[Talk] Paul Christiano on his alignment taxonomy 2019-09-27T18:37:31.475Z
Are we living at the most influential time in history? 2019-09-07T18:09:00.411Z
jp's Shortform 2019-08-29T16:18:07.727Z
What is the evidence for productivity benefits of weightlifting? 2019-06-02T19:17:35.883Z


Comment by jp on jp's Shortform · 2020-11-12T11:31:15.660Z · LW · GW

This video seems highly relevant to the luminator fans here.

Comment by jp on jp's Shortform · 2020-10-21T07:47:52.701Z · LW · GW

I hadn't before actually, thanks for the recommendation. I checked it out. Talon seems more like "teach yourself the alphabet, then program using vim." Whereas Serenade is more like "teach yourself to program using our macros." This makes Talon easier to learn, and more flexible, Serenade has the following benefits as I see them: 

  • Programming in large well defines chunks, which it makes it easier to incorporate "pick which one you meant, or continue to use our default"
  • It uses more natural language, which makes me think that the accuracy has a higher ceiling
Comment by jp on jp's Shortform · 2020-10-11T17:40:30.896Z · LW · GW

What to do if you suddenly need to rest your hands

On Monday I went from "computer work seems kind of uncomfortable, I wonder if I should be worried" to "oh crap oh crap, that's actually painful". Everything I've ever heard says not to work through RSI pain, so what now? I decided to spend a week learning hands free input. I wanted to a) get some serious rest and b) still be productive. And guess what? Learning hands free input is like the one activity that does not suffer a productivity penalty from not being able to use your hands.

When I started it was really slow going. Lots of yelling "no don't type 'delete word' delete the f**king word!!!"[1] But then I found this talk, which was just .. woah. Since then, I have been using Talon, and I am in love.

The key to understanding Talon, and the reason I think it's heads and shoulders above everything else I've tried, is the basic insight that most of the time the input you want to do is not writing. Talon has the concept of modes, and most of the time you're in the command mode.[2] And because there's a limited set of commands, there's much less ambiguity between inputs.

After one week, I'm able to dictate an existing code file painlessly albeit still slowly. The thing that feels important to me, is that I'm no longer living in fear of my career being taken away from me by my wrists. 

So my recommendation to you, if you find yourself in the situation I was is to rest your hands, and try learning this thing. If you do, reach out to me! I'm (at least currently) sufficiently excited about this that I would very much enjoy help you out.

Epistemic postscript: I'm writing this while still excited about it, which is providing the motivation to do the writeup, but also makes me believe that in the future I will be less optimistic, and you should take that into account when evaluating its implicit predictions.


[1] A word to the wise, if you have dictation software listening, yelling at your computer is the opposite of productive (this is good, a frustrating experience with quick negative reinforcement for outbursts induces a zen-like experience).

[2] This is similar to vim, if you're familiar.

Comment by jp on Habryka's Shortform Feed · 2020-06-15T19:00:34.426Z · LW · GW

I had forgotten this post, reread it and still think it's one of the better things of it's length I've read recently.

Comment by jp on Problem relaxation as a tactic · 2020-04-26T16:01:17.465Z · LW · GW

Why is this comment bold?

Comment by jp on Announcement: LessWrong Coronavirus Links Database 2.0 · 2020-03-26T05:31:36.767Z · LW · GW

Another suggestion: could this post be linked to by the link database page?

Comment by jp on Announcement: LessWrong Coronavirus Links Database 2.0 · 2020-03-26T05:30:43.751Z · LW · GW

Suggestion: a forecasting section. Already some of your "Other" links would do well there.

Comment by jp on March Coronavirus Open Thread · 2020-03-14T20:35:28.763Z · LW · GW

Does anyone want to venture a guess for the true number of cases in the Bay Area right now?

