Links for October 24

Starships, LLMs, Urban Legends, Comics, Novels

Thanks for reading! If you’d like to support Bullfish Hole, you can become a subscriber with the button below. You can also leave a one-time tip at this Stripe link.

Now onto some interesting items from around the interwebs.

Above and Beyond

You’ve probably heard about SpaceX catching the enormous Starship booster rocket with a pair of giant mechanical chopsticks. I watched the launch and catch live, and on through the reentry and soft water landing of Starship itself. If you haven’t seen it yet, the full video of the test flight is on Youtube. This shorter video focuses just on the launch and catch. There’s probably a million other edits, clips, and photos out there.

Peter Hague at Planetocracy writes about the catch as a kind of inverse Challenger disaster, in which nature defeated negative PR by showing that that the crazy idea actually worked.

At Quillette, Hague writes about why the successful test flight and catch is a big deal:

Mass is the key problem of spaceflight…. In the modern commercial era, the most important measure is cost per kg—and even with the Falcon 9 and Falcon Heavy substantially reducing costs, each kilogramme of matter sent into low Earth orbit still costs over $1,000 USD. What Starship promises is, depending on which estimates you believe, a per kilogramme cost in the range of $10–$100 USD.

The point of catching the rockets with a tower is to save mass by ending the need for heavy landing legs on both the Starship and its booster. As Musk said in his conversation with Dan Carlin on the Hardcore History Addendum podcast, “mass costs you mass.” More mass on one part of the structure means you need bigger engines, more fuel, and so forth. All of that subtracts from the payload the rocket can carry out of Earth’s gravity well.

Hague continues:

SpaceX is several years late relative to its initial timeline, and the clock is ticking for the company’s contractual commitment to NASA to deliver a human lander for the Artemis missions; and for Elon Musk’s vision of sending humans to Mars in his lifetime. Musk is demanding that once the rocket is ready, its launch rate be rapidly increased—to up to 1,000 ships per year by 2028. If SpaceX employees achieve even a fraction of the target they have been set, it will transform life on this planet and beyond.

If all that seems far-fetched, consider what the company has already achieved: The same week as the successful catch, they launched a NASA probe to Europa, sent up two rocket-loads of their Starlink satellites, and another load of satellites for an internet competitor. Falcon 9 just celebrated 100 missions flown so far this year, compared to the 135 flown by the space shuttle in 30 years of service. Oh, and while I was writing this, one of their Dragon capsules splashed down, safely returning astronauts and cosmonauts from the space station.

I’m old enough to remember a decade ago when rockets landing and taking off again wasn’t just the stuff of science fiction, but the stuff of outdated science fiction — goofily optimistic 1950s zeerust. Now it happens nearly weekly.

The cool thing about routinizing space travel and bringing down the cost to orbit is that it opens up possibilities that might not have even been invented yet. One wonders how many people working in space-adjacent industries are suffering from culture lag, working on mental models based on the costs ten times as high.

For example, physicist Casey Handmer in Palladium discusses the limitations on the James Webb Space Telescope and calls for something much more audacious:

Why does this help telescopes? One of the reasons JWST was so expensive was that in addition to being exquisitely precise, robotically operated in a pitiless vacuum, and parts of it at cryogenically cold temperatures, it also had to be feather-light and capable of folding up like origami to fit within the Ariane V payload fairing i.e. nose cone. Starship has mass, volume, and launch capacity to spare, relaxing these engineering constraints and significantly simplifying design.

He goes on to propose using the lift and launch cadence of Starship to orbit an army of mirrors that can self-assemble into one larger telescopic mirror:

This mirror is self-assembled in space from thousands of 8 m hexagonal segments, each an independent free-flying satellite and derived from the Starlink satellite design to keep costs and mass low. A self-assembling space mirror has been studied in detail with the AAReST concept, but the Monster Scope takes it to its logical extreme.

He claims the Monster Scope would have a resolution that would allow us to actually look at the surface of planets in other systems.

On X, John Carter disagrees that the planned scope could actually have this resolution, but notes it would allow us to identify more and smaller exoplanets, do analyses of their atmospheres, and identify their moons.

