A positive hit on an fMRI would a) not indicate whether practicing or non-practicing, and more seriously b) have an 8% error rate, and thus remain less powerful a predictive tool than Kaworu fandom.
Not a complete list, I don't remember all the fiction, and some of them are either so specialized in focus to be irrelevant, or I just read on a bet and a review is besides the point (eg, Minotaur Milking Farm after it became a short-lived twitter meme, which beyond its obvious problems was also just bizarrely normie).
Conventional Books -
A Market of Dreams and Destiny: Alternate universe London Underground Gaiman-esque fantasy where everything has a price. Serviceable prose, decently interesting universe, but it needed several editing passes or maybe even a serious rewrite. The author has too many viewpoint characters and too little meat to each tone, the politics go beyond overt to the point of hilarious inconsistency, there weren't any real big payoffs or conclusions, and the central questions just didn't hook me much. I was kinda hoping for a something akin to Fable Of The Swan, so might expectations might have just been too high for what's ultimately just angsty slash, but I just came away feeling meh.
Icarus Series: Scifi series rolling around spacers that operate somewhere between couriers and smugglers, with some twists to that. I'd actually read Icarus Hunt a couple decades ago, but I'd filed it away as a Zahn one-off; stumbled across the reset of the series in a B&N and splurged. I think I prefer the characters from Hunt, since for goofy publishing reasons, Plot and subsequent stories focus on a different set of main characters, but each story still works great and a not-absolute-best-tier Zahn character is still a great character. Not my single favorite Zahn series -- I think the Conqueror's Trilogy just had a better central gimmick -- but well-executed and consistently clever and easily beats Blackcollar.
Hugo Award Novels: yes, I still get the packet, though it's harder and harder to justify paying for it. Mostly a lot of meh. A Sorceress Comes To Call is well-executed prose and nothing else; The Ministry of Time has a great idea it does absolutely nothing with, Someone You Can Build A Nest In is about as shallow as xenofiction gets. Alien Clay's the only one I'd really put a vote into, and that's just workable rather than deeply memorable -- it's far from Tchaikovsky's best.
Humble Tech Book Bundle: Computer Science the Fun Way: a bunch of No Starch Press compsci books. I'm mostly self-taught (and worse, self-taught in weird focuses), so these were a kinda interesting read from a formal programming perspective. They're all pretty reasonable for their subjects, but what those subjects actually were and how closely they related to computer science versus computer engineering was rough at times. Computer Graphics From Scratch is literal 'how you'd do things without a graphics API, which you will never do', while on the other extreme Data Structures the Fun Way was a distillation of the various 'how do B-trees do and why would you actually use them'. Only real big complaint I have within the content was Code Craft, which felt both very opinionated on what coders should be doing and simultaneously fell prey to many of the same problems that mauled the old Clean Code movement. On the other side of things, The Book of I2C and Introduction to Computer Organization were the sort of writing that seem like great dives through layers of abstraction that I was looking for. Dunno that I could recommend any of these on their sticker price, but the shop regularly offers deep discounts.
If Anyone Builds It, Everyone Dies: Yudkowsky's AI safetyism to the masses. Takeaway: Not Great, Bob. There's few things more frustrating than a good writer making bad arguments for a position you think is worthwhile, and Yudkowsky's written single one-off jokes (eg, Moore's Law Of Science Fiction) that were more compelling than this book, while providing only the least-plausible defenses against his proposed horror stories. The first half of the book is telling us how People Won't Just in the face of a superintelligent system, and then the second half gives a list of things that would work if People Would Just. Few people are going to buy into the assumptions, here, and those that do won't trust the conclusions. I've long complained that post-golden era LessWrong erred by emphasizing pivotal moments and extreme runaway, but the book comes across as less grounded in its speculative fiction than Friendship is Optimal, and where a reread of FiO leaves people going oh fuck, I keep coming away from sections thinking they'll be dated in years, if not months. Maybe there's something valuable here I've missed because I've read Yudkowsky since the 00s, but I just don't see it.