I just some rough back-of-the-envelope calculations, following the method here. Currently there are 2 deaths in the Bay Area. I keep his time-till-death rate of 17.3 days, but substitute a doubling time of 3 days, based on Our World in Data's US number. I get 2*100*exp(17.3*.231) ≈ 11,000, or about 2/1000 bay area residents. Super non-robust number, take with several grains of salt.

(.231 is the rate I get when solving for r in the exponential growth function for a doubling in 3 days.)

Comment by jp on March Coronavirus Open Thread · 2020-03-13T00:10:16.517Z · LW · GW

Why not ask this as a question post?

Comment by jp on Raemon's Shortform · 2019-11-15T19:17:33.505Z · LW · GW

I intuitively think it's good, but have in fact noticed myself clicking to dismiss it despite not having read it or thought about whether I'd like to read it.

Comment by jp on Honoring Petrov Day on LessWrong, in 2019 · 2019-09-27T03:03:05.538Z · LW · GW

Rohin argues elsewhere for taking a vote (at least in principal). If 50% vote in favor, then he has successfully avoided "falling into the unilateralist's curse" and has gotten $1.6k for AMF. He even has some bonus for "solved the unilateralist's curse in a way that's not just "sit on his hands". Now, it's probably worth subtracting points for "the LW team asked them not to blow up the site and the community decided to anyway." But I'd consider it fair play.

Comment by jp on Honoring Petrov Day on LessWrong, in 2019 · 2019-09-27T00:31:55.244Z · LW · GW

It could partially motivated by lifesaving but they wouldn't have donated otherwise. Like, not if they're a perfectly rational agent, but hey.

Comment by jp on Honoring Petrov Day on LessWrong, in 2019 · 2019-09-27T00:25:01.358Z · LW · GW

Comment by jp on Focus · 2019-09-17T21:02:50.279Z · LW · GW

I feel like I'd rather live in the world where I could leave slack unblocked and still focus on my codebase that takes 25 seconds to reload on file save.

Comment by jp on Focus · 2019-09-17T20:55:11.613Z · LW · GW

I felt like this was a scarily accurate description of my focus situation. And then I read the part about the chocolate, and like, good lord, get out of my head.

Comment by jp on Ruby's Public Drafts & Working Notes · 2019-09-14T00:31:10.718Z · LW · GW

I'm a complete newcomer to information on Bacon and his time. How much of his influence was due to Novum Organum itself vs other things he did? If significantly the latter, what were those things? Feel free to tell me to Google that.

Comment by jp on Matthew Barnett's Shortform · 2019-09-13T16:14:15.462Z · LW · GW

Sometimes you need someone to give the naive view, but doing so hurts the reputation of the person stating it.

For example suppose X is the naive view and Y is a more sophisticated view of the same subject. For sake of argument suppose X is correct and contradicts Y.

Given 6 people, maybe 1 of them starts off believing Y. 2 people are uncertain, and 3 people think X. In the world where people have their usernames attached. The 3 people who believe X now have a coordination problem. They each face a local disincentive to state the case for X, although they definitely want _someone_ to say it. The equilibrium here is that no one makes the case for X and the two uncertain people get persuaded to view Y.

However if someone is anonymous and doesn't care that much about their reputation, they may just go ahead and state the case for X, providing much better information to the undecided people.

This makes me happy there are some smart people posting under pseudonyms. I claim it is a positive factor for the epistemics of LessWrong.

Comment by jp on jp's Shortform · 2019-09-12T04:04:55.667Z · LW · GW

I'm really glad to have this comment! It seems much more valuable to know that something passes a first attempt by a second party than to just hear a recommendation from one person's experience.

Comment by jp on Is competition good? · 2019-09-12T01:07:42.133Z · LW · GW

FWIW I dramatically misinterpreted what the "people" disagreed with and did not think "AMF is better than a restaurant" was the claim that would be contested.

Comment by jp on jp's Shortform · 2019-09-10T15:12:01.281Z · LW · GW

Do Anki while Weightlifting

Many rationalists appear to be interested in weightlifting. I certainly have enjoyed having a gym habit. I have a recommendation for those who do:

Try studying Anki cards while resting between weightlifting sets.

