Hammertime Intermission and Open Thread

post by alkjash · 2018-02-07T22:10:00.382Z · LW · GW · 14 comments

Contents

  Discussion Topics
    1. Sequences: Yea or Nay?
    2. Repeat or Explore?
    3. Monotonicity of Progress
None
14 comments

This post marks the end of the first cycle of Hammertime. Click here for intro.

Hammertime will return on Monday 2/19.

I want to close off the first cycle with some thoughts, and designate a place for discussion about the future of this sequence.

Discussion Topics

1. Sequences: Yea or Nay?

I’ve always felt that sequences are a valuable way to organize deeper thoughts and drive home a few central messages from several perspectives. However, the current format and culture on LW seem to radically favor short, independent chunks. (There is also the obvious problem that Sequence construction is not working.)

I’ve been posting daily for a while now but when I shifted from individual posts to a sequence, average Karma immediately dropped by a factor of about 2. It’s possible that people don’t bother upvoting the same sequence, or that my writing quality dropped, but if this is real signal that many more people would read a sequence if they are marketed as individual thoughts (and WordPress stats suggest this as well), that might be reason for me to stop writing sequences in the future, or at least collect sequences together only after they’re complete.

Possible actionable for meta: make posts in a Sequence share karma and/or a single slot on frontpage.

2. Repeat or Explore?

My original intention was to review 10 topics over three cycles, building up in the difficulty of problems solved. I think I will definitely return to and expand on several of the techniques we’ve seen already, but also add more topics. If people have favorite techniques (and hopefully references) they’d like to see in Hammertime, post them here.

3. Monotonicity of Progress

A big goal of mine is to solve the “Rationalist Uncanny Valley,” where beginning rationalists get worse at life before they get better. I can’t believe that this has to be the case; it seems to be symptomatic of a larger failure to develop the proper curriculum. I would like progress on rationality to be monotone – is there a good reason this should be difficult? It’d be great if we could compile a central list of “uncanny valley” failure modes.

14 comments

Comments sorted by top scores.

comment by Raemon · 2018-02-08T04:23:30.673Z · LW(p) · GW(p)

I was excited for the first Hammertime post... and then realized that dedicating a full month to following along seriously was in fact pretty intense and I wasn't up for it. I'm still excited for the sequence to exist overall, and that it'll continue to be valuable after it's done.

I think "once every 3 days" rather than once a day may be more practical, both as a writer and as readers - if I have a few days to catch up on a given post, I'm more likely to be able to fully invest the time into following along. (I actually had the same reaction to Conor Moreton's sequence).

Sorry the Sequence editor is currently broken, I have no idea what's up with that, will hopefully get to it sometime in the next week or so.

comment by swift_spiral · 2019-03-21T21:12:29.504Z · LW(p) · GW(p)

My thoughts after the first cycle: I found listing my bugs, Yoda Timers, and TAPs very helpful. The later days were interesting to read, but they did not help me solve many of my bugs (time calibration is a very useful skill, but I was already pretty good at it, at least for the short projects I can get quick feedback on). I think my largest benefits from reading this came from setting aside time to fix various things I had been procrastinating on, and motivating me to actually start doing things I had been planning to do for a while.

comment by Logan Riggs (elriggs) · 2018-02-08T04:47:36.486Z · LW(p) · GW(p)
  1. I've read through your series so far, and I don't believe your writing quality has dropped. Eliezer's inadeuqacy sequence went from 200 to 50 karma from beginning to end, and you'll see the same drop in views in youtube multi-part videos. I believe it's just barrier-to-entry with each additional post in a sequence because you have to read the other ones first. Posting individual posts and then compiling them as a sequence sounds like a good solution. Have you done a Yoda timer on this yet? lol
  2. I would like to see the dark side technique, which has been stated at Ziz' blog here and has a basis in Nate Soares' guilt series. Probably related to goal factoring and internal double crux just by the sound of those topics. If I was to summarize it, it'd be "Never do anything unless you know how it benefits you"
  3. What are your experiences of the "rationalist uncanny valley"? I would assume sunk cost fallacy fallcy you mentioned, but is there anything else? For me personally, it would be "expending too much social capital for truth's sake" and the above dark side technique. Both of these came from taking those ideas (Truth and Dark Side) seriously, actually trying them in real life, and overdoing it in wrong ways. I did learn from those experiences and am better for it, so trying, failing, learning, repeating was overall beneficial. I assume that's what you would call the uncanny valley?

