philh

Wikitag Contributions

Comments

Sorted by
philh20

The internet can convey much more information much more quickly, but radio is still better for creating common knowledge, in part because of its limitations (like having a fixed small number of channels).

The internet has lots of channels, but there are only a small number of official government channels.

I'm tickled by the idea of: the coup's success depends on which side first manages to convince an aide to change the password to the presidential twitter account.

philh95

Huh, I'm also British and I thought the first version looked like a placeholder, as in "no one's uploaded an actual cover yet so the system auto generates one". The only thing making me think not-that was that the esc key is mildly relevant. I bought the second one partly because I was a lot more confident I was actually buying a real book.

I guess part of what's going on here is it's the same grey as the background (or very close?), so looks transparent. But even without that I think I'd have had a similar reaction.

philh40

Feels like there might be two broad categories here?

If there's a visible autistic person and an invisible autistic person, it's likely the visible one is "more autistic" in some way that I think is meaningful though also kinda reductive. Same for two people into BDSM (more into it → does it more often → more likely to get caught, more likely to have to choose between "admit to this or pretend to do nothing with your time"), or two drug users (easier to hide a small amount of drug use than a large amount). There are other factors too, but "more of the trait → more likely to visibly have it" feels like a common thread in them.

Whereas I kinda don't feel the same with two atheists in a religious community, or two people with HIV, or two left-handers. The reason I learn one of them has HIV and I don't learn that about the other, probably (according to my current limited knowledge of HIV) isn't that the first has more HIV.

The throughline here is, one feels like the trait in question has (something like) a unimodal distribution, where being further out on the tail makes you more visible, and the other feels like it has a bimodal distribution.

Not sure if it's useful to make this distinction. We could talk about what bayesian update is correct in each case, which inferences are actually justifiable to make and which aren't, but part of the whole thing here is "the reason I'm paying attention to these things in the first place isn't necessarily downstream of sensible bayesian updates". I don't know what's a random stereotype I have and what actually correlates with the thing.


Which category is "homosexual in 1950" in? In my ontology, I'd say it's the second. I expect that mostly, homosexuals are more likely to be outed if they have more gay sex, but according to my ontology that's "being equally gay, but hornier". Could also be "equally gay, but has less privacy" or "but cares less what the rest of us think" or so on. (There is something of a spectrum from straight to gay via bi, where a bi man is "less gay" and less likely to get caught, but overall feels closer to the bimodal thing.)

But... in 1950s ontology, is that how they'd have seen things? Maybe they would have thought of it as, like, "the more gay sex you're having, the more gay you are". Is this hypothetical other ontology just as good as mine? My guess is no, and part of the reason they (hypothetically) have a worse ontology than me is downstream of the fact that they were punishing people for having gay sex and that made it harder to see the territory.

(Related: recent Ozy essay about the history of the ontology of transness.)


Feels like this comment wants some kind of conclusion but I don't have one.

philh*20

Fall of Civilizations #19: The Mongols

The Eurasian steppe is big and flat. Not many trees, lots of grass. Supports big herds. Thousands of years ago the ancestor of the horse arrived. At first it was used for meat and milk, then later for riding. The Proto-Indo-Europeans came from the area and spread themselves around a bunch. Much later the Huns came from the area and made Rome sad.

Around the 12th Century, the Mongols were one of a bunch of different nomadic groups living in the area. They had a not-so-great bit of territory in the north. They were themselves divided into smaller groups. There was lots of fighting among everyone. Lots of bits where I'm not sure if the people in question are Mongols or not.

A lot of the story comes from "The Secret History of the Mongols". Not sure if it's clear how reliable it is.

Temujin's mother fell in love and got married. She and her husband were travelling somewhere when they were attacked. She told her husband to ride away and save himself so he could marry someone else later, and he did. She got captured and taken as someone's second wife.

Temujin grew up with various siblings (all brothers?), including one older half-brother (from his father and his father's first wife). There were some tensions between them. I think it's around now he gets a blood brother. At some point he rides somewhere else, to stay with another family for a bit? He falls in love with someone there. I think he's something like 14 at this time. Before they can marry, he hears that his father has been killed by relatives of his mother's previous husband, and returns.

