How ought I spend time?

post by Quinn (quinn-dougherty) · 2020-06-30T16:53:53.787Z · LW · GW · 3 comments

This is a question post.



What factors go into your reasoning about assigning relative weights to

  1. grinding through books
  2. projects, taking a shot at the possibility of a contribution
  3. reading current research

Personally, I never maintain 1 and 2 at the same time, and I tend to have >95% of my resources on either one or the other, leaving 5% or less for 3. The vast majority of the last 4 years I've been in 1-mode, my reasoning being that I can't make a contribution if I don't know anything, and I don't want to sink my time into bad projects.


answer by dqups1 · 2020-06-30T17:55:46.495Z · LW(p) · GW(p)

I'm not a huge fan of books tbh, I find that it takes a lot of effort and wasted efficiency to shift through the noise, of which, books have a lot. I tend to prefer number 2 and then refer to 3 or 1 as is needed to advance 2. The hands on experience helps you retain and learn best in my humble opinion. Finally, I think the most important factor is your drive to do each of these. If I hit a roadblock on 2, I can try and power through in a suboptimal mindstate or take on 1 or 3 with reknewed enthusiasm. Therefore, I would prioritize 2 but use my time in such a way that I'm focused on whatever I'm most interested in at the time and in that sense pursuing the activity in which my valuable time nets the greatest result in any category.

comment by romeostevensit · 2020-06-30T19:48:31.003Z · LW(p) · GW(p)

+1 most books are bad

answer by Kaj_Sotala · 2020-06-30T17:12:30.075Z · LW(p) · GW(p)

Whether they feel motivating/exciting/interesting enough for me to actually carry them out. Things like "will this have an impact on anything" do get factored into that, e.g. completely pointless activities often feel less interesting. But my conscious mind only catches glimpses of that calculation, so I wouldn't call it "reasoning".

answer by bvbvbvbvbvbvbvbvbvbvbv · 2020-07-01T10:54:51.746Z · LW(p) · GW(p)

Funny you mention this as, starting tomorow, I am writing a script that tries to solve this issue.

The idea is to have an sql database that contains everything you want to do. The python script picks 2 entries and ask "If you die in a year from now, wich activity minimized regret the most?". The answer will probably be a cursor from activity1 <-> neutral <-> activity2. The scores are then calculated using the ELO scoring. This pair comparison algorithm allows to quickly rank your lifegoals by importance.

But that's only half of it. It will also ask "which activity takes the most time to complete?/If you had one hour, which would be a better investment?" to rank activities by time to complete (also using ELO).

If you do a few comparison a day and don't add too many entries you can then display a rank by "importance score minus by time score" to see what is the most important thing to you that also takes the less time. Or something like that. I expect some fine tuning for this.

The idea is to have an optimized and dynamic todolist that can help you to jungle between short time goals and life goals.

I'm thinking of managing my "toread" list the same way, but automatically deriving the time score from the length of the article.

What do you guys think? Any idea how to make this better? I just started and plan to learn quite a bit by trying to make this work. Any opinion is much appreciated. I will put this on github at the end. Should I publish it here also?

Btw, the idea came to me after reading through this :

Thanks gwern!

comment by Evan Ward · 2020-08-07T04:21:57.223Z · LW(p) · GW(p)

It's great to see other people thinking and working on these ideas of efficiently eliciting preferences and very 'subjective' data, and building your own long-term decision support system! I've been pretty frustrated by the seeming lack of tooling for this. Inspired partially by Gwern's Resorter as well, I've started experimenting with my own version, except my goal is to end up with random variables for cardinal utilities (at least across various metrics), and I'm having the inputs for comparisons be quickly-drawn probability distributions.

Replies from: bvbvbvbvbvbvbvbvbvbvbv
comment by bvbvbvbvbvbvbvbvbvbvbv · 2020-08-07T08:47:22.029Z · LW(p) · GW(p)

Very interesting!

Could you explain the workflow?

Also, do you intend to make the code accessible?

answer by Dagon · 2020-06-30T21:00:01.404Z · LW(p) · GW(p)

What granularity of time are you talking about? When you "never maintain 1 and 2 at the same time", is that any given minute, or any given decade? For me, "background learning" includes #1 and #3, and for a given quarter I'm usually 25-75 between learning and doing, but the ratio reverses for some periods when I don't know how to approach a project or what project I might want to do next. On the timeframe of weeks, I might be 100% on one of these, but on the timeframe of months, I _always_ have some project/build/do time and some research/explore time.

I try to never "grind" through a book. Most books are simply not that necessary, and those that are, I can usually get 80% of the value by skimming most of it. I do grind through papers sometimes, and I do grind through a chapter sometimes, but in both cases only after a bit of consideration about what I'm likely to get out of it.

comment by Quinn (quinn-dougherty) · 2020-06-30T21:14:48.829Z · LW(p) · GW(p)

What granularity of time are you talking about? When you "never maintain 1 and 2 at the same time", is that any given minute, or any given decade?

I would say every couple months is an opportunity to either pivot or continue.


Comments sorted by top scores.

comment by mingyuan · 2020-06-30T18:13:29.594Z · LW(p) · GW(p)

What is your goal?

Replies from: mingyuan, quinn-dougherty
comment by mingyuan · 2020-06-30T18:14:29.826Z · LW(p) · GW(p)

Also, see Lynette's recent post [LW · GW] on prioritizing.

comment by Quinn (quinn-dougherty) · 2020-06-30T23:43:23.413Z · LW(p) · GW(p)

I'll check out Lynette's post.

I'd like to take a shot at technical AI alignment