Last year in May I made some predictions. Now it’s time to find out how I did!
The summary is I was wrong. A lot. This should be no surprise.
First, let’s take a look at my overall accuracy:
The source data for this chart is a spreadsheet I made for my 2020 predictions. Overall, this should be fairly understandable but there’s one nuance that needs some explaining. In order to make the chart simpler, I converted any prediction for less than 50% to its inverse and graphed that.
Here’s an example. I estimated a 10% likelihood that “a vaccine is generally available by end of 2020”. But another way to think of it is that I estimated a 90% likelihood that a vaccine is not generally available by the end of 2020.
If my predictions were perfect, then 60% of my predictions of 60% likelihood would have happened, 70% of 70%, and so on. But I wasn’t even close! Only 33% of my 60% predictions happened. I did even worse at 70% (33% happened), improved a little at 80% (50%), and redeemed myself a little at 90% (80%) and 95% (100%).
The accuracy trendline goes in the right direction, but it’s way too steep. I clearly have work to do on becoming an expert prognosticator.
Let’s look at each of my predictions in detail. Only the text in bold is new. The rest is copied from my original predictions post.
Politics and Economy
- Joe Biden is the Democratic nominee come November voting: 95%
- Trump is the Republican nominee: 95%
- November election proceeds normally: 95%
- By “normally” I simply mean that the election occurs on the scheduled day in all 50 states. I think states may also expand voting by mail, but no state will cancel or postpone the election. I do expect some states to use the pandemic as an excuse for even more voter suppression, but sadly that fits the definition of “normal” in the US.
- Yup. While shit went crazy after the election, the election itself was close enough to normal that I will count myself correct on this one.
- Trump wins the November election: 60%
- Nope. I should have counted on Trump’s incredible incompetence at managing both coronavirus and the economic downturn that it caused. I think this is what cost him the election.
- Democrats maintain control of the House: 80%
- Republicans maintain control of the Senate: 70%
- This is mostly based on which states are having senatorial elections this year. Nearly all of them are Republican strongholds.
- Nope. But I stand by my 70%. This was incredibly close. I didn’t predict that Trump would completely lose his mind and engage in a series of escalatingly insane attempts to overturn the election, which I think ended up tipping the Georgia Senate votes to the Democrats.
- US GDP for 2020 is down by at least 30% year over year compared to 2019: 80%
- Nope. Looking at one source, it shows the US economy as down 3.5% for the year. Thinking back to this, I didn’t actually do any research on past annual GDP changes. If I had, I would have realized that 30% in any direction is utterly ridiculous. On the site I linked before, the biggest change was 18.9% in 1942, in the middle of WW2. Was coronavirus a bigger event than WW2? No, it definitely was not.
- Minnesota lifts stay at home order but then reinstates it at least once: 90%
- Nope. I was tempted to give myself a “Yup” here, but we didn’t get a second stay-at-home order in December. We got something that was fairly similar, but not quite the same.
- Three months or more of cumulative stay at home in Minnesota: 80%
- Nope. I didn’t realize the level of insanity that would arise around restrictions, which made it politically impossible for any governor to be this strict, regardless of whether the circumstances warranted it. But if I thought about this a bit more deeply I think I would’ve realized how unlikely this was.
- Five months or more of cumulative stay at home in Minnesota: 20%
- Over 100,000 in the US dead from coronavirus: 90%
- Over 250,000 in the US dead from coronavirus: 40%
- Nope, more than 250,000 died. Sigh.
- A vaccine is generally available by end of 2020: 10%
- By “generally available” I mean that it’s not in clinical trials, that it’s available in sufficient quantities for use as needed, and that it is safe to be given to at least 90% of the population.
- Yup, no vaccine was generally available by the end of 2020. But good job vaccine makers for nearly making me wrong here!
- A generally available therapy exists that reduces mortality by 30% or
- This can be either be preventative or something that reduces severity of the symptoms.
- WTF? How did I think I could possibly figure this out? Well, I just tried and I can’t. Looking at Our World in Data, it seems like the case fatality rate went from a peak of 6.2% to 1.7%, but I have no idea why. I didn’t factor this prediction into my accuracy chart because it’s not really assessable for accuracy.
- I have had coronavirus: 40%
- This is based on either a test or my best guess based on symptoms and contacts.
- Nope, I haven’t had it? I have had both an antibody test (November 18, 2020) and a test for the virus (in mid-January). Both were negative. However, I, my wife, and two close friends all got sick with what felt like very bad cold symptoms at the end of February, 2020. Maybe we had COVID, maybe we didn’t. We’ll never know. But I counted this as a miss in my accuracy chart.