His mother and her sons are all abandoned by their tribe and expected to die over winter. But she keeps them alive, and when the tribe comes back they're surprised and a bit disconcerted. At some point he teams up with a brother to kill the eldest brother. The tribe is a bit worried that as he grows he's gonna bear a grudge against them, and he's sent into slavery. He escapes that, goes to claim his bride that he fell in love with previously, rejoins his family, and... I guess probably isn't at this point spending time with the people who sent him into slavery?

He and his wife and family join up with some other tribe including his blood brother. He becomes a good warrior, and he's popular. He divides spoils according to who did well in battle, not according to rank. And if a warrior dies, he gives that warrior's shares of the spoils to his wife and kids. He and his blood brother get more distant, and at one point BB snubs him and his wife encourages them to leave instead of staying. They do, and a bunch of others come with them because he's popular.

I don't remember how, but this escalates into a big feud between Temujin and BB. Eventually there's a big battle. Temujin loses and has to flee. Not clear where he goes but likely China. At this time another group of steppe nomads had conquered Northern China. They had a strategy of protecting themselves against yet more steppe nomads, by hiring some of them as mercenaries to keep the others out. When the mercenaries got too powerful, they'd hire a different group instead. Temujin gets hired, and I think starts assembling a big army again.

What he does is, when his group beats another group, he kills the elites but lets the warriors live and join him, instead of whatever is the custom (killing them all?). His mum, still with him, ceremonially adopts a kid from the group.

(I don't remember what happens to his wife during all this. Maybe she's with him in China. But at some point she's kidnapped, and then he recovers her, and it's not clear whether her first kid is his or not. He keeps the kid but names him something translating to "guest".)

His group grows like this, and eventually the Mongols appoint someone (his blood brother?) as something like "Khan over Khans" to face him. But this time he wins. Blood brother flees. Some of BB's lieutenants offer to betray him to Temujin, but Temujin has them killed for disloyalty. When he does capture BB, he has him executed without spilling blood, which is a mark of respect.

Now Temujin is the ruler of all Mongols and I think all steppe nomads. Calls himself Chinggis Khan, "great ruler" or something, it's a title that hasn't been used before. (Translation through Arabic and Latin turns the ch into a soft g and then a hard g, and now he's usually known as Genghis.) Reorganizes things a bunch. Warriors are organized in groups of ten who live together; then ten of those come together and elect a leader; ten of those, and ten of those, is a force of 10,000 which is big enough to be formidable but small enough to travel fast and feed themselves. He imposes a bunch of regulations. Allows a certain amount of religious freedom (there are Muslims, Buddhists, Christians - Temujin isn't any of those). Everyone has to perform one day a week in public service.

But the steppe nomads have historically been kind of infighty and Temujin's worried they'll go back to fighting each other if he doesn't give them someone else to fight. First he goes after a smallish nearby kingdom. They ask North and South China for help, but those places are like "um, we like it if our enemies fight each other". The kingdom falls. Then North China - I think until now he's been kinda sorta a vassal of them, but their emperor dies and the new one sends an envoy out to get him to reaffirm that he's still a vassal. He declines. There's the whole gobi desert in between so China isn't worried, but the horde crosses it and starts laying waste.

Probably they didn't really intend to conquer the place, just raid and leave, but they kept winning battles so why not keep fighting them. Eventually they get to Beijing. The emperor (I think another new one) agrees that he's now a vassal. The horde makes to leave, camping on the edge of the desert until a good season to cross. But then they get word that the emperor is trying to flee Beijing, and they get pissed at that and I think this time actually enter Beijing and do lots of killing.

Finally they go back to the steppe. Next, some traders on the silk road (which now goes through Mongol territory, and maybe the traders were themselves Mongols?) get killed by some noble in, I think somewhere that's one of the modern-day 'stans. Temujin sends some ambassadors to say, I'm sure this was just a misunderstanding, send us that guy to be executed and we'll forget the whole thing happened. I think two of those are killed and one allowed to return. So Temujin starts laying waste around there too. When Samarkand tries to defend themselves, they kill one of his sons-in-law; he asks his daughter what should happen to them and she says everyone in the city should be slaughtered and the place completely ravaged so it can be used for farmland, so that's what happens. (Um, except it doesn't sound like it went quite that far.)

