Honkai: Star Rail

Well, I’ve played a bunch of Honkai Star Rail. tl;dr: Cute game, terrible leveling system.

I played some of their earlier big hit, Genshin Impact, which has the same cute anime characters, element-based combat, and gacha system, but is more of an open world. I kind of lost interest in it, but not because I ran into any barriers.

The pressing question is of course, what’s a Honkai? It’s 崩坏, which is Bēnghuài… for some reason Mihoyo, a Shanghai studio, likes to name its games for the West in Japanese. It means “collapse” or “breakdown.” Apparently the first Honkai game was a zombie shooter, while the previous one, Honkai Impact, is “post-apocalyptic.”

The Star Rail is pretty literal: there’s a train that goes around in space, you see, and you join its crew. That’s maybe the first clue that this is not hard sf. Also there are godlike beings called Aeons– one of them created the train. And things called Stellarons cause trouble for humans, which the crew is supposed to solve. All this of course gives you, the “Trailblazer”, an excuse to visit various planets, explore, and fight monsters.

There’s an introductory section on a space station where you wake up with Plot-Based Amnesia, and which is suffering a monster attack. You clean that up and then go to a planet named Jarilo VI, which is heavily based on Russia… pre-Revolution, apparently, as it’s divided into an aristocratic Overworld and a mining-based, oppressed Underworld.

The fighting is turn-based, which is awfully reminiscent of the 1990s, but hey, I liked the King’s Bounty games, and at least there’s no twitch mechanics. Different enemies have weaknesses against certain elements, and each of your characters specializes in one of those, so you want to pick the right team for the right enemy. It’s pretty satisfying if you can arrange the fight so the enemy doesn’t even get a turn. The four-character teams also make more sense here than in Genshin Impact since you see all four of them during combat.

I might add, don’t get this if you’re impatient. Turns take awhile, especially because your ultimates (which you will get several times per fight, for each character) each have a pretty long animation. (They’re also so bright that they hurt my eyes.)

Now, when I say you’d better get the team comp right, you’d better get the team comp right. Twice I ran into enemies (Svarog, then Gepard) who defeated me multiple times. Both have huge health bars and you absolutely need the characters that hit their weaknesses. If you lose a fight, the game will tell you the monster level, and if it says you’re too low, you probably are, and had better grind a bit.

Interspersed with the fights are little story missions. One charming bit is that your friends will send you text messages, often weird and baffling ones.

As you may expect, the major gotcha in this game is gacha. I haven’t had to buy anything, but to motivate buying things they’ve made the leveling system arcane and annoying. There is a plethora of special currencies or leveling-up items… separate ones for your own “Trailblazing Level”, your characters, and their “light cones” (repositories of extra stats). You can get completely blocked if you run out of one item or another, perhaps for just one character.

For the most part, there are ways to get to level 40, at least. I had to do pretty much every side mission. Plus, there are Golden Calyxes, little fights you can get into a limited number of times a day, but which reward you generously with the McGuffins you need to level up characters.

But now I’m stuck. You see, the next story quest has a level 42 monster that I can’t defeat. And there is literally no way to advance past level 40. You can’t just grind, you need to pass a certain story mission… one that’s after the quest I’m on. I can’t even buy stuff to advance, as it’s not gated to materials but to the story. This doesn’t seem intentional, or sane, so I filed a bug report, but for now at least there’s no use playing more.

(Also: in a normal game this would be fixable by turning the difficulty level down. But of course they won’t want you to be able to do that in a gacha game. So even though I can’t fix the problem by spending money, I’m still screwed by it being a gacha game.)

Edit: I got unstuck. First, some tweaks to my characters (leveling up all the accoutrements for everyone), I was able to beat Mr. Level 42. Second, it turns out that monster wasn’t necessary for the mission– it was only guarding a special chest. So I was able to move on, and even defeat Cocolia on the first try. Jarilo-VI saved, though for some reason there are a bunch of quests added after the main mission.


The Metabaron

Alejandro Jodorowsky is back with more meta-action. I can’t exactly recommend it, but it’s worth it if you’ve read the The Incal, the classic BD he wrote and Moebius illustrated in 1980-88. It comes in four fat volumes, representing eight BD albums (2015-22). There are various artists: Valentin Sécher, Niko Hendrichon, and Pete Woods– all excellent. Also Jodo just came up with the story; the actual writer is Jerry Frissen.

The Metabaron, as we’re frequently told, is the greatest warrior in the galaxy, and invincible or nearly so. The main villain is the Techno-Techno Empire. Oh, and the key plot point is a “magic white oil” named epyphite, the substance which powers anti-gravity, and is only found on one planet in the universe, Marmola, which happens to be where the caste of Metabarons started.

I’d might as well get to the bad news first: it’s small beer compared to The Incal. Since then Jodorowsky has apparently read too much Batman and Dune, lost his sense of humor, and given up on spirituality. The Metabaron is almost entirely about who can be the most badass, the Metabaron or whoever the Empire comes up with next. Spoiler: it’s the Metabaron. And half the time he’s handicapped by his own mind: he tends to lose interest in his own badassitude, though he always regains it just in time.

The Incal works in part because its center is an antihero, John DiFool. He has no powers and no real desires except for the occasional whiskey and robot prostitute– which means that he’s in real danger when something bad happens, as it continually does for six albums. The Metabaron, by contrast, is a superhero able to defeat entire Techno-fleets; the only thing that can challenge him is a similarly absurdly overpowered villain.

There was plenty of violence in The Incal, but Jodorowky’s imagination has become far darker…he spends quite a few pages detailing just how sadistic his villains are. They or he have a particular interest in amputation and body horror. (At the same time, there’s the odd trope that nothing says Gotho-Far-Future like using swords.)