The upside is high. Building the habit of studying Anki cards is hard, and if doing it at the gym causes it to stick, you can now remember things by choice not chance.

And the cost is pretty low. I rest for 90 seconds between sets, and do about 20 sets when I go to the gym. Assuming I get a minute in once the overheads are accounted for, that gives me 20 minutes of studying. I go through about 4 cards per minute, so I could do 80 cards per visit to the gym. In practice I spend only ~5 minutes studying per visit, because I don't have that many cards.

I'm not too tired to concentrate. In fact, the adrenaline high makes me happy to have something mentally active to do. Probably because of this, it doesn't at all decrease my desire to go to the gym.

I find I can add simple cards to my Anki deck at the gym, although the mobile app does make it slow.

Give it a try! It's cheap to experiment and the value of a positive result is high.

Comment by jp on jp's Shortform · 2019-09-10T15:11:27.282Z · LW · GW

Part II

The goal I ended up going with was two edited posts this week, due Sunday.



Comment by jp on How Much is Your Time Worth? · 2019-09-03T05:07:53.098Z · LW · GW

She mentioned Philip was a client. He's literally paying to be other-optimized. Also, she's citing enough evidence to get around the typical problem of a failure to generalize.

Comment by jp on jp's Shortform · 2019-09-03T04:08:04.009Z · LW · GW

Bruce Schneier's original security mindset blogpost basically just says "look for holes in things." I thought I remembered the concept as being more interesting when I read it on LW and sure enough, Eliezer's post was much more cogent. "The reason security is hard is because there's someone optimizing the system down paths that lead to bad outcomes."

Comment by jp on jp's Shortform · 2019-08-31T15:21:32.053Z · LW · GW

YA Novels and Human Talent Distributions

I take a dim view of how I spent my free time as a teenager. Reverse to how many people see it, I think my school time was great for me and my intellectual development, while my spare time often made me a worse thinker.* In particular, I'll call out my habit of videogames and YA fantasy novels. Here's a thing I wish I hadn't learned.

In YA novels, if you've ever spent 10 minutes living in the woods, you're now an A+ expert on all  things forestry. It doesn't matter if you're up against an adversary who logically would have spent years training for this, don't worry, if a single person on your team has some plausibly related piece of backstory, you're going to have an advantage.

Additionally, your primary talent is probably something where you have a god-given advantage over the rest of the world.

So fantasy novels are unrealistic. I noticed this while reading them. I still think I'd rather read books that will leave my system 1 with a more accurate understanding of talents. But what I noticed recently was that I didn't quite appreciate that these novels (and books) had discontinuities of talent. Many talents are power law distributed, to be sure, but more commonly they are normally distributed.

I've noticed myself appreciating that I/my friend/coworker/acquaintance are good at something, and then it taking a while to realize how not-special their talent is, to the detriment of my predictions about the world.


Another anti-useful learning: I spent years training my intuitive appreciation for how often a 90% accurate attack will miss on game THAT LIED ABOUT IT'S ACCURACY.

* I think I still am my best self doing productive things and often my spare time is spent unproductively.

Comment by jp on jp's Shortform · 2019-08-31T15:04:02.605Z · LW · GW

For this purpose there are two related dynamics. How much activation energy it takes to start, and how much energy it takes to leave (usually inversely related). Like an object in a potential energy landscape, being in a low energy state makes it harder to move to a high energy state. I agree there's a surprising lack of correlation between low energy states and relaxing states. Meditating is a clear example of a high energy state, but it is pretty restorative. I don't find binge watching restorative after maybe the first episode or so, but I do find reading a blog for 10 minutes to be so*.

*Possibly because by "blog" you're thinking "intellectual blog like SSC", and I'm talking about what's half the time a tumblr blog.

