In this post (the first post of skyspace4!) I reflect on at least 3 topics: (1) There are 3 types of shame: shame that causes you to take actions better aligned with your goals, shame that is maladaptive embarrassment of doing something right, and shame that causes you to pull away from others. (2) What fun pursuits do I value the most? (3) Is it actually good to try to achieve your goals?


Introduction

As many of you know, I’ve recently discontinued skyspace3.

Fortunately, I’m starting skyspace4.

The main difference is that I have an actual good site for viewing the 0stack: it has descriptions for the posts and AI generated thumbnails! I’ll also try to be better about not wasting reader time with incoherent notes that I accidentally put in 0stack.

  • Partially because of encouragement from readers. (It makes a big difference if you email me and let me know your thoughts on a blog post!)
  • Partially because I enjoyed TG’s most recent blog post.
  • Mostly because I think that reflecting on life and stuff is something that I enjoy and get value out of, and having a blog serves as an accountability mechanism of sorts for doing this and having relatively high quality thoughts.

A couple of notes about skyspace4:

  • I’ll try to post once or twice per month.
  • My vision for the posts is as follows:
    • Discuss concrete recent events in my life, or a question that that I’ve been thinking about lately.
    • Try to distill to a neat insight.
  • My hope is to keep the quality bar high.

ok, logistics out of the way, I’ll now brain dump some things that have been happening lately, and things that I’ve been thinking about. At the end I’ll hopefully distill to an insight.


Recent Events

The last couple months, especially June, have involved lots of change and decisions.

For instance,

  • I said goodbye to lots of friends in Boston.
  • I started working at ARC.
  • I met a bunch of cool people working on AI safety. Including many of my favorite bloggers such as Paul!
  • I graduated from college.
  • I quit ARC to intern at Redwood instead.
  • I made some new friends.
  • I’ve been calling and emailing friends some more.

I’ve learned some stuff about myself --- or more accurately was reminded of several things that I already kind of knew about myself:

  • I find living alone kind of lonely. This got somewhat better once I made some friends in Berkley, but in the future I’d like to prioritize finding roommates more.
  • Change and need to make decisions makes me pretty neurotic.
  • By default I have bad technology use habits when neurotic. But there exists software which mitigates ~90% of this problem.
    • Namely: blocking all apps on my phone except msging, phone calls, calendar and notes.
    • Having my computer automatically shut off at 9:30.
    • (Afaik, I can’t get around either of these restrictions, at least without a ridiculous amount of work).
    • (Note a slight confounding factor in the analysis here is that around the same time I put these measures in place, I also was getting a bit more adjusted to life in Berkley, which I’d also have expected to help out with this problem.)
  • Social skills
    • I think in some ways I have pretty good social skills.
      • Namely: reaching out to people, and organizing little events.
    • Probably the main place where I have bad social skills is that I don’t often join ā€œregularā€ activities (stuff with weekly large group meetings).
      • Notably this is partially bc I get a lot less out of those types of interactions, so it’s not even totally clear that this is a deficit. But this seems like a good way to get to know ppl.
  • Some longer points that don’t fit in a single bullet point to expand on below:
    • thoughts on ā€œshameā€
    • thoughts on ā€œfunā€
    • thoughts on ā€œconvictionā€

Shame

There are some situations, which occur every once and a while, where I feel ā€œshameā€ very strongly. I think this is partially a large strength of mine, and partially a large vice.

(Note: I’m not sure that shame is really the right word to describe this, and I am actually gesturing at a couple of distinct things here, but I think it does a reasonable job).

Here’s what helpful shame looks like for me:

  • Helpful shame is basically ā€œcognitive dissonanceā€.
  • There are about 4-6 situations I can remember where there’ve been fairly major actions that I’d been taking for a long period of time, that I’d decided where wrong.
  • In all of these cases the cognitive dissonance eventually caused me to change my actions to be in line with what I believe to be right.
    • I think this is good.
    • By which I mean something like ā€œI terminally value truth seeking + having my actions be aligned with the truthā€.
    • I don’t mean ā€œI’m happier than I would be in the counterfactual world where I ignored this factā€.
      • Bc this counterfactual does not exist for me.

Ok, but here’s what bad shame looks like:

  • Instead of thinking ā€œI’m doing bad thing X, I should stopā€ thinking ā€œI’m doing bad thing X, so I am badā€
  • Making me avoid other ppl --- out of fear that they’ll discover my shame
    • E.g., if I’m working on a research project, and it’s hitting dead ends,
    • I might avoid talking to ppl about the project
    • Because I feel like my failure to execute the project well indicates a lack of skill or smth
  • Just generally stressing me out.