If so, to improve it would be to improve that feed-back cycle. Anything that increases trying, minimizes failing, and provides better feedback is a possible research avenue. From your own series (and a couple extra):

Increasing trying: Yoda Timers, TAPs, Aversion Factorying, Design, Dark Side

Minimizing failing/error: general biases, Actually Trying, Murphyjitsu,

Better Feedback: Bug Hunt, CoZe, Time calibration, Focusing

comment by Swerve · 2018-02-08T06:45:34.448Z · LW(p) · GW(p)

My gut reaction told me that I hadn't gone through the experience of the uncanny valley. After a minute of thought, I realized that there was a period of four months last year where I divebombed in grades and class engagement. It was contextually because I had just learned about the diaspora and was rabbit-holing everything related to the disparate rationality blog. I was also rereading HPMoR. I think the historic reason for my failure was two-fold. The realization that the community wasn't dead and I wasn't alone; and the overemphasis the sequences and HPMoR had placed on epistemic rationality. I certainly learned how to have more accurate beliefs about the world, but I didn't realize there was a cost to optimizing. Come to think of it I didn't even conceptualize optimizing as a thing people could actually do in the real world. How scary.

I've got to agree with your sentiment about sequences and engagement. I've found I have a reluctance to read something that is explicitly a sequence. You may consider uniquely labeling the rest of the hammertime posts and only compiling them afterwards.

Meta:

I have a question. Er, not so much a single question as a concept with a strong sense of confusion attached to it. What made you decide on the tools you decided to cover? Do you believe all of these techniques are low-hanging, or are they more advanced? What would other advanced techniques look like? I know that other people have attempted such things in the past (with varying degrees of success). I remember reading somewhere (unfortunately I can't recall the source nor the exact quote) that three or four people on that particular thread had strongly considered creating a primer instrumental sequence.

The main issue was something along the lines of 'who am I writing this for?' Anyone reading the sequence would most likely already be famailiar with the material (a la someone on LW) and thus not really need the tools and advice (on account of the information being fairly wide-spread and easily accessable in the rationalsphere). On the other hand anyone who could really use the information is probably not going to be the kind of person reading this material in the first place.

Important note! This is not a criticism of your efforts here. This has been a concern of mine for couple of months and it's extremely gratifying to see someone trying to fix the problem I mentioned in the first paragraph. I suspect communicating the importance of both sides of the rational coin will be a productive way to break part of the uncanny valley. I'm extremely happy that we have more people actually engaging in rationality practice, and generating usable content, and I applaud you for your efforts. The above is just my feeble attempt at asking what direction the steepest gradient for rationality (instrumental or otherwise) is current in. I have a sense that more is possible, but where?

Respond to any of the prompts, or none of them. They're not the point. They're just there to illustrate the point.

Replies from: alkjash
comment by alkjash · 2018-02-08T15:28:59.513Z · LW(p) · GW(p)

A bunch of helpful comments on this thread already, I'll collect my all my thoughts so far here, in no particular order:

I know people have done/thought about writing such a thing in the past, but I approached this sequence in a completely different way. It was my New Year's Resolution to write productively every day for a year, as this seemed like the single most easily achievable and powerful mind level up I could attain. I made a bunch of progress in January but felt my thoughts going all over the place. Hammertime was the product of me sitting down for half an hour and figuring out a simple outline for writing with some system and regularity. Because I was already writing every day, the actual writing part was completely cost-free to me, whereas it sounds like possibly the major roadblocks in previous attempts.

My reasons for writing this sequence were, in clear order of importance: (a) to practice writing, (b) to review CFAR techniques for my own benefit, (c) to entertain, and (d) to teach instrumental rationality. With regards to (c), I have a belief that once you have decided on a thought to write about, the primary focus of the writing process is to make it as entertaining to read as possible, and this should come before making it useful.

The main issue was something along the lines of 'who am I writing this for?' Anyone reading the sequence would most likely already be famailiar with the material (a la someone on LW) and thus not really need the tools and advice (on account of the information being fairly wide-spread and easily accessable in the rationalsphere). On the other hand anyone who could really use the information is probably not going to be the kind of person reading this material in the first place.