That said, the first two books are pretty entertaining, not least because one of the Empire’s minions– Orne-8, the advisor to the Techno-Pope– keeps us guessing. In general, because all the main characters have to be total badasses, it’s their second-in-commands that have the more interesting character arcs. The later volumes are weaker… but that was true of The Incal too.

As in The Incal, women can participate in the badassery, but there’s also some romanticized gender essentialism going on– and this in a civilization where any organ can be regrown or replaced with metal. Maybe a reminder that Jodo is 94 by now.

For a space opera, the books have an amazing illiteracy about actual astronomy. At one point two galaxies collide– in response to events on Marmola– and this produces an immense shock wave which destroys “entire galaxies.” Either Jodo or Frissen seems to not quite know what a galaxy is. Big solid things a million miles out in space, I guess? Oh well, the endless interior of the Earth in The Incal didn’t make sense either.

I haven’t read the other post-Incal books, but the connection to The Incal is hard to trace at times. The Metabaron was one of John DiFool’s enemies and then companions, and the Techno-Technos resemble the technical caste of The Incal‘s Empire, but there’s no explanation of where the latter’s Imperoratrix came from, or where the various arhats and divinities have gone. The Incal didn’t mention epyphite and its Techno-Pope didn’t run the galaxy. But maybe this is all explained somewhere. (French Wikipedia suggests that the book takes place after The Incal, not before.)

As for the art… it’s extremely well done, like a set of paintings Yet it’s hard to compete with Moebius, and the artists too have read too much grimdark. I miss the playfulness and the sense of surprise in the original series.

Cipri and Yu

I was going to review Nino Cipri’s Defekt, but then I started reading Charles Yu’s How to Live Safely in a Science Fictional Universe, and they were so similar I had to treat them together.

To wit:

  • The protag is a young man in a dead-end job
  • They work for soulless but happy-faced corporations
  • They are almost entirely alone and don’t entirely understand why
  • They’re kind of failures, though they do their jobs well
  • Something paranormal happens to jolt them out of their rut
  • Nonetheless this brings more dangers, possibly lethal
  • There are frequent extracts from in-universe manuals
  • Some of the most compelling characters are not human
  • They both meet other versions of themselves
  • There’s a hint of escape, though nothing systemic

Defekt is set in a big retail store called LitenVärld (‘little world’), and my first reaction reading it was “Wow, is working at Ikea that bad?” The hero here is Derek, who’s a top sales associate, totally dedicated to LitenVärld, and a huge nebbish. He lives out back of the store in a storage container, and his only problem is that he seems to have a bloody cough. This will turn out to be the least of the horrors visited upon him over the next two days.

This bit is technically a spoiler, but it’s on the damn back cover of the book, so: LitenVärld allows sick days, but it doesn’t appreciate them, not one bit. Derek is reassigned to the inventory team, whose job is to find furniture and other merchandise that have mutated into monsters. Also his team is composed of alternate-universe versions of himself.

The tone reminds me a bit of Christopher Moore, whose Bloodsucking Fiends also spent a lot of time with dead-end retail workers. But apparently the times have darkened– Moore’s crew were jolly rogues compared to the dead-enders at LitenVärld. Cipri is very evocative of joyless work and abusive managers, and the bullying corporate pretense that it’s all great for everyone.

The book is short, but manages to tell a full satisfying story. Cipri is good at making a situation increasingly worse, and I didn’t expect some of the directions they took. (Cipri is nonbinary, and so are a couple of characters in the book.) The alternative-worlds idea is used for both plot and thematic reasons– naturally LitenVärld uses it in the crappiest way possible. In an afterword Cipri explains the genesis of Derek– they’ve worked with people like him, and wondered, “what was up with that guy?”

The book is a sequel to Finna, which I haven’t read, though I intend to.

In How to Live Safely in a Science Fictional Universe, the protagonist calls himself Charles Yu, and the paranormal bit here is time travel. Charles has his own time machine, the TM-31, with a depressed operating system named TAMMY, and a dog who doesn’t really exist but thinks he does. He spends all of his time– well, sometime else, except when he’s doing his job. His job involves stopping time paradoxes. People visit their past and want to change it, only they can’t, and if they insist they’ll get stuck in an alternative timeline.

Charles also has a dilemma: at one point he meets his future self. And shoots him. (Also not a spoiler, it’s in the first paragraph.) This creates a bit of a problem since this universe (it’s actually a Minor Universe) is deterministic. But Charles’s real problem turns out to be his family of origin. His father was a huge nerd, you see, who devoted years of his life to inventing a time machine, and seems to have disappeared with it.

The book is longer and more ambitious than Defekt, though I think less successful. It’s very immersive on the father’s hopes and dreams and failures, and there are a lot of cute bits– e.g. Charles’s boss is Phil, who’s a computer program but doesn’t know it, which makes things awkward.

There are also some metanarrative digressions… the title is not entirely ironic, apparently it really is a science fiction universe and all the characters know it. He doesn’t commit to the bit as in Scalzi’s Redshirts, but I don’t think the idea does the book any favors. Many authors have been fascinated by metanarrative; it probably comes with the territory, as they are acutely aware of their characters as both people and as creations. But I think it rarely works for the reader. Fortunately it’s not a deep part of the book.

I’m a sucker for time travel books, but the minimalism of this one is a bit dissatisfying. In Charles’s view, people would mainly use time machines to revisit their own past. That’s cool, but I think I’d prefer to go solve old historical puzzles, or go to the future and read the best comics.

