Archive:
In this section: Subtopics:
Comments disabled |
Fri, 28 Mar 2025
Does someone really have to do the dirty jobs?
Doing the laundry used to be backbreaking toil. Haul the water, chop the wood, light the fire, heat the water, and now you are ready to begin the really tough part of the work. The old saying goes "Wash on Monday", because Monday is the day after your day of rest, and otherwise you won't have the strength to do the washing. And the saying continues: “Iron on Tuesday, mend on Wednesday”. Routine management of clothing takes half of the six-day work week. For this reason, washing is the work of last resort for the poorest and most marginal people. Widows are washerwomen. Prisons are laundries. Chinese immigrants run laundries. Anyone with enough money to outsource their laundry does so. The invention of mechanical washing machines eliminated a great amount of human suffering and toil. Machines do the washing now. Nobody has to break their back scrubbing soiled linens against a washboard.
But the flip side of that is that there are still poor and marginalized people, who now have to find other work. Mechanical laundry has taken away their jobs. They no longer have to do the backbreaking labor of hand laundry. Now they have the option to starve to death instead. Is it a net win? I don't know. I'd like to think so. I'd like to free people from the toil of hand laundry without also starving some of them to death. Our present system doesn't seem to be very good at that sort of thing. I'm not sure what a better system would look like. Anyway, this is on my mind a lot lately because of the recent developments in computer-generated art. I think “well, it's not all bad, because at least now nobody will have to make a living drawing pornographic pictures of other people's furry OCs. Surely that is a slight elevation of the human condition.” On the other hand, some of those people would rather have the money and who am I to deny them that choice? [Other articles in category /misc] permanent link Tue, 25 Mar 2025
The mathematical past is a foreign country
A modern presentation of the Peano axioms looks like this:
This baldly states that zero is a natural number. I think this is a 20th-century development. In 1889, the natural numbers started at !!1!!, not at !!0!!. Peano's Arithmetices principia, nova methodo exposita (1889) is the source of the Peano axioms and in it Peano starts the natural numbers at !!1!!, not at !!0!!: There's axiom 1: !!1\in\Bbb N!!. No zero. I think starting at !! 0!! may be a Bourbakism. In a modern presentation we define addition like this: $$ \begin{array}{rrl} (i) & a + 0 = & a \\ (ii) & a + Sb = & S(a+b) \end{array} $$ Peano doesn't have zero, so he doesn't need item !!(i)!!. His definition just has !!(ii)!!. But wait, doesn't his inductive definition need to have a base case? Maybe something like this? \begin{array}{rrl} (i') & a + 1 = & Sa \\ \end{array} Nope, Peano has nothing like that. But surely the definition must have a base case? How can Peano get around that? Well, by modern standards, he cheats! Peano doesn't have a special notation like !!S!! for successor. Where a modern presentation might write !!Sa!! for the successor of the number !!a!!, Peano writes “!!a + 1!!”. So his version of !!(ii)!! looks like this: $$ a + (b + 1) = (a + b) + 1 $$ which is pretty much a symbol-for-symbol translation of !!(ii)!!. But if we try to translate !!(i')!! similarly, it looks like this: $$ a + 1 = a + 1 $$ That's why Peano didn't include it: to him, it was tautological. But to modern eyes that last formula is deceptive because it equivocates between the "!!+ 1!!" notation that is being used to represent the successor operation (on the right) and the addition operation that Peano is trying to define (on the left). In a modern presentation, we are careful to distinguish between our formal symbol for a successor, and our definition of the addition operation. Peano, working pre-Frege and pre-Hilbert, doesn't have the same concept of what this means. To Peano, constructing the successor of a number, and adding a number to the constant !!1!!, are the same operation: the successor operation is just adding !!1!!. But to us, !!Sa!! and !!a+S0!! are different operations that happen to yield the same value. To us, the successor operation is a purely abstract or formal symbol manipulation (“stick an !!S!! on the front”). The fact that it also has an arithmetic interpretation, related to addition, appears only once we contemplate the theorem $$\forall a. a + S0 = Sa.