All of jp's Comments + Replies

I would also expect extraneous details like, "got sick and fell of the wagon" or similar to add significant noise. And with only one data point each, it'd be hard to know the variance to use. I'm guess I'd trust this study more?

To be clear I don't know what I'm doing really. I do think that it failed to get the precise thing I was looking for though.

2habryka
😡

"Cruxy" is a useful term to have in my vocabulary. I use it relatively loosely to refer to the type of thing I look for in a double crux. A consideration is more "cruxy" if it's closer to a but-for support for a proposition. Interestingly (mildly) this is very similar to the definition of "crucial," and in fact the etymologies are the same.

The California vibes in that video are immaculate. Even mentioning Windham Hill Records!

I'm a big fan of land reclamation, but hadn't heard of most of these barriers, thanks!

You might be interested in this "Best of LessWrong" post, Make more land.

This post is very cute. I also reference it all the time to explain the 'inverse cat tax.' you You can ask my colleagues, I definitely talk about that model a bunch. So, perhaps strangely, this is my most-referenced post of 2022. 🙃

My explanation of a model tax: this forum (and the EA Forum) really like models, so to get a post to be popular, you gotta put in a model.

I've referenced this post several times. I think the post has to balance being a straw vulcan with being unwilling to forcefully say its thesis, and I find Raemon to be surprisingly good at saying true things within that balance. It's also well-written, and a great length. Candidate for my favorite post of the year.

2Raemon
I think this comment is probably getting at something I was unconsciously tracking while writing the post, but I don't quite understand it. Could you say this again in different words?

You get paid if Wave does well whether they like you or not.

Private companies can and do prevent people they don't like from selling stock to private buyers. So as an addition to this comment, I'd note that "the ability to cash out before an IPO 5+ years in the future" is a strong reason not to make an enemy of your former startup.

I've historically said 1-2 weeks of skilled engineering work. That will lower by a factor of 2 after this branch, plus some follow ups, get merged.

Thanks for writing about ForumMagnum! This software is so much of my life, but understandably gets little attention an an object in its own sake.

That's changing a bit now, and more people are reaching out about using it. — I think it's the best forum software out there.

If someone reading this wants to build an instance, feel free to reach out.

4Viliam
How difficult it would be to run an instance with minimum customization? That is, suppose that I am happy with all the default options (wherever that makes sense). I just need a forum; I don't care about details.

Talon Voice, which I use for voice-commanding my computer, is very fast and has a linux version. I don't know if it would run on a small machine, but it seems worth a shot. And it seems perfect for your command-centered use-case. To get more information, I would ask in their slack.

2jefftk
Looks like Talon uses (their own) W2L Conformer as their recognition engine. I'll poke at it!
jp2116

I love this post. Both the content, and the writing — I felt like you were happily telling me about your interest, and it made me happy.

This comment made me happy! Thanks for the positivity!! 💖

My mom and her siblings report learning their phone number this way. It's effective enough that I know that the house phone of my grandparents half a century ago ends in seven.

This is extreme JP bait.

Exactly the sort of mild optimization that I will become obsessed with.

I think you and the previous commenter would both do well to read the short, hyperlinked definition. (Sorry.)

2Three-Monkey Mind
Oh, that's something significantly different from what I had in mind. Thanks for pointing me to the page that explains the concept.

We use Asana for this. (It's broadly great, and I don't understand how Jira isn't getting its lunch completely eaten. And I'm not just saying that because of my funder.)

I agree with a lot of this, though I think the mental attitude here is still extremely useful. For example, you may be dealing with something outside of your usual ticket system, or a ball may be smaller than something that would justify an Asana ticket.

5Viliam
My model of Atlassian business model goes like this: Take a popular open-source product, and make a crappy version of it with half the functionality and inconsistent UI, but make it possible to create plugins and sell them at your marketplace (where you take a cut). Then sell it for lots of money, and when anyone asks whether it has some functionality X, say "there is a marketplace with many plugins, I am sure there are many high-quality implementations of X". Write in the contract that you only provide support for the basic installation without any plugins; if plugins are installed, proper functioning of the system is no longer your responsibility, and the customer should call the author of the plugin instead! This seems to somehow work in practice. On one hand, you can check the checkbox "we provide support", and also check the checkbox "our product supports X" when someone makes a plugin for X (so you soon provide more functionality than the original open-source project you copied... so when someone later proposes to replace your product with open-source, some manager will ask "but does the open-source solution provide integration with SharePoint?" or something like that), but when the customer installs the plugin for X, the support is no longer your problem. In theory, your product can be configured to do anything, but in practice, the customer is supposed to configure it in their own time, which most customers will refuse to do (somehow, spending $100,000 to buy your software is okay, but spending $10,000 to pay their employee his salary until he figures out how to configure it properly is not), so most customers will use the default settings, or the default settings plus a plugin. (And the integration with SharePoint does not work anyway, but no one is complaining about that, because the people who actually use the software do not care about SharePoint integration.) This is too cynical, but I haven't heard a better explanation that would fit the known data.
jp110

