Introducing my personal prediction database
I show off my predictions, and answer some questions
I’ve been interested in forecasting for some time now. I frequently post questions to Metaculus, and spend a lot of my time thinking about how to predict the future. That’s why I’ve decided to create a public database of my predictions, which I plan to update semi-regularly. You can access the database here.
FAQ
What are your predictions so far?
I started by filling out 10 starter predictions. Here they are,
The database link contains more information about each prediction, including a potential justification and sources used for resolution. I aim to make sure each question has little to no ambiguity in the way that it’s resolved.
What’s the point?
This way I can be held accountable for incorrect predictions, by having a public log of a lot of the stuff I believed. That should be good for others who are skeptical of my forecasting abilities and seek to expose my unreliability. Or it might be good for me to brag about how I’ve been correct in the past.
Aren’t existing platforms for prediction logging good enough?
Maybe, but I wanted to have more control over the database, and especially what information is used for resolution.
Are you open to betting?
Yes, but it depends. In the spreadsheet, there’s a “Confidence” tag which should loosely correspond with how open I am to betting on the prediction. Roughly speaking, I will probably be OK betting on questions with “High” confidence, and might be OK betting on questions with “Medium” confidence. However, in some cases, my prediction might be seriously out of date, and I reserve the right to say “that was an old prediction, and I’ve changed my mind since then.” I also won’t guarantee large bet sums.
Are you open to a friendly prediction competition?
Yes, absolutely. Give me a set of interesting questions, a reasonable time frame for resolution, and a reliable arbiter for their resolution. I’ll want to do it.
Might this make you more closed-minded?
Being open about one’s beliefs is sometimes bad because humans have a tendency not to want to switch their beliefs in public (when politicians do it in the US, they are accused of “flip-flopping” and this is supposed to be bad). I’ve thought about this effect, and I agree it’s a problem. But, I think I’m surrounded by peers who are abnormally good at not judging me for changing my mind; in fact, I may get judged for not changing my mind enough.
Couldn’t you just lie and change your beliefs retroactively to make yourself look smart (since you control the database)?
Yes, but that wouldn’t be very ethical. Feel free to distrust me if you wish. That said, Google Drive keeps a copy of every change you made to a given document, so if you really doubt my integrity, it should be possible to verify it in principle.
At some point I may establish a feed which provides links to each of my predictions, with a trusted third party that can instantly verify that I indeed made that prediction at that particular time. For now I just have this spreadsheet.