Comment by jp on jp's Shortform · 2019-08-31T14:46:28.416Z · LW · GW

For sure. This is actually something that made me curious about the internet writ large. If the WWW was just one thing built on the internet, then why do I get mostly answers for the WWW when I look up history of the internet? (Because the WWW was so useful.) Why did the internet already exist when the WWW came along? (Because it was just the phone lines and some modems.)

Comment by jp on jp's Shortform · 2019-08-30T18:20:48.824Z · LW · GW

Some thoughts on free time

I've had discussions recently about how to spend free time. I'm blessed with a job with relatively well-specified boundaries (I don't typically work on weekends or outside the hours I'm at the office), but I often still feel like I'm sucked into "unproductive" things in my free time. Here are some things I could want to do during my free time:

 1 Maximize global utility directly

 2 Maximize moment-to-moment hedonic happiness

 3 Maximize long term hedonic happiness

 4 Maximize mental recovery for later productivity

 5 Use a virtue heuristic for doing things that seems "worthwhile"

 6 Do whatever one feels like in the moment

 7 Try to accomplish things that sound cool

(1) Seems penny-wise, pound foolish, and paradoxically hurt my altruistic efforts. All of the ones with maximize I endorse caring about to some extent, but not maximizing. I'm especially interested in 4. I feel like 5 has led to some great victories for me. 6 obviously guides a lot of what I do. I'm happy that I do it, but I don't want to elevate it the way some people do. I like doing 7 sometimes, but often it trades of against 1, 2, 4, and 6, in which case I mostly end up not doing it. Thus I think most of my accomplishments have come during work hours. I'm basically ok with this.

Sometimes 6 ends up tanking efforts to do 1-5,7. Here's a list of some things that I don't endorse doing:

 1 Binge watching netflix, youtube etc.

 2 Scrolling through facebook, twitter, etc.

 3 Playing a videogame that grows to consume my time

 4 Writing code < 1.5 hours before I'd like to go to bed

 5 ~Half of the times I stay up late at parties

I claim that these are usually bad for essentially all of the other goals.

Here are some things that I endorse doing:

 1 Visiting individual blogs are reading through what I've missed

 2 Churning my personal Evernote todos

 3 Trying to answer something I'm curious about

 4 Hanging out with friends/boyfriend

There's a gap here where the unendorsed list has a bunch of things I can do even when I'm exhausted, and the endorsed list usually requires at least a little bit of awake-ness. This often makes me very reluctant to take stimulant holidays. The problem I think is that zero effort things are very low points in the energy landscape. This is what makes individual blogs so useful. They're pretty low energy, but they run out of content quickly.

I'd be interested in other recommendations for low energy but finite activities and additions to the goals list.

Comment by jp on jp's Shortform · 2019-08-30T18:16:17.026Z · LW · GW

I'd really like to write more. I've noticed that some ideas become much better after I write them up, and some turn out to be worse than I initially thought. I'd also like to expand my ability to have conversations to include online spaces, which, as a confirmed lurker, I didn't really have much of until after I found myself writing code for the EA Forum. I'm going to try writing a shortform post a day for a week. Acceptable places to post include here on LW, the EA Forum, Facebook, and my org's slack. I'd like to go for at least one each.

If that goes well, my next step might be to try this thing called editing and post every other day. After that I'd like to try writing some top level posts.



Sunday: ✓ (FB)


Tuesday: ✓ (Slack)

Wednesday: ✓ (Slack)


Comment by jp on jp's Shortform · 2019-08-29T16:18:07.882Z · LW · GW

Here's a conceptual clarification that deconfused me somewhat. – The zeroth instantiation of the internet was the telephone network. The alpha version was when someone first made a modem to let computers send 1s and 0s over the telephone system. And that was it. You had the ability for computers to talk to other ones in different cities.


A different 0th version was computers hooked up directly to other computers.


The beta version involved Usenet, emails, etc.