I have a classic rationality comment on the EA Forum that's reasonably popular. I thought I'd crosspost here. The context is "What are work practices that you’ve adopted that you now think are underrated?"

***

CEA (my employer) has long had the concept of "who owns this ball."[1] I'm gonna have a hard time in this answer conveying exactly how much this has become a whole encompassing working philosophy for me.

Level 1: The alarm bells about dropped balls

If you are having a conversation and someone's like "we should do X"... Someone should really be the p... (read more)

3Viliam
This is also called "tickets". Tickets for tasks that happen repeatedly can have a workflow assigned to them, which can be as simple as a linear sequence of steps, but also more complex, and you can have a software that tracks them. Like, you click a button on your smartphone, and it will show you a list of tickets currently assigned to you, maybe also ordered by priority. When you move the ticket to someone else, it disappears from your list and immediately appears on their list. So the tasks are always remembered and always have a person assigned, until they get closed. Tickets are usually used for predictable, repetitive tasks with known workflow. But nothing prevents you from creating a generic "workflow" containing only one step "solve the problem", and using it for everything else. When a new ticket is created, unless the owner is specified explicitly, the system should automatically assign it to someone.  Then you just need the person "on call" who registers everything that happens, and creates a new ticket for that. There are commercial solutions, such as the infamous Atlassian Jira (most companies use it, most users hate it with a passion), or free solutions... I don't know what is currently the state of art. There is a list on Wikipedia. I think every organization should use something like this; many individuals would probably also benefit from a personal ticket system. But there is an overhead with setting up the system (installing the software, setting up the workflows). Not sure of there is an online system where one could just register and start using it... if there is, someone please give me a link; if there is not, sounds like a business opportunity. If you understand how the system is supposed to work, you could implement a poor man's version using a spreadsheet. Each row is a ticket. Columns: ticket id, who is it assigned to, workflow type, workflow phase, description. Maybe create a wiki page for each ticket, and have it linked from the sprea
3Raemon
This is pretty great and I think honestly makes a good top-level post (albeit with maybe some fleshing out)

We've been thinking about this for the EA Forum. I endorse Raemon's thoughts here, I think, but I know I can't pass the ITT of a more transparent side here.

jp157

This comment feels like wishful thinking to me. Like, I think our communities are broadly some of the more truth-seeking communities out there. And yet, they have flaws common to human communities, such as both 1 and 2. And yet, I want to engage with these communities, and to cooperate with them. That cooperation is made much harder if actors blithely ignore these dynamics by:

  1. Publishing criticism that could wait
  2. Pretend that they can continue working on that strategy doc they were working on, while there's an important discussion centered on their organizat
... (read more)
2TekhneMakre
What are your ideas for attenuating the anti-epistemic effects of belief lock-in and group think and information cascades?

I basically probably endorse this for you, but would also suggest whether you could do more automatic red-shifting and dimming of your lights in the evening.

Is that common?

My model was that this is the thing going on for many night owls. I believe I had studies at one point that would back this up, but could not immediately find them.

if you benefit from a visual indication of wake time blacking out the external light and replacing it with light under your control seems much better, if you can get it bright enough?

I'm not in principle opposed. The approach you mention has super conceptual benefits under the model you and I share. In practice, I find my friends often have lights that go from zero to very bright ... (read more)

When I had the chance to drive an electric car recently, I turned off "one pedal driving". I bet you could in a Tesla. Note for next time!

2jefftk
I looked in the settings and online at the time and couldn't find an option to configure this. Instead I found people saying it wasn't an option (ex: forum post)
jp1-3

This sounds like some dark magic to me.

This post was a great dive into two topics:

  • How an object-level research field has gone, and what are the challenges it faces.
  • Forming a model about how technologically optimistic projects go.

I think this post was good on it's first edition, but became great after the author displayed admirable ability to update their mind and willingness to update their post in light of new information.