For the 4-6 major situations where I had cognitive dissonance that was later resolved, the time that it took to resolve the dissonance was actually extremely long often. That is very unfortunate.

I think what often happens is I weigh all the ā€œsunk costsā€ or whatever that are holding me back from just doing what I know is right, and it’s just too heavy. And this kind of disables me for a bit. I only change my actions once there is literally no other way. And it seems like I often have to go to ridiculous measures to not feel like I’m taking scary actions. For instance, if I really need to send someone an email or something, but I’m anxious about sending the email, then I’ll schedule send it and forget about it, or else I couldn’t send it.

Well. What’s my advice to future Alek when he has this problem next?

  1. Rely on friends ASAP.
    1. For instance, if you feel strongly that you should do thing X but are scared or whatever, tell your friend ā€œI’m going to do thing X tomorrowā€ and have them bug you about it.
  2. If you are being anti-social bc of feeling bad about stuff, then just be social.
  3. Life is probably short, so prioritize resolving the dissonance ASAP. (Subject to the constraint of, if doing this too rapidly would cause larger problems, then resolve in whatever temporal schedule is best).
  4. Use weird hacks to force yourself to resolve the dissonance soon. E.g., send an email rather than talking in person if this is easier.
  5. Be kind to yourself.

It’s also worth noting that shame is just an emotion. It can be wrong. Often it is wrong. For instance, I might feel ashamed to ask a friend for a favor. As discussed elsewhere, this is dumb. But again shame is not really the right word for this feeling, and I think that this might be a genuinely different feeling from the other notion of shame here, such that I can actually distinguish between them.

Fun

I think it’s pretty healthy to do some things besides work, and I think this is important for achieving my values (not just instrumentally).

Here are the main ways that I’m trying to do this right now, and some thoughts on their merits.

  • Walking around:
    • My favorite way to do this is with one other person, and to have a good conversation.
      • I think this should be my go to way of inviting ppl to hang out:
        • ā€œHey do you want to go walk around locationX at 7ā€?
    • By myself, it can be a good for thinking, or for listening to some music and maybe running a bit to pretend like I exercise.
  • Calling friends:
    • I’ve been convinced that it’s more fun to use google meets for this than phone call.
  • Read textbooks and books:
    • I read some of Kolmogorov complexity the other day.
    • I found it very beautiful.
      • It reminds me of days spent hashing and working on allocation, which are the some of the most beautiful things that I know of.
    • I enjoy talking to ppl about cool math facts.
    • And trying to solve some of the problems.
    • I also read Tegmarks’ ā€œMathematical Universe Hypothesisā€ book recently.
      • I thought this was pretty fun.
      • Although I bet Feynman’s treatment of relativity and quantum is better. I plan to check that out at some point.
  • Play a game:
    • Again this is a pretty easy way to set up a social interaction, and I generally enjoy it.
    • One caveat is that it depends on the game.
    • The game has to be very interactive, have chatting as the main part of the game, or I won’t enjoy.
    • Charades is pretty good. As is Boticelli.
  • Writing:
    • I enjoy writing.
    • One of my favorite ways to write is to send ppl letters (ok emails).
      • Maybe partially because this is better at eliciting ppl sending me writing than blogging due to the bystander effect :P.
    • But also bc it adds a level of concreteness to the conversation.
      • I’m talking to personX about problemY, eventZ or emotionW.
    • In light of this observation, I might actually chose to alternate between blogging and sending individuals letters.
    • You could make an argument: but there are 5 ppl (NS, KZ, AW, MW, TG) that read your blog! So you’d have to believe that individual messaging is 5x higher value for the messaged individual than a blog post to think that this is more valuable.
      • However, this logic is flawed for several reasons.
      • (1) it’s totally fair game for me to just optimize for what I get out of the writing rather than what other ppl get out of reading.
      • (2) it actually seems totally plausible that me + conversation partner get >5x the value from a letter vs a blog post.
    • Anyways, I think there’s a time and place for both blogs and letters. I’ll try to do both.
    • One thing which I can’t emphasize enough is that I’d love to receive emails with questions or asking for advice as prompts to send you an individual blog post. Here are some fictional examples to illustrate what this looks like.
  • Shatar: Hi Alek, I was thinking about getting married to Blobby. Do you think this is a good idea?
  • Alek: Nope.
  • JJ: Hi Alek, I was thinking about quitting my job. Here’s some context: I don’t really buy the vision of the group I’m working with, and I think (but am not sure) that I could do good work with another group. What do you think?
  • Alek: Hi JJ, thanks for reaching out. I think that’s potentially a really great idea. Some advice I’d give is to talk to the other group you’re thinking about working with, get an actual offer on paper, and then actually read that offer, and then probably go for it if the offer seems reasonable. Let me know if you have any other questions, wishing you the best with this decision!
  • Blobby: Can I be the art on your new blog?
  • Alek: Hi Blobby, Thanks for your thoughtful question. I’ll have a chat with my content creator about that.
  • Blobby: yum!

