My current daily routine

Here is my current daily routine, which I’ve been following for a few months now with some modifications over time.

Morning

• Check email, Messenger, Facebook, LessWrong, EA Forum; I usually do this from my tablet while I’m still in bed.
• Check news (RSS subscriptions, urlwatch, and Andy Matuschak’s notes). I’m considering doing this only once every three days or so, because I’ve noticed that this is pretty time-consuming, even when I try to go through it quickly.
• Check Slack (for a project I am working on).
• Check Telegram.
• Start doing work: writing down ideas, reading things, thinking about things. This step continues throughout the day. My writing/reading/thinking process is pretty complicated and I won’t try to describe it here.

Day

• I continue working.
• Throughout the day I also continue checking email, Messenger, LW, and EA Forum. I should probably stop doing this, but it’s not a huge time-sink so I haven’t put much effort into getting myself to stop.
• Late in the afternoon I check Hacker News once. I think HN adds very little value to my life, but I still feel an urge to check, and there’s occasionally something interesting on there. (I think if there was never anything interesting on there, or if I could reliably be sent just the things I find interesting, then I wouldn’t feel an urge to check it so often.)

Evening

• Tao Analysis Solutions blog: I try to write one more blog post. Some of the exercises have multiple parts or are otherwise time-consuming, in which case I often make partial progress on writing a post without actually publishing.
• Anki review. This consists of a few things:
  • All my normal cards (e.g. about math, AI safety, economics)
  • Incremental reading cards; see Incremental reading in Anki for details.
  • Math problems cards; see Spaced proof review routine for details.
  • Russian; I currently have a very simple alphabet deck and a “frequently used words” deck, both of which I just downloaded from shared decks on AnkiWeb.
• Duolingo (Russian): I average around 40 XP per day, which takes 5-15 minutes.
• Russian immersion: At the moment I’ve been spending about 30 minutes per day just watching random videos on YouTube, e.g. setting my location to Russia and going to “Trending”, searching up random topics I am interested in, searching up names of TV shows or cartoons. I’m still figuring this out, e.g. how to make it more enjoyable.
• About once a week, I check Twitter, using this list (nitter).
• About once every couple of days, I check Jonathan Blow’s Twitch stream. If he happens to be live, I watch for a few minutes (usually around 5 minutes, maybe up to 30 if it’s especially interesting). Regardless of whether or not he’s live, I also check out the new clips.

2019-02-19

I continued with the caffeine reset (I again drank a very small amount of coffee).

I continued reading “The Psychological Foundations of Culture” by John Tooby and Leda Cosmides, as well as some other random things.

2019-02-18

I continued with the caffeine reset (I had a very small amount of coffee to help with the withdrawal).

I started reading again “The Psychological Foundations of Culture” by John Tooby and Leda Cosmides. This isn’t directly related to AI safety, but Eliezer Yudkowsky keeps mentioning this paper so I decided to read it. (See also this page.)

2019-02-17

I rested on this day (I stayed in bed and sat around daydreaming the whole day, and minimized the amount of reading and writing I did) and started a caffeine reset.

Regarding the severe form of rest, I was curious to see if I would feel more rested by cutting out reading and writing. (If you spend any time reading rationalist writing, you will hear people talking about sabbath.) As of 2019-02-19, I think it helped a little but probably not much. My guess is that I get enough rest each night and that I don’t have any personal problems that are helped by this sort of intensive rest/introspection. It could also be that caffeine withdrawal is making it hard to tell how rested I am.

2019-02-16

I continued with causal inference.

2019-02-15

I continued learning about causal inference.

2019-02-14

I returned to learning about causal inference, mostly from Pearl’s book but also this overview paper.

I added a bit to the do operator page.

2019-01-13

Not much math; I spent the day relaxing, reading things, socializing, and thinking about some things.

I made some additions to the List of important distinctions in mathematical logic page.

2019-02-12

I again didn’t do much math.

2019-02-11

I didn’t do much math; I wrote some scribbles about the page List of important distinctions in mathematical logic.

I spent the day reading random things (I was curious about top LessWrong posts I had missed, so I spent some time looking through highly upvoted posts, e.g. the top list for 2018).

2019-02-10

I started writing List of important distinctions in mathematical logic.

I think I read more of Peter Smith’s Gödel book.

2019-02-09

I thought a bit more about the “two flavors” of incompleteness.

I thought about the Löb’s theorem cartoon and puzzle. My take on the cartoon is that it didn’t help me understand the theorem any better, even if it is a cute cartoon; I found myself having to translate the cartoon proof back into logic, which seems like evidence against the native architecture idea that led to the cartoon (or at least, evidence against this specific execution having succeeded). I started working on a page about the puzzle.

2019-02-08

I reviewed the “two flavors” of incompleteness theorems and the easy incompleteness proofs in Peter Smith’s book (chapters 6 and 7).

I started thinking about undecidability in logic, and wrote the page for the Entscheidungsproblem. I made some embarrassing realizations, like finally understanding that all the common undecidability statements (for first-order logic) are equivalent, and also realizing that undecidability for a logic is different from undecidability for a theory. It seems totally insane to me that this isn’t emphasized in the resources I have been consulting.

2019-02-07

I continued thinking about the diagonal lemma (I mostly used Peter Smith’s book, some online postings, SEP, and Gaifman’s paper). I think it was on this day that I realized that the proofs of the Rogers fixed point theorem and the diagonal lemma are basically the same, and that the theorems themselves are basically the same (I found a couple of papers and some online postings stating this connection, but I wish more textbooks talked about this).

I wrote the diagonalization lemma page.

2019-02-06

I continued trying to understand Gödel’s first incompleteness theorem, diagonalization lemma, etc. Mostly via Leary & Kristiansen’s book, but also Peter Smith’s book. Also this page on SEP.

I started the expresses versus captures page.

2019-02-05

I spent some time reading about heavy/fat/long-tailed distributions and the distinctions between these terms. I’m frustrated that people seem to use two different visualizations (the quantity vs individual visualization, and the frequency/density vs quantity visualization) but usually don’t mix the visualizations and don’t prove the equivalence between them either.

I started the Multiplicative process page.

I also started reading more about Gödel’s first incompleteness theorem (Leary and Kristiansen’s book).

2019-02-04

I re-read the proof of existence of a computable infinite binary tree without any computable paths, in Stillwell’s Reverse Mathematics. I was able to follow the proof. If you remember, I was having trouble with this proof a while back. One of the errors I had made earlier is to assume that the tree nodes were integers, rather than finite binary sequences. Stillwell actually clarifies this in like the first sentence of the proof, but I still had the misconception! I must have been tired… A picture would have helped, and Stillwell does include one, but in an earlier section (that he references).

I did some searching on sets/relations versus sets/relations. They seem to be equivalent under some definitions, but I wasn’t able to figure out when they are vs when they aren’t.

Started on the function versus algorithm page.

2019-02-03

I worked through the proof of completeness of first-order logic in Goldrei and Leary & Kristiansen. I think I was able to resolve one of the questions I had about the proof (namely, why it is that we iterate the process of adding constants rather than just adding constants once). I then started working through some of the exercises in Leary & Kristiansen for that section.

2019-02-02

I spent the day reading random things and not doing math.

2019-02-01

I did various first-day-of-the-month bureaucracies and didn’t do math.