Overall I must reluctantly only give this post a +1 vote for inclusion, as I think the books are better served by more general rationality content, but I'm terms of what I would like to see more of on this site, +9. Maybe I'll compromise and give +4.

This is great. Encouragement to turn it into a top level post if you want it.

Spot the problem with this statement:

The Galleri test did not detect DNA methylation patterns that are associated with cancer in your blood sample. In a clinical validation study, fewer than1% of individuals with this result were projected to have cancer.

It seems like a large amount of work of this post is being done by:

So people seem skeptical that we can cover large areas with these lamps.

Maybe the experts are thinking of large-scale deployments in schools, hospitals, airports, conference centers? I feel like numbers seem important.

2Roko
It's false IMO, see my comment below. https://www.lesswrong.com/posts/4Zjm8ycWhg6PzFwnF/far-uvc-light-update-no-leds-are-not-around-the-corner?commentId=NLHt7atxYX4Esuww5

It would help your signal-boosting if you hit space after pasting that url, in which case the editor would auto-link it.

(JP, you say, you're a dev on this codebase, why not make it so it auto-links it on paste — yeah, well, ckeditor is a pain to work with and that wasn't the default behavior.)

I think the problem with zoom meetings is not the meeting itself, but instead the bounds of the meeting. It's easier to have better coordination if you can freely wander in and out of a casual conversation. It's hard to get super-in-sync over, say, 60 minutes a day of facetime. To put another way, zoom does fine for "full meeting" mode, but much worse for casual, semi-meeting mode. VR does nothing to solve the second category, so I'm skeptical.

I really buy the argument Sinclair makes about reducing trivial inconveniences here. Let’s make a model.

Ambiguity has two main negative effects, according to me:

  1. Reducing precision in the prediction, because some of the prediction is based around interpretation of the creator’s foibles rather than a “true” resolution of a better-specified question.
  2. Making the forecaster’s lives worse, because they want to be forecasting world events, not guessing as to the creators behavior in unclear situations.

Let’s set 1. aside for now. 2. seems like a big deal for sure. ... (read more)

This is really neat. Thanks for helping me build a technical base (eyyyy) for understanding my partner's work.

One thing that wasn't clear to me is: if you can only sequence 150 bases at a time, how do you build a complete picture of a genome? One might come away from your post thinking that next gen sequencing is only useful for getting the edges of the genome, say for simple identification purposes. Based on "chopped up your fragments" and my preexisting knowledge, I expect you do some sort of chopping and then reassembly, but I'd be curious to learn more... (read more)

3Tapatakt
>how do you build a complete picture of a genome You put a lot of identical DNA strands in a sequencer and use an algorithm that identifies overlaps and restores the strand sequence. Toy example: if you have reads "GCTTTAGCCCCTAGG" and "TAGCCCCTAGGAATCC", there is propably "GCTTTAGCCCCTAGGAATCC" in the original DNA. Of course, real reads and overlaps are usually much longer. (Epistemic status: studied bioinformatics for 1 year, in 2016-2017, forgot a lot)
5tgb
Yeah, you break the big strand randomly up and sequence the starts and ends of each of the little fragment. Then you have to assemble all the little bits together, which is called either assembling a genome (if you don’t have anything else to work off of) or alignment (if you’re just comparing you sequenced data to an existing reference genome). These are computationally nontrivial, though alignment at least is well solved these days through tools like bwa or STAR. For something like wastewater survey you do ”meta genomics” since it comes from many bacterial/viral genomes. This process wasn’t really enough to get a complete picture of the genome. The reads were too short and so some parts of the genome were hard to read: anything that is highly repetitive cant be assembled accurately since it all looks the same. The recent “T2T” genome is really the first complete human genome we have despite being two decades after the human genome project finished. But the earlier reference was good enough for most things. Actually the very start and end of the chromosomes, the telomeres, were some of the hardest part part to sequence, hence the name of “telomere 2 telomere” genome for the new build. Older builds have like a million ”N”s at the start and end of every chromosome, denoting an unknown sequence (ACTG all possible) in the telomeres.

NB: the title no longer appears in bold in the first sentence, contra the style guide.

2abramdemski
I didn't fix it, but I de-bolded all the other technical terms that I spuriously bolded, so that distributional shift now sticks out more even though it is not in the first sentence. Not quite sure how to get it in the first sentence in a clean way, since I really feel I have to define IID first in order to define distributional shift properly.

