally I wouldn't be sure it was worth continuing to talk about politics if I was barely adding anything to the average pundit. But given that I got one or two things right that everyone else got wrong, maybe having an extra person giving additional perspectives is worth it on wisdom-of-crowds and marketplace-of-ideas type grounds.
On the other hand, couldn't I make the same argument even if I'd done much worse than the average pundit, as long as I got at least one thing right? I'm not sure. It seems like we're located somewhere between an ideal marketplace of ideas (where perfect Bayesians update the exact right amount on each new idea they hear, so each idea adds something) vs. a flock of sheep (who just update to whatever the last thing they heard was, so if you're worse than average you're bringing things down)
I currently intend to blog about current-events-style politics less than before but still a non-zero amount, as an average of those considerations. You might have already noticed this. I will still be recording predictions about the Biden administration, and depending on how I do, I might increase or decrease the current-events-blogging for the next administration.