I strongly disagree with this statement. I've been reading LW for about five years and "knew" about most of these tools abstractly without ever getting anything practical out of it. You can check the comments to Hammers and Nails and extrapolate that even longtime LWers have each only practiced only 5% of all the techniques we have, and that 5% varies wildly from person to person. It's not clear if what I've written so far actually helps in this direction, but I think a properly written sequence will actually inject readers with the moral fire to do the thing that they've known about for years.

This is part of the reason (the other reason being laziness) that I've eschewed science and citations - I think most LWers can produce on demand the psychological science to back up most of my techniques and claims, and adding them is low value. Some of this work has been filled in in the comments.

On the uncanny valley, I think what you're referring to is something different and benign: unavoidable opportunity cost, not so much practicing rationality causally making you worse at life. To me the uncanny valley refers to about three phenomena (that have all happened to me):

  1. People becoming hyperaware of status games and deciding to play them consciously in a calculated, off-putting way. Anecdotally this seems to be a relatively common failure mode for students who go through SPARC. This is also my read on a good deal of unproductive LW conversation. Hint: high status isn't always better!
  2. People learning about biases and deciding to solve them by pasting a post-processing fix on top(e.g. double all time estimates to fix planning fallacy), or making all decisions by System 2 because they no longer trust System 1. This is sort of classic "Straw Vulcan." One of the big failures of the Sequences, imo, is not detailing the important and serious reasons that many biases can actually be quite locally useful, even if they cause problems in areas they're not designed for. We need more Chesterton's Fence.
  3. Rationalists beginning to see non-rationalists as normies, NPCs, or otherwise sub-human. It's not uncommon to see conversations and word choice on LW like: using "aliveness" or "more human" as largely synonymous to being better at rationality, saying in perfect seriousness "it's impossible to talk to normies because they have no concept of objective truth," or putting oneself in the tiny set of unlikely heros who are "actually trying."
Replies from: Swerve
comment by Swerve · 2018-02-10T00:26:22.193Z · LW(p) · GW(p)

After thinking about your reply for a while, you've made me update strongly towards believing that I had overestimated my own efficacy. In particular:

I've been reading LW for about five years and "knew" about most of these tools abstractly without ever getting anything practical out of it. You can check the comments to Hammers and Nailsand extrapolate that even longtime LWers have each only practiced only 5% of all the techniques we have, and that 5% varies wildly from person to person. It's not clear if what I've written so far actually helps in this direction, but I think a properly written sequence will actually inject readers with the moral fire to do the thing that they've known about for years.

This struck me as I hadn't considered that I was missing so many tools. I feel like I've had my life improved a lot by rationality. I can cite many many cases where things get resolved specifically because I have the proper information and training. Yet even with everything I currently do, I now realize that I don't impliment the vast majority of useful things we as a community have managed to come up with. If one of your primary goals with this series it to imbue people with some of that "moral fire", then consider it a success, at least regarding my personal experience. One thing to note is that while the series itself sets the context for this sort of thing to happen. It was only the meta-level conversation, and that specific information about inadequacy that helped me viscerally update. That might be series-relevant information.

Anyway, thanks for helping me fix my models and pointing out a glaring blind-spot.

Replies from: alkjash
comment by alkjash · 2018-02-11T00:15:52.122Z · LW(p) · GW(p)

At the beginning of this project I said to myself that I would be happy if it moved one other person in a signficant way. Right now, I'm very happy =).

I'll try to inject more of this moral fire into the coming cycles. One thing that I've come to understand is that people really don't aim high enough. There's a mindset where you try to self-improve until you reach a satisfactory level, something like 90th percentile among the peer group, and relax there. There's an alternative mindset where you believe that the better you are, the quicker you will improve by learning new tools, since each of them is a force multiplier. If you follow this mindset far enough, it's almost dizzying.

comment by Qiaochu_Yuan · 2018-02-08T08:10:38.771Z · LW(p) · GW(p)

I think sequences are good and will get much better once the sequence thingy is up and running. I think most of their value doesn't become apparent until further in the future, when someone stumbles upon one of the posts and now knows that there are a bunch of other related posts they could read.

I think I'm more excited to see what you do with repeating than with exploring.

I haven't seen the rationalist uncanny valley be a big problem in practice. To the extent that it is, I have several possible mechanisms in mind, but I'm not sure which kinds of cases you're thinking about. Maybe the worst thing I personally did was tell a friend of mine a bunch of un-PC thoughts I'd been having about feminism, with the result being that she stopped talking to me for several months.