SF is a balance: there has to be an SF idea, or it’s not SF at all; but like all fiction it need a plot and characters too. Cipri manages this balance: their book functions both as an adventure story, and as a satire on late capitalism. Yu I think doesn’t: the SF part is fun, but is neither developed as a concept or exciting as an adventure in itself. So it doesn’t support the heartfelt family story very well.

Take these gruntles with a grain of salt, because it usually isn’t a great idea to read the same kind of book one right after the other. (It wasn’t intentional!) These books might look quite different read in the opposite order!

With me in charge…

So, Mastodon user Samuel Wade and Twitter user Alexander Boyd highlighted this billboard in China, which contains a delightfully double-edged message attributed to Xi Jinping:

Boyd helpfully transcribes the message, left to right and then right to left and provides these translations:

Today’s efforts will bear fruit after I’m gone.

With me in charge, failure is guaranteed.

Backwards reading is not as esoteric in Chinese as it is in English: the traditional reading order for Chinese was top to down, right to left, and a title or slogan like this could be read right to left. I recently looked up the words on the Chinatown Gate in Chicago (because I wanted to build it in Minecraft), and the slogan there (天下为公 “the world is for everyone”) is read right to left. (Er, I mean what actually appears on the gate is 公为下天.)

I know just enough Mandarin (不多) to want to know exactly how this works. So here’s a gloss

功成 gōngchéng – achievements, outstanding work

不 bú – not

必 bì – must

在 zài – in

我 wǒ  – I, me

Google Translate suggests, reasonably, that this is more like “Success doesn’t have to depend on me.”  That matches the glosses. Backwards, it gives simply “I will fail”. Better, I think, would be “I must fail.” Zài in this context, before a verb, forms the progressive, e.g. 我在睡觉 “I am sleeping”.  I don’t think it combines well with bì ‘must’, but that’s more semantic than syntactic.

Note that 成功 chénggōng is interpreted as “success”. Mandarin often uses compounds where Old Chinese used single morphemes— gōngchéng is a compound of gōng “merit” and chéng “finished”. It happens to be lexicalized in both orders.



We just finished watching Wednesday, though we did it on Saturdays. It still works.

This is of course the Netflix TV show, which updates the movie, which updates the ’60s TV show, which updates the ’50s Charles Addams cartoons. That’s a lot of transformations, and I’m leaving out some (the ones I never saw).

I used to watch the TV show when I was a kid– it was already in re-runs. My wife prefers The Munsters, but I liked Addams Family. Both shows had the metajoke that the monster family is actually perfectly benign– if there’s any villainy it’s due to the ‘normal’ people outside. This is actually pretty sophisticated for an era where the typical TV comedy involved a talking horse or a heartwarming sheriff or both.

The 1991 movie went back to the far more impish cartoons. These roughly fall into two categories:

Jokes where the family is just being goth, or reversing normie values:

  • Grandmama telling a story: “Then the dragon gobbled up the handsome young prince and lived happily ever after”
  • “Are you unhappy, darling?” “Oh, yes, yes! Completely!”
  • The children get a giant iguana as a pet
  • Gomez makes a torture rack with his kids, explaining it’s more fun to make it yourself
  • Lurch serves a two-headed pig for dinner

And jokes where the family is hinted as actually being murderous:

  • Fester releasing a hawk while neighbors release pigeons
  • Morticia borrows cyanide from the neighboring witch
  • Puggsley has bricked up Wednesday in the furnace
  • The family drops boiling oil on carolers

Addams never actually showed the actual murders, and there were never any consequences– Wednesday was evidently not burned up, and just reappears in the next cartoon. This shtick is similar to his cartoon about a patent attorney trying out a client’s invention: “Death ray, fiddlesticks! Why, it doesn’t even slow them up!”

This grimmer humor is echoed in the ’90s movies, though those are still mostly about normies attempting to swindle the Addamses.

Now to the Netflix show Wednesday, which as you may know focuses on Wednesday. She’s now about 16, and she’s sent to a boarding school, Nevermore, which her parents went to decades ago. It’s built next to a very normie town, Jericho, whose main attracton is called Pilgrim World. Nevermore is intended as a school for “outcasts”, including werewolves, sirens, gorgons, psychics, and vampires. (None of the previous versions of the story leaned quite so hard into the supernatural.)

Wednesday is for some reason on the outs with her parents, and initially has trouble fitting into the school. But her attention is drawn by a murder mystery– there’s a monster murdering outcasts and normies alike– and she takes eight episodes to solve the mystery, make some friends, alienate said friends, and then realize at the end that she needs them.

All of this depends heavily on Wednesday the character, and her actor– fortunately this is Jenna Ortega, who does an amazing job.

Wednesday is a carefully balanced fantasy concoction. The character is so antisocial that she’s basically an antihero; but as in (say) Sin City, the narrative trick is to pit her against even worse villains. Then, for comic relief, you contrast her with likeable people who just want to get to know her: her bouncy colorful roommate Enid, the nice normie boy at the coffee shop Tyler, the appropriately goth artist Xavier. Plus a bunch of other relationships that go up and down as the plot dictates– with her parents, with the grumpy local sheriff, with the imperious principal.

I saw an interview with Christina Ricci (the 1991 Wednesday, and a Nevermore teacher in this version) where she praises this Wednesday as, more or less, a feminist superhero. She does what she wants, she refuses to let anyone put her down or stop her, she doesn’t give a fuck about traditional femininity or who disapproves of her. She is smart, witty, plays a mean cello, dances like no one is watching, and duels expertly with the rapier.

All true, but she also has the ability of the antihero to say and do things that the rest of us can’t– or that we’ll regret if we do. E.g. the series begins when she attacks the boys who have bullied her brother with piranhas. It’s so over the top that it’s more comic than horrible, but really, hasn’t anyone who’s been bullied wished they could fight back that way?