$$ There is nothing like this in Peano. It's things like this that make it tricky to read older mathematics books. There are deep philosophical differences about what is being done and why, and they are not usually explicit. Another example: in the 19th century, the abstract presentation of group theory had not yet been invented. The phrase “group” was understood to be short for “group of permutations”, and the important property was closure, specifically closure under composition of permutations. In a 20th century abstract presentation, the closure property is usually passed over without comment. In a modern view, the notation !!G_1\cup G_2!! is not even meaningful, because groups are not sets and you cannot just mix together two sets of group elements without also specifying how to extend the binary operation, perhaps via a free product or something. In the 19th century, !!G_1\cup G_2!! is perfectly ordinary, because !!G_1!! and !!G_2!! are just sets of permutations. One can then ask whether that set is a group — that is, whether it is closed under composition of permutations — and if not, what is the smallest group that contains it. It's something like a foreign language of a foreign culture. You can try to translate the words, but the underlying ideas may not be the same. Addendum 20250326Simon Tatham reminds me that Peano's equivocation has come up here before. I previously discussed a Math SE post in which OP was confused because Bertrand Russell's presentation of the Peano axioms similarly used the notation “!!+ 1!!” for the successor operation, and did not understand why it was not tautological. [Other articles in category /math] permanent link Tue, 18 Mar 2025We want to adapt baseball to be played on the moon. Is there any way to make it work? My first impression is: no, for several reasons. The pitched ball will go a little faster (no air resistance) but breaking balls are impossible (ditto). So the batter will find it easier to get a solid hit. We can't fix this by moving the plate closer to the pitcher's rubber; that would expose both batter and pitcher to unacceptable danger. I think we also can't fix it by making the plate much wider. Once the batter hits the ball, it will go a long long way, six times as far as a batted ball on Earth. In order for every hit to not be a home run, the outfield fence will have to be about six times as far way, so the outfield will be !!36!! times as large. I don't think the outfielders can move six times as fast to catch up to it. Perhaps if there were 100 outfielders instead of only three? Fielding the ball will be more difficult. Note that even though the vacuum prevents the pitch from breaking, the batted ball can still take unexpected hops off the ground. Having gotten hold of the ball, the outfielder will then need to throw it back to the infield. They will be able to throw it that far, but they probably won't be able do it accurately enough for the receiving fielder to make the play at the base. More likely the outfielder will throw it wild. I don't think this can be easily salvaged. People do love home runs, but I don't think they would love this. Games are too long already. Well, here's a thought. What if instead of four bases, arranged in a !!90!!-foot square, we had, I don't know, eight or ten, maybe !!200!! or !!300!! feet apart? More opportunities for outs on the basepaths, and also the middle bases would not be so far from the outfield. Instead of throwing directly to the infield, the outfielders would have a relay system where one outfielder would throw to another that was farther in, and perhaps one more, before reaching the infield. That might be pretty cool. I think it's not easy to run fast on the Moon. On the Earth, a runner's feet are pushing against the ground many times each second. On the Moon, the runner is taking big leaps. They may only get in one-sixth as many steps over the same distance, which would give them much less opportunity to convert muscle energy into velocity. (Somewhat countervailing, though: no air resistance.) Runners would have to train specially to be able to leap accurately to the bases. Under standard rules, a runner who overshoots the base will land off the basepaths and be automatically out. So we might expect to see the runner bounding toward first base. Then one of the thirty or so far-left fielders would get the ball, relay it to the middle-left fielder and then the near-left fielder who would make the throw back to first. The throw would be inaccurate because it has to traverse a very large infield, and the first baseman would have to go chasing after it and pick it up from foul territory. He can't get back to first base quickly enough, but that's okay, the pitcher has bounded over from the mound and is waiting near first base to make the force play. Maybe the runner isn't there yet because one of his leaps was too long and to take another he has to jump high into the air and come down again. It would work better than Quiddich, anyway. [Other articles in category /games] permanent link Sat, 15 Mar 2025