Lawyer Dog adds: “This is glorious, and suggests (not begs!) the question - what commercial applications might a scaled-down version of this have? E.g. for asteroid detection/ analysis, for later commercial exploitation.”

I lost the reference, but I saw someone else claim that even if we just send up another regular space telescope, Starship can carry one with a larger mirror than JWST without necessitating all the crazy folding, meaning it can be made more quickly and cheaply.

As for all else in space, here is video from NASA’s New Horizons flyover of Pluto, a planet that for most of my life was known as a little blurry dot. Now we can see the ice mountains clearly. But, oddly enough, we can’t see the black citadels of the Mi-Go.

Spooky Stuff

Yakubian Ape relates the modern legend of Polybius, a strangely addictive arcade game that debuted in 1981.

Another account - this one from the purported owner of an arcade who apparently only agreed to speak under the condition of anonymity - claimed that players would often try and stay after closing to play longer, some begging, some even bargaining, offering to pay what little cash they had just for some extra time with Polybius.

Soon the addicted players reported weird symptoms like hallucinations, amnesia, and missing time. Then, after a month, the mysterious games were pulled from locations, never to be seen again.

Oddly enough, those who recall playing the game can’t quite describe what the game was like. The company that made it is also shrouded in secrecy:

Though it seems that these men with the company would mostly come to the arcades after closing, there’s more than one report from patrons that they were seen going about their business during operating hours. By all accounts, the men were always dressed in black suits, black coats, and black hats.

I like that humans update our spooky stories to match current technologies. See also “Ghosts of the Digital Dimension” at After the Hour of Decision.

But we still get new incarnations of traditional monsters-in-the-woods. See, for instance, the Not-Deer of Appalachia, a creature that looks much like a deer but “there is an uncanny intelligence behind their eyes (which are often forward-facing the way a predators seem to be), and they have joints that bend in weird ways.” They sometimes move in an unsettling and jerky manner.

From an X user:

i have seen the not-deer btw, i was out for a walk at my college at dusk and saw a herd of does walking across the train tracks and two of them walked different than the others and their eyes were in the wrong place.

I mentioned some other legendary monsters in my post on teaching “American Folk Tales” at a local homeschool co-op. In addition to the famous Mothman and slightly less famous Flatwoods Monster, West Virginia also boasts the Grafton Monster, and sightings of monster birds and white-furred beasts.

Here’s a list of monsters of Central Pennylsvania, including the squonk and the Albatwitch.

I don’t recall any cryptids from the Eastern Shore, other than sightings of Chessie on the bayside. But I know the Eastern Shore has its share of ghost stories. My own great-grandfather on Chincoteague Island had two firsthand accounts — one about being approached by a recently deceased friend on a dark road at night, and another of seeing a headless man walking across the marsh on a misty morning.

When Life Gives you LLMs…

I’ve seen all this talk of LLMs helping productivity, and so I mess with them periodically, hoping to get in on the action. The results so far are mostly disappointing. But that might be because I’ve been trying them with domains where I’m already an expert, or at least have very particular standards that generic pablum isn’t going to meet.

My first real LLM success came from something unfamiliar: I know practically nothing about coding but asked Claude AI to walk me through building and customizing a data scraper for a survey project. It told me what programs to download, gave the basic scraper code, and told how to inspect the web page code and customize the scraper. Every time something went wrong, I could just feed in the error message and get an explanation of the error and suggestion for fixing it. Eventually I got things working!

I still can’t really say I’ve coded in Python, but at least I’m more familiar with it than before. I certainly learned far more than if I’d bought one of those Python books to sit on my shelf until “things calmed down a little” and a I had time to digest it.

I’ve also found a potential use for Google Notebook: using the podcast feature to get audio summaries of stuff I don’t have time to read but might get the gist of by listening to the two AI characters discuss it while I drive or do dishes. Important that it’s only “get the gist of it,” though — when I put one of my own articles into this feature, it took three minutes before it gave an incorrect example of a concept. (This is what I mean by not finding AI useful for things I already know a lot about.)