Replies from: flipflopchip
comment by flipflopchip · 2018-02-08T18:32:33.590Z · LW(p) · GW(p)

I had something similar happen to me (with a different but equally inflammatory topic).

I didn't say anything I don't believe, but I think I could have said it in a more diplimatic way.

Is that the case with you too?

Replies from: Qiaochu_Yuan
comment by Qiaochu_Yuan · 2018-02-08T21:42:17.855Z · LW(p) · GW(p)

My story about what happened is more that I was saying things that were very triggering to my friend and that I wasn't modeling that in sufficient detail; she even said something like "I'm happy that you're trying to think through this and come to your own conclusions, but you just can't do it around me." So I stopped.

comment by Richard Meadows (richard-meadows-1) · 2018-11-10T15:23:26.908Z · LW(p) · GW(p)

Progress report at the end of cycle one: So far I've taken three days on average to think or act on each post. I don't think this was a bad thing, and was probably better than attempting a frenetic daily pace.

I've resolved 10 of ~50 bugs, with progress underway towards another 10. That's roughly an even split between TAPs, Yoda timers, and bugs that resolved themselves with no action required. I'm most pleased with dramatically cutting down on smoking - it wasn't actually difficult; I just needed the impetus of a lovely spreadsheet with conditional formatting. Looking forward to cycle two.

comment by flipflopchip · 2018-02-08T18:35:53.559Z · LW(p) · GW(p)

Writing something every day seems incredible to me, I often don't have the energy to read every day let alone write.

Are there any techniques you (or others) use to boost your mental energy?

Replies from: alkjash, Qiaochu_Yuan
comment by alkjash · 2018-02-08T19:35:07.123Z · LW(p) · GW(p)

I am an endless repository of writing advice, mostly superstition and a bit of which I've written down. The most useful thing I've read recently is this excerpt from Pirsig's Zen and the Art of Motorcycle Repair:

Phaedrus found that rhetoric at the University level was taught as a branch of reason alone. He was also having trouble with students who had nothing to say. Especially one girl, who was a serious, disciplined, and hardworking student. She wanted to write an essay about the United States. Phaedrus told her to narrow it to Bozeman but she couldn’t think of anything to say. Phaedrus told her to narrow it down to the main street of Bozeman. Still nothing. He then said “Narrow it down to the front of one building on the main street of Bozeman. The Opera House. Start with the upper left-hand brick.”The next day she returned with a 5,000 word essay on the front of the Opera House on the main street of Bozeman, Montana.
We get blocked from our own creativity because we just repeat what we have already heard. Until we really look at things and see them freshly for ourselves, we will have nothing new to say. “For every fact there is an infinity of hypotheses. The more you look the more you see.”

The difference between "trying" and "doing" is something I touched on in Day 2: Yoda Timers. I think there's a closely related difference between "trying to write" and "writing." Namely: "trying to write" feels like looking for things you already know, whereas "writing" feels like looking at with fresh eyes. This is something a high school art teacher taught me as well, about drawing realism. If you want to write more, pick something and examine it in as high resolution as possible. The amazing thing about reality is its endless, almost unnerving detail.

comment by Qiaochu_Yuan · 2018-02-08T21:37:57.882Z · LW(p) · GW(p)

"Mental energy" is a gigantic rabbit hole; you've asked a much deeper question than you think. I would set the frame less as "how do I boost my mental energy" and more as "what kind of crazy crap is sapping my mental energy," which suggests quite different sorts of ideas to try. I'm also quite worried at the thought of having so little energy you can't even read.

For starters, take a good look at your health in a broad sense: how's your diet? Are you sleeping enough? Getting enough exercise? Drinking enough water? Getting even the basics right here can make an enormous difference to your quality of life.

Beyond those basics there's mental / emotional / social health: do you have a strong support network? Are there people you can rely on for emotional support? Do you touch and get touched by others regularly?

Unfortunately most of the really helpful interventions here are really hard to do if you're starting out in a bad place. The fastest way to get out is probably to have someone else do it: you can swap "debug days" with someone else where someone spends an entire day trying to figure out with you your biggest bottlenecks and then tries to help you resolve them, and then vice versa.

Edit: There are also "tricks" that people sometimes discuss, such as splashing water on your face and so forth. They probably wouldn't hurt, especially the ones that are more health-oriented like drinking water and going for walks, but both my experience and my models suggest that most of them will not help in the long run.