Hangeul sign-engraving machine
Last summer I was privileged to visit the glorious Letterpress Museum in Paju Book City, where I spent several hours and took a collection of photos that are probably not of interest to anyone but letterpress geeks, and perhaps not even to them. Looking back at the photos it's not always clear to me why I took each one. But some of them I can remember. For example, this one: This is not exactly letterpress. It is a device for engraving lettered signs on thin strips of metal or perhaps plastic. Happily I don't have to spend too much time explaining this because Marcin Wichary has just published an extensively-illustrated article about the Latin-script version. The only thing different about this one is the fonts, which are for writing Korean in Hangeul script rather than English in Latin script. (Here's my real-quick summary. There is no ink. A stylus goes into the grooves of those brass templates. The stylus is attached with a pantograph to a router bit that rests on the object that the operator wants to engrave. When operator moves the stylus in the template grooves, the router bit follows their motions and engraves matching grooves in the target object. By adjusting the pantograph, one can engrave letters that are larger or smaller than the templates.) Hangeul has an alphabet of 24 letters, but there's a difficulty in adapting this engraving technique for written Hangeul: The letters aren't written in a simple horizontal row as European languages are. Instead, they are grouped into syllables of two or three letters. For example, consider the consider the Korean word “문어”, pronounced (roughly) "moon-aw". which means “octopus”. This is made up of five letters ㅁㅜㄴㅇㅓ, but as you see they are arranged in two syllables 문 ("moon") and 어 ("aw"). So instead of twenty-four kinds of templates, one for each letter, the Korean set needs one for every possible syllable, and there are thousands of possible syllables. Unicode gets around this by… sorry, Unicode doesn't get around it, they just allocate eleven thousand codepoints, one for each possible syllable. But for this engraving device, it would be prohibitively expensive to make eleven thousand little templates, then another eleven thousand spares, and impractical to sort and manage them in the shop. Instead there is a clever solution. Take a look at just one of these templates: This is not a Hangeul syllable. Rather, it is five. The upper-right letter in the syllable is the vowel, and the template allows the operator to engrave any of the five vowels ㅣㅓㅏㅕㅑ to produce the syllables 잉 엉 앙 영 양 pronounced respectively "ing", "ông", "ang", "yông", and "yang". Similarly this one can produce six different syllables: The upper-left part can be used to engrave either of the consonants ㅅ or ㅈ and the upper-right part can be used to engrave any of the vowels ㅣㅓㅏ, to produce the combined set 싱 성 상 징 정 장. I'm not sure why this template doesn't also enable vowels ㅕㅑ as the other one did. In the picture at top you can see that while the third template This ingenious mechanism cuts down the required number of templates by perhaps a factor of five, from ten boxes to two. Addendum 20250325A great many of the 11,000 Unicode codepoints are for seldom-used syllables that contain four or even five letters, such as 둻. I studied Korean for a while and I think I learned only one with with more than three letters in a syllable: 닭 means “chicken”. I don't see templates for these syllables in any of my photographs, which probaby accounts for much of the great reduction in templates from the 11,000 possible syllables. But there must have been some way to engrave the syllables with the machine. Maybe there was a template that had a small four small ㄷsymbols, one in each of the four corners of the template, and another with four ㄹ symbols, and so on? Then the operator could have composed 닭out of bits from four different templates. [Other articles in category /IT] permanent link Fri, 07 Mar 2025
Claude and Merle Miller let me down
ClaudeMy relationship with Claude has its ups and downs, and I'm still trying to figure out what to use it for and what not. It's great at rewriting my job application cover letters to sound less like an awkward nerd. Last week I was in the supermarket and decided to try asking it something I've been wondering for a long time:
I thought Claude might do well with this. I had had a conversation with it a while back about Pixies songs, which I was satisfied with. But this time Claude let me down:
(I thought: What? Am I supposed to believe that
is about a doll?)