Finally we get to what people talk about when they say the "invention of the internet" and a web of pages that pointed you to other pages when you clicked on links. I think the reason this gets so much attention is because the previous versions were just immensely less powerful.


There's so much more about this I wish I knew.

Comment by jp on Cognitive Benefits of Exercise · 2019-08-16T00:42:58.434Z · LW · GW

My model is something like: "these effects dance around the edge of significance and have a bunch of uncontrolled variables that differ between studies, so it shouldn't be surprising that the results aren't consistent." Now, that might cause you lower your credence in the whole endeavor, which in fact I do.

Comment by jp on Cognitive Benefits of Exercise · 2019-08-14T22:17:23.333Z · LW · GW

See some related work here:

I wonder if hereisonehand would be interested in a repeat performance.

Comment by jp on Why do humans not have built-in neural i/o channels? · 2019-08-14T18:12:34.279Z · LW · GW

You can ask a related question with the LW UI.

Comment by jp on jacobjacob's Shortform Feed · 2019-08-14T04:14:12.824Z · LW · GW

The Sequences could use a refresher possibly.

Deep Work was my immediate thought after Elizabeth's spot check.

Comment by jp on Do you do weekly or daily reviews? What are they like? · 2019-08-08T20:19:01.389Z · LW · GW

Here's the terrible hack of a bash command (not script actually)

pbpaste | perl -pe 's/\n/%0A/m' | (read pastebuff; curl -X POST -H 'Content-type: application/json' --data "{\"content\": \"$pastebuff\"}"
Comment by jp on Do you do weekly or daily reviews? What are they like? · 2019-08-07T15:27:17.428Z · LW · GW

1) Evernote quick entry on my mac

⌘⌃N <text> ⌘⏎

2) Evernote quick entry on my phone

Theres a setting to keep it it the notifications menu. Then I can just pull it down click the new note card, type text and hit save.

3) Ok Google tell Evernote ___

IFTTT integration

4) My Livescribe notebook which syncs to Evernote

My pen and physical paper are digitized and synced with evernote. I then usually create regular notes from the notebook pages during my personal review.

5) Aquanotes in my shower which I manually add to Evernote

Waterproof notepad

6) Keyboard shortcut that pipes my mac's system clipboard to Asana

Better Touch Tool shortcut to run a bash command which hits a Zapier webhook API. I write the text, cut it, and hit ⌘⌃A

7) Ok Google tell Asana ___

Ditto above

Comment by jp on Do you do weekly or daily reviews? What are they like? · 2019-08-05T02:24:52.991Z · LW · GW

I have two task inboxes, one for work and one for personal. At the end of the day for work and sporadically for my personal life, I triage the incoming tasks*. For work this involves assigning between:

  • Asana upcoming for smalls or the “Possible Sprint Goals” project for larger tasks
  • The correct project backlog, tagged appropriately
  • Doing any that are less than 2 minutes

For personal this is a pretty quick:

  • Do now if < 2 minutes
  • Put in Personal Tasks, possibly tagged with a date and category

I do the work ones at the end of every day. I have some tasks that rely on me doing them at the end of the day that are valuable (I block slack during parts of the day, so sending messages happens at end of day), which keeps me motivated to do the review. It takes 10-40 minutes depending on how many tasks I generated. Most of the time is spent doing the < 2 minute tasks.

I select work tasks to work on at the start of every day from my “Upcoming” section and put them into the “Today” section of Asana.

There are several good environments for me to do personal triage. When I’m resting in between weightlifting sets at the gym, when I’m on public transit, and when I’m taking rideshare. (Aside: these are also good places to do Anki.) Combined, they’re more than enough to finish the triage. I’d say it averages about 5 minutes per day currently, though in the past it was more.

I work on Personal tasks on the weekends, often at a cafe.

* Tasks can range anywhere from “order measuring spoons on amazon”, or “figure out how recommendations should work on the Forum”. I have a bunch of low friction systems for taking notes which I can expand on if people are interested.