Bunch of broken images in this one

1Aprillion
I mostly fixed the page by removing quotes from links (edited as markdown in VS Code, 42 links were like []("...") and 64 quotes were double-escaped \") ... feel free to double check (I also sent feedback to moderators, maybe they want to check for similar problems on other pages on DB level)

Sam's talking about the rich text editor.

2Vladimir_Nesov
Somehow this reply didn't create a new/unseen reply notification for me (with the black bell), instead there is an already-seen reply notification on the top of the list of recent notifications. I vaguely recall already having seen this bug in another Shortform reply, though I'd expect that this can't matter. So I probably got confused by clearing notifications about the other thread I got replies from at about the same time.

There is, by coincidence, a recent PR to fix this.

Is there a way to hide the curated sequences from the frontpage?

2Raemon
Not yet but I’ll likely be adding it soon. (I’m also going to be overhauling how curated sequences work and may do that first. I might actually end up re-disabling recommended sequences for older accounts until I’ve iterated on it more, then re-enable it, this time with a dismiss button)

I disagree here. Every time I've seen someone strong-self-vote their comment, I've felt it was pretty bad. It's rare, and not obviously worth the effort to fix, but I wouldn't agree with Ray's take.

1the gears to ascension
on my shortform i used self-strong-upvote to sort the videos list in a way that other users could vote on

I would also expect less exposure in a normal office meeting than at a dance

How do I square that with:

so I would say our ventilation was successful at keeping us from going above standard indoor exposure despite the number of people and active movement

I think "faces being further apart" is basically what I mean in the final two sentences.

2jefftk
Yes, sorry, that was confusing. In thinking about how risky a contradance event is I model it as [general air quality] * [closer to people's faces] * [effect of masking]. I was only trying to talk about the first term, since that's the part that can be controlled by choosing whether to have the dance inside/outside, and how ventilated you can make the space.

It's much harder to change a tag ontology once created.

3niplav
True. I just think there's so little activity here on the tags portal that marginally less caution is better than marginally more caution. Also strong-upvoted your tags contribution :-)

I think this is really cool. I nevertheless expect perhaps that I would be that a normal office meeting would result in less exposure than this dance. I'm not actually sure where my intuition is coming from, but I'm going to say it's due in part to the increased exertion of the dancers causing more potential virus to get mixed into the air, and that this effect is bigger than the increase in CO2 from exertion, if you follow me. The other part would be that perhaps the measurement is too far away from the center of the dance? Hypothetically, if I imagine finding out that the CO2 levels stayed below 1000ppm even in the middle of the dancers, I'd be less skeptical.

2jefftk
I would also expect less exposure in a normal office meeting than at a dance, but I think my expectation is mostly coming from people's faces being farther apart? I think this isn't quite the same as what you're saying: if at a future dance I brought the meter to the middle of the hall and stood there and found that CO2 levels were similar, that doesn't mean that if I were to strap the meter to the my face (and route my own exhalation elsewhere) I would see the same levels?

Tag suggestion: "Air Quality". There's a bunch of things in a cluster of space around here, you could imagine one or more tags. Carbon dioxide, air particulate pollution, and aerosolized respiratory pathogens. The last one  may seem a bit of an odd duck, but the techniques for dealing with it are often the same as the others.

4niplav
It probably took me less time to create the tag than it took you to write that comment ;-)

I wear a P100 sometimes, and it's incredibly hard for other people to hear what I'm saying when I'm behind one, which would be a factor.

2jefftk
I think in this particular situation it's not bad? It's not an activity where people do that much talking.

I get an access error for boston-effective-altruists.

I agree with Mingyuan that Google hasn't put a lot of work into making the UI a good user experience in a long time, but agree generally that email lists are probably a good option. I do think organizers should still post on Facebook though?

2jefftk
Sorry! The group privacy settings were wrong. Fixed now! (It was set so that the group was only visible to people in the group, which wasn't intended)
3Jeremyy
I also get an error for boston-effective-altruists. "Content unavailable Click here to try again." Perhaps a box needs to be checked to make it publicly available?

Economic efficiency and a 5 year old bet

5 years ago, my then-boss and I realized we disagreed about something. He thought the oil and gas sector was clearly on the decline, and would be a bad place to put your money. I was a big fan of the efficient market hypothesis, and though, well, shouldn't that already be priced in?

He was confident and willing to give me good odds, but wanted to be clear he was talking over the very long term. So we agreed: 5 years term, 70% odds. If the O&G sector trailed the S&P 500 by more than 25%, I'd pay him ¢30, if it ... (read more)

Load More