By the time he's done, he has the largest empire the world's seen.

Some battle strategies:

  • They have Chinese engineers building siege engines.
  • If a city surrenders, they let the inhabitants live. If not they kill them.
  • They force peasants to flee into the city. Then the city has lots of people and no food.
  • They capture peasants and make them do some of the siege engineering work, so the defenders have to spend arrows shooting their own citizens.
  • They can travel fast. No need for fires to alert people to them. Instead of cooking meat, they can put it under their saddle to tenderize it. If there's no food, they can drink their horses' blood. (Each warrior brings like five horses.)
  • If they want to seem like a larger force than they are, they can tie branches behind their horses to whip up dust.

Maybe that's the end of Temujin's large scale conquest? He's old by now and thinking about immortality, and when a wise man tells him that's not really an option he thinks about succession. He has four sons by his first (favorite) wife and gathers them to him.

  • #1 is Guest, he's always treated him like a legit son but Chaggatai resents him.
  • #2 is named Chaggatai.
  • #3 is well-liked but an alcoholic.
  • #4 is ruthless even by Mongol standards, and doesn't always spare surrendering cities. By Mongol custom, the youngest son often inherits titles, though not titles that are meant to denote merit (like Khan).

His plan is to split the empire into three, have one son ruling over each part and another ruling over those three. Who gets to be top dog?

Temujin asks Guest to speak first, and Chaggatai gets annoyed at that and he and Guest argue. Eventually they realize they're not helping their cases, and agree that the alcoholic should rule.

Eventually Temujin dies en-route to pacify some rebels. His burial place is... kept secret to the point where there are rumors that 20,000 people who just happen to come across the funeral procession are killed, and also guarded by a load of arrows stuck into the ground and by warriors?

(Notable omission: my understanding is he had a lot of children, and I don't remember any mention of that.)

So his alcoholic son becomes Khan. I don't remember his name, but I think from now on the Khans go by "(first name) Khan". This one is very profligate, keeps giving people gifts. Also decides he wants a city. So he builds a big one, Karakorum. It's very impressive. But it's in a silly place for a city, no agriculture so it needs wheat imported from far away, and silk road traders need to be bribed to come visit because it's not on the way to anywhere.

So he starts running out of money, and decides to fix that by invading Europe. Now there's an idea that maybe the whole world should be united under Mongol rule.

The Mongols entered eastern Europe once before. (That caused a bit of confusion, because there was an idea at the time that there was a great Christian king somewhere in the east and one day he'd come forth and reclaim Jerusalem. Obviously that didn't happen.) They chased some other steppe nomads there. I think the story was something like: they told the powerful locals "we just want these nomads, let us kill them and leave and we won't bother you". The powerful locals weren't scared of the Mongols and refused. That ended badly for the powerful locals.

Now they invade once more, and do a lot of damage in eastern Europe. They eventually turn back because the alcoholic Khan dies?

I think at this point the empire fractures into four, or maybe there's another Khan in between. I've mostly lost track of the story from here. Monke Khan eventually inherits part of the empire. He wants to invade southern China and gets his... brother? son? Kublai to do it, but Kublai isn't very successful. When he dies, there's a power struggle between Kublai (who's spent most of his life in China) and a brother. The brother is a good diplomat but not a good strategist and loses, I think with Kublai attacking Karakorum. Kublai rules from China and adopts a bunch of Chinese customs and the Mongols back home kinda feel like they're not being ruled by one of them any more. There's an invasion of Japan, where Japan is saved by the kamikaze (divine wind, causing shipwrecks). Eventually Kublai is overthrown by what becomes the... Qin dynasty? and China starts looking inwards, doubling down on building its wall and burning its big fleet of ships.

Stuff happens in other parts of the world, too. At one point caliph of Baghdad is killed by having molten silver poured in his mouth, eyes and (ears? nose?).

Marco Polo shows up in the story too. He travelled at some point to somewhere in the Mongol empire, stayed there for a while, and then left. When he went there he travelled by land, but by the time he left the Pax Mongolica was less strong and he travelled by boat.