Claude just flubbed over and over. I wonder if the grammatical error in “Mary Gray Staples, who was the name of …” is a kind of a tell? Perhaps Claude is fabricating, by stitching together parts of two unrelated sentences that it read somewhere, one with “Mary Gray Staples, who was…” and the other “… was the name of…”? Probably it's not that simple, but the grammatical error is striking. Anyway, this was very annoying because I tend to remember things like this long past the time when I remember where I heard them. Ten years from now I might remember that Anne Sexton once had a doll with a very weird name. Merle MillerA while back I read Merle Miller's book Plain Speaking. It's an edited digest of a series of interviews Miller did with former President Truman in 1962, at his home in Independence, Missouri. The interviews were originally intended to be for a TV series, but when that fell through Miller turned them into a book. In many ways it's a really good book. I enjoyed it a lot, read it at least twice, and a good deal of it stuck in my head. But I can't recommend it, because it has a terrible flaw. There have been credible accusations that Miller changed some of the things that Truman said, embellished or rephrased many others, that he tarted up Truman's language, and that he made up some conversations entirely. So now whenever I remember something that I think Truman said, I have to stop and try to remember if it was from Miller. Did Truman really say that it was the worst thing in the world when records were destroyed? I'm sure I read it in Miller, so, uhh… maybe? Miller recounts a discussion in which Truman says he is pretty sure that President Grant had never read the Constitution. Later, Miller says, he asked Truman if he thought that Nixon had read the Constitution, and reports that Truman's reply was:
Great story! I have often wanted to repeat it. But I don't, because for all I know it never happened. (I've often thought of this, in years past, and whatever Nixon's faults you could at least wonder what the answer was. Nobody would need to ask this about the current guy, because the answer is so clear.) Miller, quotes Truman's remarks about Supreme Court Justice Tom Clark, “It isn't so much that he's a bad man. It's just that he's such a dumb son of a bitch.” Did Truman actually say that? Did he just imply it? Did he say anything like it? Uhhh… maybe? There's a fun anecdote about the White House butler learning to make an Old-fashioned cocktail in the way the Trumans preferred. (The usual recipe involves whiskey, sugar, fresh fruit, and bitters.) After several attempts the butler converged on the Trumans' preferred recipe, of mostly straight bourbon. Hmm, is that something I heard from Merle Miller? I don't remember. There's a famous story about how Paul Hume, music critic for the Washington Post, savaged an performance of Truman's daughter Margaret, and how Truman sent him an infamous letter, very un-presidential, that supposedly contained the paragraph:
Miller reports that he asked Truman about this, and Truman's blunt response: “I said I'd kick his nuts out.” Or so claims Miller, anyway. I've read Truman's memoirs. Volume I, about the immediate postwar years, is fascinating; Volume II is much less so. They contain many detailed accounts of the intransigence of the Soviets and their foreign minister Vyacheslav Molotov, namesake of the Molotov Cocktail. Probably 95% of what I remember Truman saying is from those memoirs, direct from Truman himself. But some of it must be from Plain Speaking. And I don't know any longer which 5% it is. As they say, an ice cream sundae with a turd in it isn't 95% ice cream, it's 100% shit. Merle Miller shit in the ice cream sundae of my years of reading of Truman and the Truman administrations. Now Claude has done the same. And if I let it, Claude will keep doing it to me. Claude caga en la leche. AddendumThe Truman Library now has the recordings of those interviews available online. I could conceivably listen to them all and find out for myself which things went as Miller said. So there may yet be a happy ending, thanks to the Wonders of the Internet! I dream of someday going through those interviews and producing an annotated edition of Plain Speaking. [Other articles in category /tech/gpt] permanent link Thu, 06 Mar 2025Around here, these metal things are commonly found on streetside utility poles, attached maybe a meter off the ground. ![]() Metal reflector
![]() Plastic reflector
When I first noticed one of these I said “I wonder what the holes are for. Maybe to make it more visible? And what do they do with all the leftover rectangles after they've made one?” I eventually got a better idea: The little metal rectangles are the primary product, and after they have been die-cut out of the metal sheet, there is this waste material left over with all the holes. Instead of throwing it away someone nails it to a utility pole to make the pole easier to see at night. I felt a bit silly that my first idea had been exactly backwards. I later learned that only the older ones are made of sheet metal. Newer ones are made of some sort of plastic, maybe polyethylene or vinyl or something, about the same thickness. They look pretty much the same. I can only tell them apart by feeling them. Still I wondered what the little rectangles had been used for. It turns out that the purpose is this: That's according to an old Philadelphia Inquirer article, Why yellow grids are on some Philly-area utility poles. (Patricia Madej, Aug. 31, 2019.) But I measured them to make sure. They matched.