- I am hospitalized for coronavirus: 5%
- Yup, I was not hospitalized.
- At least one person I know dies of coronavirus: 60%
- This hinges a lot on the definition of “know”. Let’s say it’s people I’ve met in person more than once and I remembered how I know them when I learned about their death. Public figures and celebrities don’t count (not that I’ve met many).
- Nope. I know more than a few people who’ve had it, but no one I know personally has died. Glad to be wrong.
- Fatality rate in the US is estimated at less than 1% in retrospect,
excluding any newly developed treatments from #13: 70%
- Note that even a “low” fatality rate like this is still very dangerous when combined with rapid spread and no vaccine/immunity.
- Nope with a side of WTF. This is phrased so unclearly that I’m not sure how I would’ve handled the whole “excluding any newly developed treatments” thing. But that’s moot since the fatality rate is clearly above 1%.
- I am still working for ActiveState: 95%
- Yup. And I still like working there.
- I have attended at least one non-Perl conference (in person or online): 60%
- What “attending” means for an online conference will be on the honor system.
- Yup. I attended RustConf 2020 back in August of 2020. It was great.
- I have released at least one CPAN module every month for 20
- Yup. I did this one.
- My weight is 210 pounds or below and has been since November 1: 70%
- See I Weigh Way Less for more than you ever wanted to know about this topic. For reference I was at 216 last time I weighed myself.
- Yup. My weight has been around 201-204 most times that I’ve checked over the past few months. There were a couple of days where it dipped below 200, but that was very temporary. I’m pretty happy with where I’ve gotten to. I don’t know how I’d lose significantly more weight at this point without adopting a much more drastic diet change than I’m up for.
- My weight is 200 pounds or below and has been since November 1: 10%
- Yup. My weight was not below 200 consistently.
- I have climbed at least one climbing route (top rope or lead) rated
5.11(-/a) or higher: 60%
- This mostly reflects my prediction about access to climbing gyms and outdoor climbing this year. Absent coronavirus I’d have put this at 95%. Note I consider this to have happened even if I fall or rest during the climb and then continue. This is just about whether I can top out at all.
- Nope. I haven’t been back to the climbing gym since February of 2020. I did take a top rope anchors class, and I did some outdoor top rope and lead climbing, but not nearly enough to improve, and I haven’t been climbing in several months now. I will not be surprised if I’m not even able to do 5.10+ next time I actually go to the gym.
- I have a bouldering wall in my garage: 80%
- Nope. My mother threw a monkey wrench into this one by dying last September. That set in motion a plan where my father moved here and we purchased a new duplex with him. My father is currently living in his part of it that is his, and we’re having major renovations done on our part with a plan to move in April or May. Once that started it was obvious there was no point in having a bouldering wall built in our current garage.
- I maintain my three hours (ish) per week weight training schedule (modulo
illness or injury that prevents me from doing so): 95%
- There’s a little leeway here. For example, yesterday I did 45 minutes instead of 60 because of some neck & shoulder issues, but I’d still count this week as a success.
- Yup. I’ve had some days/weeks off because of illness and injury, but overall I’ve stuck with my exercise plan. I’m probably doing about 150 to 165 minutes per week, which I will count as “three hours (ish)”.
- I’m still vegan: 95%
- If I let myself score 100% this would definitely be 100%.
- Yup. In retrospect I’m not sure if this sort of prediction is worth putting on the list. I could also put things like “I won’t be hit by a meteor” or “I will go shopping for food”. While these are likely to be accurate predictions, they don’t really do anything to evaluate my prognositication skills. I think I just juked the stats with this prediction.
- I spend time in Taiwan this year (whether on vacation or as a temporary
- Yup. I did not go to Taiwan in 2020. We’re hoping to go later this year, though.
I looked at the stats for my three categories and I didn’t see any huge difference in accuracy between them. The only obvious highlight is that things I predicted at 95% were the most accurate. But those were basically all sure things.
My main takeaways are as follows:
- If my initial certainty on a prediction is lower than 90%, I probably need to do some research or deeper thought to improve my accuracy.
- I should be careful to consider how I will actually figure out whether the predicted thing happened. There were a couple cases where I couldn’t really figure out how to do that this year.
- I’m not a superforecoaster. But I would’ve predicted that to be the case, so maybe I am, at least when it comes to predicting my forecasting abilities.
Stay tuned for 2021 predictions. Maybe. Right now I don’t have as many ideas for things I might predict this year. Politics and coronavirus are (hopefully) going to be less interesting this year, and I’m not sure there are other really interesting fields I can even attempt to make predictions in.