A Reasonable Approximation

Latest posts

99% shorter

I listen to podcasts while doing chores or at the gym, and often feel like I'm learning something but then can't really remember anything afterwards. So for the past ~month I've been doing an experiment where I write brief summaries of them afterwards, usually same-day but sometimes a bit later. Generally I avoid all forms of fact checking, both "what did the episode say" and "what is actually true", though I don't stop myself if I feel like doing it.

I've been posting them to my shortform on LessWrong. Mostly in reply to a single comment for tidiness, but two of them I accidentally posted to the top of the thread and one as a reply to another. Initially that was just because I wanted to be able to write them from both my work and personal laptops. (This blog is published through git. I don't have the repo on my work laptop, and probably shouldn't do.) But I kind of like it. A downside is it's slightly less convenient to include episode numbers or air dates or even titles unless I remember them. So I might be less consistent about that, though it feels good to have a fairly straightforward way to look up the episode given my summary.

I've skipped all the fiction podcasts I listen to, because that's not why I listen to fiction. Also most interviews, those seem generally hard to summarize, though it would probably be reasonable to extract a few individual points. And one Planet Money episode seemed like it would be irresponsible to summarize carelessly, and I didn't feel like trying to be careful about it. But I've summarized every episode of 99% Invisible and History of English, and all but that one episode of Planet Money, that I've listened to in this time. Also one episode of Rationally Speaking and one of Corecursive.

I'm not really sure what this experiment was trying to test. I was curious, so I did it. Some things I've maybe vaguely learned: first, I think I can generally do a decent summary. I frequently get distracted for a few seconds at a time, but there's a difference between that and being actually distracted, and I think it shows. (I listened to 99% Invisible, "Tanz Tanz Revolution", while looking at parking restrictions in my area and anticipating my evening plans. There's significant parts of that episode that just rolled right past me.) I don't think you could reliably pick out the episodes I didn't do same-day, I mostly don't even remember myself. (Hot Cheetos was several days later, but I did go over it a bit in my head in between, which I wouldn't have done if not for this project.)

Second, I seem to retain less of History of English than other podcasts. That feels like it matches my intuition. I expect Causality, too, but that only recently released a new episode and I haven't listened to it yet.

Third, this causes a marginal shift from podcasts towards music for me. I predicted this and I'm okay with it, I feel like I don't listen to music enough relative to how much I enjoy it.

If I wanted to make it more of an experiment, I could randomly select some episodes to write up quickly, some to write up the next day, the next week, next month. Not sure how much I feel like this, but maybe. Another thing would be quizzing me on the episodes I did and didn't write up, without rereading. And I'm curious whether I'll ever actually want to refer to these, and if so whether I can predict which ones I will.

I haven't been tracking my time closely, but I think they mostly take me 20-30 minutes to write? I listen between 1.5x and 2.1x speed depending on the podcast, so that's considerably longer than they usually take to listen to.

Dunno what I'll do in future. Maybe write up some summaries, of episodes I found particularly interesting?

I'm including here three that I particularly liked, but if you feel like reading the rest see the LW thread.

Planet Money (14 May 2021): Blood Money

America lets you sell blood plasma for money. The centers will call it a donation, but you get money for it, so. You can come in a couple of times a week and especially for low income people it can be a significant addition to their income. There are referral bonuses, and you get paid more if you come in more often. (Seems weird?) These centers are mostly located in low-income areas.

There's a history here involving Nicaragua. Under a dictator there was a center doing this, and a journalist was writing about concerns, and eventually the dictator had the journalist killed. Riots in the aftermath left the center burned down and eventually the dictator got deposed. At some point the WHO wrote up an agreement or something not to allow blood plasma to be sold. Almost everyone's signed, but not the USA. Now almost everyone gets their blood plasma from the USA. Four exceptions, who allow it to be sold and are self-sufficient: Germany, Austria, Czech Republic, Hungary.

(Farmer's Dilemma!)