Jay, my friend, your wife is smarter than you are. Listen to her. The article also tells us that the rectangular leftover is called a “grid reflector”. With a little more research I learned that one manufacturer of grid reflectors is Almetek. They cost $3.50 each. Pricey, for something they would have had to throw away. (Here's the old South Philly Review article that put me on to Almetek.) What kicked off this article was that I was walking around and I saw this similar reflector grid, which felt to me like it was a bit of a farce, like a teenager sneaking into a bar wearing a fake mustache: Hey, those aren't holes! When I saw this one I wondered for a moment if I was suffering some sort of mental collapse, or if none of the others had had real holes either. But no, they had, and this one really did have fake holes. (Also, it has been installed sideways. Normally they are oriented as the two above.) This isn't the first time I have written about ID numbers on utility poles hereabouts. [Other articles in category /misc] permanent link Sat, 01 Mar 2025[ Content warning: angry, contemptuous ranting that accomplishes nothing. ] I didn't really know who Jonathan Chait was until last week when I unfortunately read this essay of his (from February 2016) on “Why Liberals Should Support a Trump Republican Nomination”. I've said a lot of dumb things in my life but I don't think I've ever been as wrong about anything as Chait was about this. I sure hope I haven't. But if I do ever find out I had been this wrong about something, I would want to retire to a cave or a mountaintop or something. “Hey, remember Dominus? Whatever happened to him, anyway?” “Oh, he said he was going away to cleanse himself of error, and might not be back for a long time.” And yet this guy is still shamelessly writing. And why not? Editors are still buying his essays and maybe people are even still reading them. Why? You'd think that people would look at this essay and say “yeah, that's enough Chait for me, thanks, next time I need an opinion I'll try someone else.” I get it, nobody's right all the time. Whenever you read anyone's essay you're taking a risk, like rolling a die. Sometimes the die rolls high, sometimes it rolls low, and some dice might have higher numbers to begin with. I've usually been well-served by Daniel Dennett's dice, and Robertson Davies'. But here people have an opportunity to toss a totally unknown die that they haven't tried before but that most likely rolls numbers from 1 to 6, and instead they toss the Jonathan Chait die when they know it has at least one side with a -1000.
I don't think anyone could have predicted the extent of the current fiasco, but I do think it should not have been hard to predict, in 2016, that liberals should not, in fact, have supported a Trump Republican nomination. Anyone can be wrong, even the wise cannot see all ends. But I think this one was maybe not so hard to see. Chait spends a lot of time comparing Trump with Arnold Schwartzenegger: both nominally conservative, both inexperienced in government, both assholes. I think the part that Chait ignored was that by 2016 — no, scratch that, by 1990 — it was perfectly clear that Trump was a liar, a thief, a racist, and a deadbeat, and that he had no respect for law or truth or ethics or anything other than his own convenience of the moment. (Here are just two examples. More recently, his ridiculous years-long insistence that Barack Obama was born in Kenya. And earlier, his equally ridiculous lies around his destruction of the Bonwit Teller building.) In that old essay I said:
I looked around a little to see if Jonathan Chait had written an essay titled “I was wrong, I was so, so wrong, I just couldn't have been wronger” but I didn't find one and I also didn't find any recent essays that said anything like “here's why I think this new essay is more reliable than that embarrassing Trump one I wrote for The New Yorker in 2016.” I don't understand how Chait still has a job after writing this essay. Why isn't he selling shoes? How does a writer come back from this? Isn't there some charitable society for the protection for the public that could pay to have someone follow Chait around, quoting out loud from this essay, as a warning to everyone he meets for the rest of his life? It least now I've been warned. Now when I read “Jonathan Chait said recently…” I'll remember: “Oh, you mean Jonathan ‘Why Liberals Should Support a Trump Republican Nomination’ Chait! Thanks, I'll pass.” [Other articles in category /politics] permanent link |