Wednesday: I know I’m stubborn, single-minded, and obsessive. But those are all traits of great writers.
(Thing makes a gesture)
Wednesday: Yes, and serial killers.

Morticia: We are not the ones who got you expelled. That boy’s family was going to file attempted murder charges. How would that have looked on your record?
Wednesday: Terrible. Everyone would know I failed to get the job done.

One of the characters calls her toxic, and on one level she is. She doesn’t have the psychopath’s or narcissist’s ability to be charming on demand. On the other hand– just like the cartoons– she is more bad attitude than bad behavior. E.g. when she starts hanging out with Eugene, the dweeby bee boy, she is dismissive as usual; but when he’s injured she regularly visits him in the hospital.

And honestly, the series leans hard into Wednesday being a teenage girl– far shorter than most of her antagonists, and disarmingly beautiful. Imagine a gender-reversed show called Puggsley. Would a story about a toxic male teenager be nearly as compelling? I think we’ve had all too many of those.

As many reviews have noted, Ortega rises over the limitations of her monotone and her perpetual unblinking scowl. There are moments of surprise, alarm, and concern… even a few moments of real joy.

A couple more performances stand out. Emma Myers would be too saccharine in a normal story, but she’s the perfect complement to Wednesday, and the sweetest moment in the story belongs to her and Wednesday. Victor Dorobantu only appears as a disembodied hand, but he’s perhaps the most winning character in the series. I liked Joy Sunday as the mean girl who has a change of heart of her own. Hunter Doohan as Tyler is kind of trapped playing a character on loan from a normal teen movie, but he does a good job anyway.

Gwendoline Christie plays Principal Weems, and she nails the surface politeness and underlying menace that she should have had as Lucifer in Sandman. I really wonder what happened there– she was stiff and not scary at all. She does great here.

There were a couple bits that didn’t quite make sense, but I can forgive them because the whole is executed so surely, and it ends well. They give Wednesday no less than three possible love interests, and the plot has to go into convolutions to make them move forward. I like the mystery angle, but also think it doesn’t quite work as a mystery, because it’s mostly a dance to keep everyone a possible suspect as long as possible. Some of the most fun bits of the show are actually unconnected to the plot: Wednesday’s cello solos, the canoe race, the school dance.

The show has been renewed, and I hope they can continue the magic. It’s going to be trickier than it sounds, because they’re probably going to have to undo some of the high note the first season ended on; and it’ll be frustrating if they undo Wednesday’s character arc.

Apparently the show has been very popular, which raises the question: why do normies like it? My guess is that it’s taken in different ways by different audiences. Anyone who feels alienated or oppressed can see themselves in Wednesday, or in the outcasts. But more mainstream audiences can take it as whimsical not-really-horror, like most of Tim Burton’s other work.

Smith vs. Devereaux

Do you have a take on last year’s row between Bret Devereaux and Noah Smith? I generally like both of those guys, but I think on that occasion, Smith had the better arguments. When you’re trying to make points about possible or likely futures based on your knowledge of the past, I’d say at least to some extent, you’re theorizing. 


I remember that quarrel, and didn’t have very strong opinions about it. Though re-reading those pieces, I find I have some opinions, which you’re welcome to.

The quarrel reminds me of a low point in syntax: linguists in the 1970s loved to lecture each other on philosophy of science, rather than, say, on linguistics. On the plus side, I learned something about Popper, Kuhn, and Feyerabend. But it generally felt to me like missing the point, as if Khrushchev and Nixon had decided to argue not about communism vs. capitalism but about Hegelianism. Maybe relevant in some deep sense, but it’s better when people argue about what they know and leave the philosophy to experts.

The argument, such as it is, is about what words like predictive and empirical mean. On a very narrow level Smith is right: historians don’t just say “what was”, but make pronouncements on what is likely to happen now. He then goes on to admit that this is perfectly OK! (“I am not claiming that these analogies are misplaced or that these predictions and recommendations are wrong.”) He’s not as good as explaining why well-placed analogies and correct predictions are bad.

And on just as narrow a level, Smith is wrong when he asks that “theories ought to be subject to empirical tests”, implying that historians don’t do this. Historians love to do it when they can. Devereaux has lots of good examples on why it’s hard: the data is bad or entirely lacking; historical analogies are always problematic; restricting yourself to the better-documented areas will absolutely distort the data. You can’t treat history, as Smith would like to— as political science. And for a lot of the questions we’d like to ask (how did Greek democracy hold up or fail? Why did Rome fall? Why did it fall when China kept regenerating itself?), the data is unavoidably human and messy, as far from billiard ball mechanics as you can possibly get. Yet we can learn even from lossy history.

Devereaux talks a lot about what historians are trying to do, but I think it could be explained more simply: history is fundamentally about what happened. When you can say why it happened, that’s great; when you can draw lessons, that’s fine, but the field is not at root a machine for creating what Smith calls “punditry” or what he calls “empirical testing procedures”. A historian would be thrilled to discover, say, a new ancient text giving the Persian side of the Greek wars, or an attestation of the Buddha two centuries earlier than we now have, or a full understanding of Quechua khipus, or better date for the settling of the Americas, even if no modern-day lessons could be drawn. The idea that all science is predictive is physics-worship… and even physics has given up on that, at the quantum level. (Biology is a science too, even if 9/10 of it is a catalog of weird animal and plant facts.) History is also a humanistic field, with ties to literature, psychology, and religion.