We speak to a Canadian healthcare person about why Canada doesn't allow it. Three concerns. He's not too worried about incentives for people to sell bad blood, apparently we can sanitize it, even of HIV. He's also not too worried about health impacts on sellers; they get a checkup every four months to make sure they're still good, and there's some anecdotal evidence that maybe it should be more frequent but basically it seems fine. He seemed more concerned about "if people can sell plasma, will they do other things like regular blood donation for free?" I don't remember the commentary on that. I think he said that if the USA didn't allow selling they'd probably have to in Canada, but as long as they do it's unlikely to change.

We also speak to a Brazilian doctor saying that plasma and the things it's used for are essential, there are people who will die without it, get over yourselves.

Concerns that if either demand raises (finding new uses: there are studies showing promise in Alzheimers) or supply drops, there might not be enough. In fact supply has dropped during Covid: possible reasons include "sellers need the money less thanks to stimulus"; "if your kids are at home all the time you might be too busy"; "a lot of the sellers near the border are Mexicans who can't come over any more".

99% Invisible #434 (9 Mar 2021): Artistic License

This is the most American story. (I actually wrote it up before starting this experiment.)

After states start requiring license plates, Idaho realizes they can be used for advertising, and start boasting about Idaho potatoes on their plates. North Idahoans grumble because that’s more of a Southeast Idaho thing. Tourists start stealing plates as souvenirs, causing people to be very confused when they’re pulled over because who checks whether they still have a license plate.

Anyway, New Hampshire’s state motto is Live Free or Die, I don’t know if that’s just a generic America thing or a specifically fuck-communists America thing. But a super fuck-communists guy gets them to put the motto on the license plate, presumably for fuck-communists reasons but I dunno if that was explicit or just subtext.

And then a Jehova’s Witness is like, no, I don’t want to, God gave me life and I’m not gonna give that up for freedom. So he starts covering up that bit with tape. And he gets arrested and the fuck-communists guy is now governor and not inclined to give an inch, so it goes to the Supreme Court who split 6-3 but the pro-freedom side wins, the government is not allowed to compel you to express your love of freedom.

Later: Texas allows specialty plates, some group designs a plate to support a cause and then you pay a little extra for it, some going to the group and some to the state. Most of these designs are just rubber stamped. But this is Texas, so the Sons of Confederate Veterans want a license plate supporting their cause, and they want the Confederate flag on it. The state says no, they sue the state, and the Supreme Court sides with the state 5-4.

Corecursive (2 May 2021): Etherium Rescue

"Daryl" was an ETH user who fat-fingered a transaction. Went online for help, guest said sorry, nothing anyone can do. Then later guest went o shit maybe there is.

Daryl was playing with uniswap, a smart contract letting people provide liquidity for exchanging crypto, e.g. ETH for USDC. Normally when providing liquidity you'd do two things in one transaction, with something like a try/catch letting you do them atomically. I guess Daryl had only done one of them? Anyway, his money was just sitting there, and as soon as anyone tried to take their liquidity from uniswap they'd get Daryl's money as well.

Guest realized this and went to check, and the money was still there. But! He also remembered stories of generalized ETH frontrunners. These will examine the pending transactions, see if there's something in there they can use to make money, and if so, submit their own transaction with a higher fee so it gets executed first. Guest worried that one of these would show up if he tried to recover the money. He asked on a group chat if others would also be worried, some of them were, and they got together to try to figure something out.

Ultimately they'd need to do some kind of obfuscation so that a bot wouldn't try the thing they were doing. They settled on two separate transactions in one block, where the second one wouldn't do anything unless the first had already happened, hoping bots would only try them separately. But there's stuff set up to protect you from making transactions that don't do anything, and it was stopping them from making the second.

Guest was tired and stressed and the money might disappear at any minute, so eventually Guest said YOLO we'll do them in two different blocks and hope. The second transaction got front-ran and they lost the money. On the plus side his worries were vindicated.

Guest and Adam (host) discuss Meditations on Moloch. The thing they take away from it is that you need regulation/Leviathan. Guest says for Hobbes the Leviathan was hereditary monarchy, recently we've been trying democracy and that seems better overall, but he's optimistic that smart contracts will be another solution.

Posted on 27 May 2021