Comment by jp on What is the evidence for productivity benefits of weightlifting? · 2019-06-16T22:43:51.824Z · LW · GW

This comment wins. This is beyond what I was hoping I would get and I'm really glad I asked. Thanks hereisonehand! Message me with paypayl/venmo/etc info to claim your prize.

Comment by jp on What is the evidence for productivity benefits of weightlifting? · 2019-06-13T16:40:56.731Z · LW · GW

Really nice job. I think I'm persuaded by your conclusions, and I intend to try to mix in more aerobic exercise. How did you do the search?

Comment by jp on What is the evidence for productivity benefits of weightlifting? · 2019-06-13T15:50:11.601Z · LW · GW

Reminder that the prize will be awarded this Sunday the 16th.

Comment by jp on What is the evidence for productivity benefits of weightlifting? · 2019-06-03T03:03:34.305Z · LW · GW

That much closer to what I was hoping to find than I found initially, and I'd say it passes the initial bar. A few knocks against it are the focus on the elderly (the mediate age was 70) and the high variability in the outcomes. I'm not familiar enough to know what to expect, but those plots look all over the place to me.

Comment by jp on What is the evidence for productivity benefits of weightlifting? · 2019-06-03T02:21:41.695Z · LW · GW

Great title.

Comment by jp on What is the evidence for productivity benefits of weightlifting? · 2019-06-03T02:19:04.154Z · LW · GW

Good question. I think two weeks is reasonable. I can extend it if someone messages me saying they're still working.

Comment by jp on Data Analysis of LW: Activity Levels + Age Distribution of User Accounts · 2019-05-15T06:03:56.257Z · LW · GW

Paul Graham is such a fun read, but when I have my skeptical hat on I don't find myself convinced. What's the mechanism of action? If LW doesn't die it will eventually achieve its aims because .. ?

I do like the lens of product-market fit, and I tentatively agree LW doesn't really have it. I guess you could say that you have successfully avoided dying and now you get to continuing swinging at ideas until you hit a home run.

Comment by jp on Data Analysis of LW: Activity Levels + Age Distribution of User Accounts · 2019-05-15T05:41:52.929Z · LW · GW


The hypothesis is: the value of the site technical functionality to the average user is mostly the posts, comments and votes. That's already built so the work on the margin hasn't increased the value that much. The real room for variation is the value of the content (writing) on the site. The value of that content is modest (not huge) and static (not growing).

Comment by jp on Data Analysis of LW: Activity Levels + Age Distribution of User Accounts · 2019-05-15T02:13:51.041Z · LW · GW

Here's a pessimistic takeaway. One should expect a site that was providing a lot of value to its users to grow, even if it wasn't explicitly trying to.

Perhaps most of the value of the site is in the fact that it has posts, comments and votes. Beyond that it's the value of the content, and that is modest and static.

View is lightly held, and I'm not drawing much in the way of conclusions from it. I am still impressed with the feature roadmap and the velocity of improvement.

Comment by jp on What makes people intellectually active? · 2019-01-12T20:40:39.625Z · LW · GW

I've rigged google assistant to be easy to speak out notes to for situations where I don't have use of my hands.

Comment by jp on Moving Factward · 2018-11-30T03:27:06.187Z · LW · GW

😛 Dangit, my American intuition wasn't good enough. I was kinda close, only a factor of 2 maybe.

Comment by jp on Moving Factward · 2018-11-29T23:37:44.782Z · LW · GW

It seems plausible that it's correct to treat a piece of information as approaching very close the fact-ness. If I say that a cup contains an kilogram of water, there's some sense in which this is not a pure fact, but it seems very close to a pure fact, and although pure-factness about water in this cup may be unattainable, it is not infinitely far away.

Comment by jp on The new Effective Altruism forum just launched · 2018-11-08T18:12:50.108Z · LW · GW

Lead developer here. Yes. By "based off of," he means "forked from with fairly minimal changes."