The discussion of a specific example— tyranny in ancient Greece— is particularly unedifying, because both thinkers push themselves into indefensible corners. Smith seems to want a database (“we should rigorously and systematically check the historical record”) as if only statistical methods create knowledge, and as if shoehorning messy human societies into mathematical data is not highly arbitrary and distorting. At the same time Devereaux is hobbled because he can’t quite admit that his present-day pronouncements are predictive and are punditry.

His argument should be that punditry is good! He acts as a pundit when he opines about Trump. Smith is a pundit— he’s play-acting at the beginning of the article when he talks about studying physics and economics; he isn’t either a physicist or an economist. I’m acting as a pundit in writing all this. Orwell was a pundit when he brilliantly analyzed European politics in the 1940s. Punditry obviously is fallible, especially if you compare it to physics. It’s not “academic knowledge”— though that is fallible too.

You mention theorizing; here again Smith is narrowly right— there are historical theories— but it doesn’t mean what he thinks it means. This is the least interesting part of the exchange, to my mind: there is almost never any real gain in arguing about what is or isn’t “a theory.”

Nona the Ninth

To my surprise, Nona was available at the library. I thought I wouldn’t get my hands on it for a year. This is, of course, the third book in the Locked Tomb series by Tamsyn Muir.

Possible spoilers for the Gideon the Ninth and Harrow the Ninth below so go read them already.

This one switches things up a bit, in that it’s way less goth. The basic situation, hinted at at the end of HtN: someone in Harrow’s body is being kept safe by Palamedes Sextus, Camilla Hect, and Pyrrha. There are a lot of strange soul + body combinations in this book, starting with each of these folks. The soul in Harrow’s body may or may not be Harrow– if it is she doesn’t remember it, and goes by Nona. Palamedes (male, necromancer for the Sixth) and Camilla (female, his cavalier) are time-sharing a body, which happens to be Camilla’s. Pyrrha (female) was the cavalier absorbed by Gideon the First (male) an eon ago, and is now sole occupant of his body as Gideon was killed by a Resurrection Beast.

Hey, stuff like that happens, and all kinds of people can make a family. Roughly the first half of Nona is in fact how they get on together in their daily life. It’s told from Nona’s point of view, and she’s both childlike and very trusting– she loves almost everyone she knows. She couldn’t have picked better adoptive parents; unfortunately they’re living in a wartime city filled with refugees, as the Blood of Eden group is fighting the necromantic Houses. Nona is also an assistant teacher at an impromptu school run by some of the refugees, and gets to know the older kids in that group. This rank is only due to her apparent age: she can’t read or write, so she mostly helps the teachers.

In interspliced chapters, John– the Emperor, the Necromancer Divine, etc., etc.– reminisces about his past: how he went from a researcher to Emperor, with stops at Murdering Everyone and Resurrecting A Bunch of Them. This in effect does the worldbuilding that connects the baroque sci-fi-medieval-necromantic empire we learned about in the first two volumes to an Earth not long past our era. (A cute touch is that John is a New Zealander, like Muir. He favors Australia plus Oceania for his resurrection.)

I’ve read a couple of quite contradictory reviews, but I agree more with the positive one. The negative one was basically “why couldn’t we have Book One again”, and that’s never a good path for a reader. Muir has changed gears once again: Nona is sweet and loving where Harrow was acerbic and self-hating; we see the Empire from outside rather than inside; we only get glimpses of what the Emperor is up to. It’s clear by now that Muir likes to switch things up (HtN was a jarring change from GtN, too); also that she’s a master of voice. You could tell who’s speaking even if the names were removed. As for the outsider perspective, I think it’s necessary; it develops what we learned in HtN, and we really can’t evaluate John without knowing, not just how he behaves at a dinner party, but what his empire is like. (Hint: like any empire.)

John is about my favorite supervillain, because he’s evil, but not eeeevil. He’s not a Dark Lord, and almost everything he does makes sense. If his account here can be trusted (but nothing is as it first appears in Muir), his early crimes were well motivated, because of even worse assholes. He was even the most well-adjusted person on the Mithraeum, though that’s an extremely low bar. If you had to spend time with him, you probably wouldn’t sense a stench of psychopathy, just a great neediness. His great failing– shared with Blood of Eden– is an inability to think past patriarchal, hierarchical structures. (In a world where necromancy works, are pseudo-medieval noble houses the best way to organize it?)

The negative review also complained that Nona had very little agency. This is like complaining that Dr. Watson never solved a case. The whole point is to see things happening (and things do happen) from off to the side. It’s customary in sf to talk about world-ending crises from the viewpoint of the tiny number who can start or end them, but some crises (e.g. nuclear war or climate change, both referenced here) need to be understood from the viewpoint of the victims.

But it turns out Nona is not powerless at all; when she does act, it’s all the more important. If I understand the plot, Nona is the spirit of humanity that John destroyed and then remade; thus she is arguably human, but has never lived an ordinary human life. She is presumably the Alecto the fourth volume will be about.

I should note that you may or may not like Nona’s teenage friends. There’s maybe more about them than is strictly necessary, but they’re very important to Nona, and trigger a few plot points, and they’re part of the process of humanizing her.

In the middle of the book we see Gideon Nav return… under a new name, and in her own body, though it’s somewhat the worse for wear, as she’s dead. But as in ancient Hindu literature, death in Muir’s books is just a weird transition, a chance to change worlds or bodies or status. What’s curious is that Gideon comes off as kind of a dick here. She’s acerbic as usual, but she’s not the sweet friendly goth from the first book; she doesn’t seem attached to anyone or anything. I wonder if this is because we’re seeing her from the outside, or because she’s been corrupted by love-bombing from Ianthe and John, or because she’s pining for Harrow and can’t show it.