Back to Claude: The newest update allows it to actually take over and use your computer. Video tutorial here, though I myself haven’t yet tried it.

At One Useful Thing, Ethan Mollick describes the potential and frustrations of something close to being a deployable AI agent:

On the powerful side, Claude was able to handle a real-world example of a game in the wild, develop a long-term strategy, and execute on it. It was flexible in the face of most errors, and persistent. It did clever things like A/B testing. And most importantly, it just did the work, operating for nearly an hour without interruption.

On the weak side, you can see the fragility of current agents. LLMs can end up chasing their own tail or being stubborn, and you could see both at work. Even more importantly, while the AI was quite robust to many forms of error, it just took one (getting pricing wrong) to send it down a path that made it waste considerable time.

Zvi Mowshowitz at Don’t Worry About the Vase also discusses the new Claude AI updates here and as well as in this AI news roundup.

The roundup also contains some more ominous bits of AI news.

There was a post on Twitter of an AI deepfake video, making false accusations against Tim Walz, that got over 5 million views before getting removed. There continues to be a surprisingly low number of such fakes, but it’s happening

The Zvi also relates the recent legal complaint by a family whose teenage son committed suicide after becoming emotionally involved with an AI character. His final exchange on Earth before shooting himself was with the AI persona “Dany,” saying he loved her and was “coming home” to her.

Comics and Novels

I backed Graham Nolan’s crowdfunded Return to Monster Island comic and was pleased to recently get my copy in the mail. Monster Island and Return to Monster Island are adventure stories with a retro feel — not just in that they’re set sometime during the Cold War, but there’s a kind of earnest wholesomeness to the characters and dialogue, and the art has strong lines and a clean look. The whole thing just screams “fun,” and I imagine the big, bold monsters were fun to draw.

Nolan’s got an open campaign for the third book in the series, Escape from Monster Island. You can buy the entire trilogy as one of the perks. Recommend.

I bought the Monster Island books for myself, but my 5-year-old son saw them on the dresser and his eyes locked on the word “monster” and he wanted in. As I said they’re pretty wholesome — there’s no cussing or gore, and the romance mostly goes over his head — so I let him read them, and he got super into it.

Other comics he’s into come from the recommendations of Chuck Dixon. The Marsupalami’s Tail, by Franquin and Batem, is the first in their series about the prehensile-tailed jungle animal. It’s a European comic that isn’t solely aimed at kids but is kid-friendly — they won’t get all the wordplay, but they’ll like the cute animals and slapstick hijinks.

He also really likes two collections of Carl Barks’s famous work on the Disney Ducks: Christmas at Bear Mountain, one of the Donald Duck collections, and The Seven Cities of Gold, one of the Scrooge McDuck collections. The Donald stories are comedies of error revolving around the temperamental duck causing his own problems, while the Scrooge stories are high adventures leavened with comedy.

Speaking of Chuck Dixon: They’re not comics, but I’ve been consuming his Levon Cade novels like popcorn this fall.

The first one, Levon’s Trade, is all you’d expect in a vigilante justice thriller from a guy who used to write The Punisher: Rough loner works his way through the underworld hierarchy, killing as he goes. There’s a great sense of both him and the bad guys being in over their heads as each side realizes the scale of the trouble they’ve started.

The second one, Levon’s Night, is a slow build as the author tries to engineer a way for our protagonist to get in trouble again. Once he does, it’s a lot like Die Hard.

The next few books in the series are a linear progression, each story dealing with the fallout of the one before. I hesitated before picking up the third, Levon’s Ride, as there’s only so many times one can read about an implacable man eliminating bad guys with double-taps. But Dixon is smart enough to mix things up, and I was pleasantly surprised that the third book dialed back the violence and introduced some humor. That was enough to keep me around for further sequels, as our protagonist goes from fugitive to solider to prisoner.

All are short and fast-paced. It’s not high art, but it doesn’t have to be.

Thanks for reading!

Substacks cited above:

Don't Worry About the Vase

;

Palladium Editors

;

Planetocracy

;

One Useful Thing

;

After the Hour of Decision