And at one point there was a religious debate between a Christian, Muslim and (Jew? Buddhist?). The participants also had to drink a lot while debating. The Christian claims to have won easily, but no one converted.

Ultimately the winners were western Europe, who had been kind of a backwater previously because they weren't on the silk road. They didn't get invaded but did get a bunch of the benefits of Mongol technology.

philh21

I'm inclined to agree, but a thing that gives me pause is something like... if society decides it's okay to yell at cogs when the machine wrongs you, I don't trust society to judge correctly whether or not the machine wronged a person?

Like if there are three worlds

  1. "Civilized people" simply don't yell at gate attendants. Anyone who does is considered gauche, and "civilized people" avoid them.
  2. "Civilized people" are allowed to yell at gate attendants when and only when the airline is implementing a shitty policy. If the airline is implementing a very reasonable policy - not just profit maximizing, but good for customers too, even if it sometimes goes bad for individual customers - "civilized people" are not allowed to yell at gate attendants.
  3. "Civilized people" are allowed to yell at gate attendants when they feel like the airline has wronged them.

...then (broadly speaking) I think right now we live in (1), but I think I'd prefer to live in (2). But I'm worried that if we tried, we'd end up in (3) and I tentatively think I'd like that less.

philh60

Are ads externalities? In the sense of, imposed upon people who don't get a say in the matter?

My initial reaction was roughly: "web/TV/magazine ads no, you can just not visit/watch/read. Billboards on the side of the road yes". But you can also just not take that road. Like, if someone built a new road and put up a billboard there, and specifically said "I'm funding the cost of this road with the billboard, take it or leave it", that doesn't feel like an externality. Why is it different if they build a road and then later add a billboard?

But if we go that far, we can say "people can move away from polluted areas"! And that feels wrong.

Hm. So the reason to care about externalities is, the cost of them falls on people who don't get to punish the parties causing the externalities. Like, a factory doesn't care if I move away from a polluted area.

So a web ad, I go to the website to get content and I see ads. If the ads are unpleasant enough that they're not worth the content, I don't visit the website. The website wants me to visit (at least in part so they can show me ads), so they don't want that outcome. The ad can't bring the value I get from the website below 0 (or I'll not go), and it can't dissuade so many people that it overall provides negative value to the website (or they won't show it). They can destroy value relative to the world with no ads on the website, but not relative to the world without this website. For, uh, certain definitions of "can't". (And: ads on websites that also want me there for reasons other than showing me ads, are probably less bad than ads on websites that only want me there to show me ads.)

For a billboard on the side of the road... well, who's getting paid? Is it typically the person who built the road? If not, the incentives are fucked. Even if it is... like the website the road is giving me value, and the ad still can't bring the value it gives me below 0. 

One difference: most of the value to me is getting me somewhere else, and the people and businesses at my destination probably value me being there, so if the value I get from the road goes to 0 and I stop taking it, that's bad for other people too. Another difference: what's the equivalent of "the world without this website"? Maybe if there's no road there I could walk, but walking is now less pleasant than it would have been because of the road (and the billboard that isn't narrowly targeted to drivers). Also feels significant that no one else can build a road there, and give me a choice of tolls or ads. The road itself is an externality.

I think my main conclusions here are

  • Billboards still feel like an externality in a way that web ads don't
  • If I want to think about, like, idealized models of free-market transactions, roads are a really weird thing to model.
philh171

Absolutely. In adversarial setting (XZ backdoor) there's no point in relaxing accountability.

Well, but you don't necessarily know if a setting is adversarial, right? And a process that starts by assuming everyone had good intentions probably isn't the most reliable way to find out.

it's exactly the case when one would expect blameless postmortems to work: The employer and the employee are aligned - neither of them wants the plane to crash and so the assumption of no ill intent should hold.

Not necessarily fully aligned, since e.g. the captain might benefit from getting to bully his first officer, or from coming to work drunk every day, or might have theatre tickets soon after the scheduled landing time, or.... Obviously he doesn't want to crash, but he might have different risk tolerances than his employer.