Overall, I liked the book a lot. It’s a welcome change of tone from the tortured-goth trauma of Harrowhark, and Muir manages to show us what the Empire amounts to for the commoners without losing her trademark thing of letting a few strong personalities at each other in a cramped space.

Now to wait for Alecto the Ninth, which should be a rip-roarer. (And come on, let’s get Harrow and Gideon to kiss, at least!)

Lucy Cooke on animal sex

I recently read Bitch: On the Female of the Species (2022), by Lucy Cooke. Great book, unfortunate title.

There is by now an entire bookshelf of books by women scientists and journalists examining sex and gender in animals, and gently skewering the bias of male scientists.  In brief: if you ever heard that males are dominant in nature, that males are naturally aggressive and unfaithful and interested in spreading sperm widely while females are demure, concerned only with offspring, and prefer a single mate… all that was widely believed well into the 20th century, but it’s pretty much all wrong.

Nature can be unkind to females… but the females fight back, and sometimes they’re unkind to males. And sometimes it’s just mondo weird out there.

Cooke’s book is organized by overall myth, and it largely consists of descriptions of one animal after another. This makes it easy to read (I can eat up wacky animal facts like popcorn) but hard to review. So rather than review the book I’m going to list a bunch of facts from it that caught my fancy.  There are a lot more in the book.

If you have a conworld, this book is a great resource for evolutionary oddities. Why not use some for your sentient species?

Ch. 1: You can’t tell a male by the “penis.”

• Moles have an enlarged citoris that looks identical to the male’s penis. Hyenas too.

• Among barklice, females are more aggressive, and have erectile ‘penis’ that inserts into a male ‘vagina’. Sperm still travels from the male into the female, through this connection.

• Spider monkey females have a long pendulous clitoris, always on display. Males tuck their penises away inside!

Testosterone is not a “male” hormone and estrogen a “female” one. Both sexes produces both, and both derive from progesterone. It’s all about the timing and relative amounts. Female hyenas produce a load of testosterone in their ovaries, which results in their ‘masculinized’ genitals.

There’s a gene called SRY, on the Y chromosome, that triggers male development— but it works in concert with 60 sex-determining genes, which exist in both sexes and most animals; they interact in very complicated ways to build either testes or ovaries. (In the 1950s Alfred Jost stated that “default” development was to produce a female, but this is now seen as naïve: it’s just as complicated to build an ovary as to build a testis.) Moles achieve their unusual anatomy with changes to just two genes.

The platypus has the same set of 60 genes, but no SRY gene— though it has no less than 10 sex chromosomes rather than our two. Females are XXXXXXXXXXX and males and XXXXXYYYYY. A different gene takes over the role of SRY.

For unknown reasons the human Y chromosome is losing about ten genes every million years, and has only 45 genes left. So we could lose the whole thing in 4.5 million years. (See chapter 10 for species without males.)

Birds and some reptiles don’t have our XY system at all; they have ZW for females and ZZ for males.

Turtles don’t rely on SRY at all: eggs incubated at 88° or higher become females, those incubated below 82° become males. There are all sorts of other environmental triggers used in various species.

Females, and estrogens, came first: at least 600 million years ago.  Males appeared about 300 million years ago. As they derived from females, it’s not surprising that a lot of male function depends on estrogen.

Ch. 2: Female choice in mating

Females were assumed to be “passive” in their mate choice, but this is not true at all. E.g. sage grouses participate in “leks”, displays where males dance to attract mates. It’s hard not to focus on the males, who are doing the hard work. But the females are actively watching, and responding to their feedback is key to a male actually succeeding.

Ch. 3: Monogamy, or more likely not

Females were assumed to be naturally monogamous, while it was in the interests of males to cheat. Not so. It turns out in 90% of bird species, the females play around. Social monogamy (i.e. raising offspring together) does not imply sexual monogamy.

It was once assumed that males philander because producing sperm is cheap.  Producing one sperm is cheap compared to an egg— but no male generates just one sperm at a time. In mammals, at least, the energy needed to produce one ejaculation is more than that needed to produce one egg.

Often sperm includes nutrients or other substances useful to the female— sometimes in the form of a ‘sperm packet’. This also make philandering a good strategy for the female!

Ch. 4: Eating males

You can’t get more of a picture of females in control than in species where eating the male is common. Note, this is an even better protein source than a sperm packet. Mantises are famous for this, but also many species of spiders. One spider keeper offers a tip for keeping males alive: make sure the female isn’t hungry. Some species have figured this out themselves, waiting to approach till they see the female eating. The nursery web spider takes an even safer approach: bondage. He ties up the female in silk before sex. (Don’t worry, she can free herself, but it gives him time to escape.)

Ch 5: All about genitals

Genitalia are amazingly diverse. And these days you can’t just look at penises, as biologists used to do— you have to look at vaginas.

Birds don’t normally have a penis, but some do. It’s likely that the lack of a penis is the innovation, not the opposite. Ducks have one, and often use it ruthlessly in forced copulations. But female ducks have a complex vagina, with dead end pouches. She seems to have control over these, and can divert an unwanted penis into the pouches. The result is that while 1/3 of duck matings are forced, only 2–5% of pregnancies result from forced matings. 

Also, both males and females ducks have seasonal genitalia. They get larger and more complex during mating season.

Some biologists have complained that the clitoris is useless— the female doesn’t need to feel pleasure to have sex. But, all vertebrates have a clitoris, which suggests it is useful. And we can easily find species where females don’t get pregnant without arousal— if nothing else, they may just reject the male. (This is easy for animals with cloacae, since then mating requires female cooperation.) There is some evidence that orgasm promotes conception in humans. Also, vaginal contractions may be necessary to get the sperm to the egg— it often can’t do it on its own. 