Conviction

About a year ago, I decided that the goals I was pursuing were not ultimately the right goals. I adopted the following goals for ā€œhow I want to spend my lifeā€: (1) Making the universe a better place (2) Solving interesting and useful problems (3) Socially connecting with other people.

Q: How have I done at achieving these goals?

I think I’ve done moderately well.

  1. I haven’t done too much to make the universe better yet, but I’ve plausibly gained useful skills towards doing this, and have maneuvered into a state where it’s relatively easy to make the universe a better place.
    1. Probably the biggest progress here is work I’ve done on clarifying my goals here, thinking about what actions are likely helpful for achieving this goal, and strengthening my resolve to do this.
  2. I think my main problem solving was on the philosophy side.
    1. Overall I think I get a fairly low rating here.
  3. I had some good social connections. Def are ways I could improve here still ofc.

TG recently asked me ā€œdo you still endorse these goals?ā€

I still endorse (1) and (3). In particular goodness(universe) still seems like a basically fine definition of how good the universe is; not that I’d be comfortable optimizing hard against it ofc.

I’d like to revise (2) as follows:

  • 2.1 Do good work.
    • This is a process-based criteria, not outcomes-based criteria (outcomes are quite important, but they fall under ā€œmake the universe betterā€ instead of this category).
    • It means something like: being diligent, organized, and strategic.
    • Talking a lot with people about your work, having a vision for it.
    • Trying to find new ideas for good projects.
  • 2.2 Be intellectually curious.
    • This looks like learning random things, e.g., by reading textbooks. (If you have recommendations for well-written textbook recommendations, please send my way! I’ve enjoyed Kolmogorov complexity recently.)
    • Blogging about interesting topics. Searching for truth by distilling life experiences feels like a pretty neat type of intellectual pursuit.
    • Chatting with people about philosophy and trivia.

I think the main distinction from the previous (2) is that I’ve removed the need for work to be ā€œintellectually interestingā€. If your working is making the universe a better place, and you are doing good work, then it’s totally fine by my value system if the work is not intellectually interesting. Although in this case, I do hope you do some other intellectually interesting stuff! One of the smartest ppl I’ve met works for the government now, instead of doing crazy math/alignment stuff. It’s not intellectually interesting afaik. But I think that this is great, and want this kind of situation to be covered by my value system. More personally, I think that empirical ML research might sometimes be an engineering grind rather than ā€œsuper intellectually interestingā€, this seems like a pretty good thing to do, and I’d like my value system to endorse this.

Note: I suspect that many of my friends are strongly opposed to this change. Well, fight me.


Is it actually good to try to achieve your goals?

Recently a certain quantum physicist told me something to the following effect:

Procrastinating on making big life decisions is actually helpful in many situations. Some people make big life decisions too rapidly / impulsively, and making decisions is usually a higher variance deal, with possibly large downsides. The low variance choice of ā€œgoing with the life flowā€ of doing what is conventional has ā€œwisdom of the crowdā€ baked into it. Or at least these options are pretty well tested, unlike departures from this path that you might take. For an avg intelligence person, you probably shouldn’t be able to do very well in expectation by betting against conventional wisdom.

I almost hesitate to include the above excerpt, because I think that it’s the wrong advice for most readers of this blog.

But I think there is some truth to the sentiment. And it does resonate a bit with me.

More specifically, sometimes in the course of trying to actually pursue my goals in an approximately mathematically sane way I cry:

Isn’t this supposed to be easy?
I’m trying to do the freaking right thing, why do I have to deal with all this crap?

But, writing this sentiment down, it instantly becomes clear how silly it is, in many dimensions.

Why would it be easy? In fact, it’s natural that there are challenges.

Also looking at it from another angle, it’s ridiculous how much altruistic people have done to try to make things maximally easy.

Yes there will be challenges.

Some will be easy to handle, and I should def do so.

Sometimes, all I can do is handle challenges one step as a time, as they come up.


Ultimately, I’m extremely glad for the times when I’ve taken actions with conviction.

I stand by my resolution to work hard, seek knowledge, connect with others and to make the world a better place until the end.