(Back of the envelope: google says United Airlines employs ~10,000 pilots. If a pilot has a 50-year career, then United has 200 pilot careers every year. Since 2010, they've had about 3000 careers and 9 accidents, no fatalities. That's about 0.3% chance of accident per career. An individual pilot might well want to make choices that have a higher-than-that chance of accident over their career.)

Something was swept under the carpet here and that may have been avoided if the investigation was careful not to assign blame.

Stipulated, but... suppose that the investigators had run a blameless process, the captain had told the truth ("I disabled the EGPWS because it would have been distracting while I deliberately violated a bunch of regulations") with no risk of getting fired, and they found some rot in the culture there. Can they do anything about it? It seems like if the answer is "yes", and if anyone benefits from the current culture, then we're back at people having incentives to lie.

philh161

Blameless postmortems are a tenet of SRE culture. For a postmortem to be truly blameless, it must focus on identifying the contributing causes of the incident without indicting any individual or team for bad or inappropriate behavior. A blamelessly written postmortem assumes that everyone involved in an incident had good intentions and did the right thing with the information they had. If a culture of finger pointing and shaming individuals or teams for doing the "wrong" thing prevails, people will not bring issues to light for fear of punishment.

This seems true and important to me. But also... some people really are incompetent or malicious. A blameless postmortem of the XZ backdoor would miss important parts of that story. A blameless postmortem of Royal Air Maroc Express flight 439 would presumably have left that captain still flying? At any rate, as long as "the captain gets fired" is a possible outcome of the postmortem, the captain has incentive to obfuscate. (Which he tried unsuccessfully. Apparently we don't know if he was actually fired though.)

I guess there are cases where "ability to get rid of people" is more important than "risk that people successfully obfuscate important details", and some cases where it's not. I don't know how to navigate these.

philh42

Huh, thanks for the correction.

Smaller correction - I think you've had her buy an extra pair of boots. At $260 she's already bought one pair, so we apply  thirteen times, then multiply by 1.07 again for the final year's interest, and she ends with no boots, so that's $239.41. (Or start with $280 and apply  fourteen times.)

Not sure why my own result is wrong. Part of it is that I forgot to subtract the money actually spent on boots - I did "the $20 she spends after the first year gets one year's interest, so that's $21.40; the $20 she spends after the second year gets two years' interest, so that's $22.90..." but actually it's only $1.40, $2.90 and so on. But even accounting for that, I get $222.58. So let's see...

Suppose she only needs to buy two pairs of boots. According to your method she goes $40 → $21.40 → $1.50. (Or, $40 and no boots → $20 and boots → $21.40 and no boots a year later → $1.40 and boots → $1.50 and no boots a year later.) According to mine, of her original $40, $20 of it earns no interest and $20 of it earns a years' interest. But that assumes the interest she earns in that year is withdrawn, she gets to keep it but it doesn't keep earning interest. So that's why I got the wrong answer.

philh20

Well, I should like to see some examples. So far, our tally of actual examples of this alleged phenomenon seems to still be zero. All the examples proffered thus far… aren’t.

From https://siderea.dreamwidth.org/1477942.html:

"These boots," I said gesturing at what I was trying on, on my feet, "cost $200. Given that I typically buy a pair for $20 every year, that means these boots have to last 10 years to recoup the initial investment."

That was on January 17, 2005. They died earlier this month – that is in the first week of December, 2018. So: almost but not quite 14 years.

So, purely as an investment, they returned a bit under $80, which is a 40% ROI.

But... if we're talking about this just as an investment, we need to compare to other investments. Let's say the S&P 500 returns 7% consistently (I think that's pessimistic - note, not adjusting for inflation because the $20 boots haven't changed with inflation either).

  • In one world, Siderea buys $200 boots and invests $80. After 14 years, she has  and no boots.
  • In another world, Siderea buys $20 boots and invests $260. A year later she withdraws $20 and buys boots, and so on. After 14 years, she has... finite geometric series, , I think she has $483 and no boots.

So if we think of this as a purely financial investment, I guess it was a bad one?

(This is also often missing when people talk about buying versus renting. Yes, the mortgage is often lower than rent, and house value is likely higher at the end, but you gave up investing your deposit. How do those effects compare? Probably depends on time and place.)

Load More