As a last resort, the egg itself may exercise some control over which sperm get to it. Eggs can secrete chemicals that act like a pathway, and sperm react to these differently, meaning that th egg is testing something.

Ch 6: Female care.

Only 1/10 of mammals exhibit direct male care. But 9/10 of birds do— probably tied to their lack of a penis and lactation.

For weird, look at the seahorse, of course. The female, er, inovulates the male, transferring eggs to him through an ovipositor. The male inseminates and feeds them in his pouch and eventually pushes them out in the water.

In mice, at least, caring for infants is controlled by one set of neurons, attacking them by another. Both sexes have both neuron types, and stimulating one produces the same behavior (caring or attacking) in either sex. Most animals have these neuron types too, including us.

Maternal care is by no means universal or automatic.  First-time monkey mothers can be very bad at it (mortality rates for firstborns are 60% higher than for laterborns). It may be adaptive to sometimes neglect the child, especially in very stressful or low-resource situations.  A kangaroo fleeing a predator sometimes ejects the near-independent joey from her pouch. This is bad news for the joey, but the mother survives, and she always has reserves: a fetus and a blastocyst held in waiting; they move to the next stage when the joey is lost.

Ch 7: female vs female. 

Another myth is that males do all the fighting. The topi are African antelopes; females fight each other for the chance to mate with the top males.

Male gorillas have small testes, usually a sign of little inter-male competition— the males dominate a harem. But the females compete to mate. Females may mate outside of estrus in order to drain his sperm; or high-rank females can harrass low-rank ones to prevent them from getting pregnant.

It was once thought that only male songbirds sing. In fact 71% of female songbirds do. It’s more complicated though: European and North American female songbirds sing less.  It seems female birds sing to compete with other females or to lure males. In Europe and North America, they are in the vicinity less because of migration; lacking time for competition, they concentrate instead on picking between males.

Ch 8 : Matriarchy

There are 90 species of lemur on Madagascar that are female dominant. The females frankly terrorize the males (and lower-ranking females)— all the more remarkable because they’re the same size. OTOH the males have evolved an ace: copulatory plugs, made of coagulated sperm; they don’t prevent further mating but have to be removed, and given the lemur’s very short mating period (sometimes just part of a day, once a year), it can help.

Ch 9: Menopause

The grandmother hypothesis is that active grandmothers make for healthier babies who wean earlier. This was demonstrated among the Hadza people in Tanzania.

Orcas are similar— they live to be over 60. And male offspring live with them till they die, basically.  Indeed, their death rates spike if their mother dies, especially if she’s post-menopause.

Ch 10: Doing without males

Laysian albatrosses, who nest near Hawaii, often live in lesbian pairs: 39 of 125 nests.  This may be because it’s a newish colony, and the females are more adventurous. Males tend to stick around where they were hatched.  This was discovered because there were two eggs in the nest!  But only one survives, because a bird can only brood one egg, and the partner has to go out foraging.

Or there’s the mourning gecko, also of Hawaii, which has no males at all. A gecko can reproduce all by itself— a trait shared with about 100 other vertebrates. Sex is seen as necessary for adapting to new environments, so it’s thought that asexual species are short-lived (100,000 years instead of several million). On the other hand, they can colonize a new area quickly as they can produce offspring so fast.   And double OTOH, some species (like the Ambystoma salamander) can grab genes from related species. It must work: the species is 5 million years old.

Some asexual species still have (lesbian) sex— because for them only sex stimulates egg production.

Ch 11: Fluid genders and genderfluids

Gender itself is fluid and nonbinary.  E.g. bluehead wrasse can be seen as having three genders: males, females, and females that turn into males. Curiously, the latter are far larger than the ‘other’ males and much more territorial. The small always-males tend to form coalitions to mate.  Different conditions favor each, so the mix remains.

Chalk bass can change sex pretty much at will. This allows them to take turns laying eggs, which spreads out the burden nicely.

Most sex-changing fish start out female, but the anemonefish at least goes male to female. Curiously the process starts in the brain— the gonads can take years to catch up.  A female-brained fish acts like a female and is recognized as such by other fish. In all of these fish species, the sex change is operational: you get functioning reproduction.

The hagfish is hermaphroditic and can self-fertilize, probably for good reason: it lives a very solitary life. It’s worked for 300 million years.

AI gets creepy

By now we’re used to AIs based on “deep learning”. They’ve moved far beyond the early and amusing stage of “put dog noses on everything“; they can create photorealistic images and now, very convincing prose.

It’s worth putting these things in context by starting with Markov machines. I explain how these work here and give access to some working generators. Here’s a sample, using a corpus of Alice in Wonderland plus H.P. Lovecraft:

she stretched herself up on tiptoe and peeped over the hill, with some scientific mechanism of greater weight and intricacy. finally, with no jury or judge, quite a commotion in the light of a lease, in 1804, the duchess’s arm that was sitting on the site he had built up an international reputation by the paris parliament and shut in a melancholy tone;” we’re all mad here. i’ll take no denial: we must have seen only a seething, dimly phosphorescent cloud of fungous.

Considering how simple the model is, I find these efforts a little troubling. Again, these are simply based on three-word patterns within the corpus fed to the program. Yet they reproduce the patterns of English, and hover on the edge of coherence. A Markov generator absolutely understands nothing, but maybe the human brain isn’t as amazing as we think it is.

Now consider the output of the ChatGPT program. Alex Dobrenko puts it through its paces here. Here’s an example of the sort of thing it can produce:

My parents missed the traditions from their hometown and wanted me to suffer with them. This, coupled with my teenage angst and need for freedom led to a huge conflict in my high school years. I ultimately got kicked out of their house for four months. When my mom found weed in my pocketbook, we took turns screaming at each other and throwing things for 14 hours (I wasn’t the only one who did drugs).

Now, I emphasize that like my Markov generator, this is ultimately based on a corpus— just an unimaginably large one. Wikipedia says it was trained on about 500 billion “tokens”— I’m not sure what that measures, exactly, but the inputs were basically all of the web, plus a bunch of books. All of Wikipedia is in there, making merely 3% of the input. A little bit of the magic drains out when you realize that the program is regurgitating what’s in its database. It “knows English” because it has a ginormous corpus of English to draw on.

The real magic is that you can ask ChatGPT to refine a text based on parameters you specify, as Dobrenko does in his essay. He tries asking it to make a couple of paragraphs of his “more intense”, “funnier,” or more “absurd and vulnerable.” He asks it to mimic George Costanza or David Sedaris.

Now, let’s be honest here, almost none of its efforts are very good. E.g. its attempt at Sedaris sounds nothing like Sedaris. But the remarkable bit is that its attempts are all coherent and apropos. As just one example, Dobrenko’s original text contains a complaint from his parents, which starts “You want to go sleep on someone else’s floor,” my parents would say. In one text the AI changes this to “Sasha, you want to go sleep at someone else’s house,” my folks would ask.

That’s not better than Alex’s wording, but it’s quite impressive as a paraphrase. It added in the correct vocative (“Sasha” was in the original text, but without an explicit statement that he was the son). Normally if someone can change “someone else’s floor” to “someone else’s house”, or “my parents” to “my folks”, we’d assume they’re an intelligent fellow human being. (Well, using “ask” for something that isn’t a question is a bit of a telltale.)

The one really amazing piece the AI comes up with is a pastiche of Dr. Seuss. It gets the gist of Dobrenko’s text and it rhymes. It does rhyme days and place, but jeez, how does it even know any rhymes, being unable to speak or hear?

Now, Dobrenko concludes that we don’t have to worry that writers will be replaced— yet. Someone on Mefi suggested that ChatGPT sounds like “a middling 8th grader.” I’d be a little more dismissive: it sounds like a smart 10th grader. (For non-US readers, we’re talking about 14 vs. 16 year olds.) What I mean is: a smart high schooler can write a pretty good pastiche— I certainly could at that age. I wrote a long bit of pseudo-Asimov featuring exquisitely dull accounts of Martian politics… I didn’t know anything about politics, but I sure thought I did. (Another Mefi commented that ChatGPT is a perfect impersonation of “a confident male with chronic answer syndrome.”) The thing is, 8th graders are less polished but also more likely to be goofily original.

Is there any type of writer AI could replace now or in the near future? Sure, precisely the ones that should be trundled onto Golgafrincham Ark B: SEO text writers, Russian disinformation producers, 419 scam letters, QAnon theorizers, web commenters, fundraiser bullet pointers, clickbait/listicle writers. That is, stuff that’s already supposed to be characterless, high-volume, and low-accuracy.

Some people have already brought up school essays— already the product of a huge plagiarism market. The problem here is detection, not production. Unfortunately, people will have to be cannier now, because AI will enable cheating. Faking an essay is definitely a task some people want to do; but it’s useless for the professor, because it’s not a test of what you’ve learned. Smart professors will have to respond by testing in a different way (and not just pretending that the Internet, and chatbots, don’t exist.)

Once the faddishness level drops, would you actually want to read ChatGPT doing news stories, or writing Wikipedia articles, or writing pop science or history books, or doing software documentation? No, because we actually want these things to be truthful, and regurgitated web text can’t do that. (If that’s not clear, look at the excerpt above, which is supposed to be a continuation of Dobrenko’s story. It does kind of continue the story, but it emphatically is not telling the truth! It’s just making up something plausible, which is not what you want in any kind of nonfiction.)

Would you perhaps read a novel written by ChatGPT or its successors? In a sense I expect we will— someone will do one as a stunt, and lots of people will be fooled. Here I expect there’s still a big gap: the best writers are interesting in a sense ChatGPT isn’t yet. They’re able to communicate human experience, because they’re humans, and they can be original in a way you just can’t be by relying on a corpus. Case in point: could a chatbot write Gideon the Ninth? In 2023, maybe, if that book is in its corpus. But that level of riotous newness is probably out of reach for now. I only read a few novels a year, so I can be very picky, and stick with humans as writers.

If you’re an AI researcher or something, I have to say: good work, simulating a 10th grader is actually pretty amazing. AI is far more than a toy at this point. Back when you only had things like Eliza and chess algorithms to point to, we could maintain that you were maybe 1% of the way to Real Intelligence. Now it feels more like halfway. And I’d add: maybe start thinking about ethics? You’ve created something that is most suited for doing evil things, like political disinformation and depriving graphic artists of a living. You’ve got the proof of concept down; maybe start worrying about how real people are going to abuse the processes you created.

Grammar of Sarroc

One of the things I’ve been doing as part of the Almea+400 project is filling in old holes in the world– especially projects that got started 20 or more years ago and never got done. One was a grammar of Sarroc, and that’s now done.

Sarroc is the language of Sarnáe, the country east of the Ctelm mountains. If you’ve seen maps of Almea in 3480 it was always part of Dhekhnam, but as you can see on the above map, that’s changed. The grammar is written from a point of view of 3650, so Sarroc is an interesting mixture of Dhekhnami, Kebreni, and Verdurian influences.  Plus I think it has a nice phonology, kind of Italianesque.