The AI Con: Takeaway: Even Worse, somehow. Where Yudkowsky seems likely to be dated quickly, The AI Con seems like it was dated before it even started writer. Computers Make Mistakes, IQ Doesn't Exist, Stochastic Parrots, Water Consumption, yada yada. I don't know if the authors originated this stuff or just absorbed it from those around them, but they sure as hell didn't care whether anything was meaningful or correct. Even where it should have been strong, on the economics and social impact, it still couldn't bother: "bossware" is up there with eyetracking for something that could become a dystopian hellscape, and deepfake ransoms are already making the world cyberpunk in the worst ways, and the authors can't actually sell those stories. Finally, the policy proposals are a grab-bag of the impossible and/or useless.
((There was some "how to deal with AI for normies book", too, but the part where I can't remember the name tells you about how valuable it was: think three-hundred-plus pages of 'ah, but LLMs do X' that were marginal predictions for Llama.))
The Unplugged Workshop : woodworking handtool-focused crafting guides. Hard to review this because, outside of watching a few youtubers, I just don't know the topic that well, and it's impractical enough that few people I know do either. Worse still, it's definitely an intermediate-level work, with only occasional accommodations for beginners like myself. Still, highly readable, well-organized, functional, and good project layouts, and the stuff I've tried so far has... well, not always turned out well, because I'm used to chisels as a de-riveting tool, but at least been more limited by my skill than that of the writer. Even if you're really into woodworking, a marginal buy, decent borrow.
Complete Guide To Sewing: I was told this was The Standard Book on the topic, and I can see why. Where a lot of 'beginners to experts' books tend either, this covers the whole spectrum from before your first stitch all the way to deep project work I couldn't even begin to understand. Not an enjoyable read by any means, but a good reference. The projects are my only big complaint: not only were they clearly marketing to a specific demographic that had zero overlap, these felt more like they were trying to help guide people who were already working from a pattern, leaving you to really guess at sizes and shapes. If you're doing anything sewing-wise more serious than patching torn pants, worth a buy... but get a used copy.
Comics -
Promethea: another one I'd read before, but that was borrowing it from a library, and now the local comic book store had the full series in trade edition. It's an Alan Moore comic, with all the benefits and costs that involves : wildly metafictional, deeply detailed, uncomfortably sexual, not quite as clever or as dedicated to its principles as the writer wanted it to be, and with an unfulfilling conclusion. Still, if you like Common Grounds or Astro City, it's worth looking at, and far more approachable and optimistic than the typical Moore comic.
Black Summer: second verse, same as the first: read it in the Obama era, and now could find it in full TPB. It's a very late GWOT story, and intensely political about it. The first pages have a superhero with blood-drenched hands lecturing the White House Press Corps about how 9/11 was planned and the Wars In the Middle East were just filling for corporate greed and the last two Presidential Elections were stolen, but he's Taken Care Of The Problem. The only men and women who can challenge him are the five(ish) surviving self-enhanced members who once worked with him... if they want to, and can get past a government presuming they are his allies. It'd run into political disfavor before it had even finished, as by 2008 concerns about a President's legitimacy had become much more popular in the wrong side of the aisle. In 2025, a man surrounded by floating eyes talking about stolen elections has rather different political valiance (and Ellis would get cancelled for other reasons). But where Ellis's other works in the same time period were either pointless gore porn (No Hero) or have a couple interesting scenes trying to cover up threadbare plot and nihilism (Supergod), Black Summer remains interesting enough to keep on the shelves, even if (or because) it'd never get written again. Not quite good enough to recommend as a buy -- the conclusion just doesn't feel earned, to the surprise of no one familiar with Ellis -- but might be worth a borrow.
Online Published -
Contention I and II: average Joe gets isekai'd into an alien or post-human world without even the clothes on his back, and gets to do the Primitive Technology speedrun. There's some really good plot seeds here -- the main character's obviously flawed in relevant ways without being an absolute asshole, there's a big driving question about the local precursors that's escalated really well, the not!magic system is useful and compelling without overriding a lot of the discovery and exploration bits. But it's also only two novellas into a story that seems built to go on for another two or three books minimum, the author only started on Book III in the tail end of last year (currently patreon-only), and that makes it hard to recommend.
Kitty Cat Kill Sat: basically Rimworld - with one of the more sadistic storyteller options - meets cozy fiction, where an (accidentally) self-uplifted domestic cat pits herself against all the plural of several post-apocalypses. My gold standard for xenofiction is Book Of Night With Moon, and Argus doesn't quite get that high. The plot's a little too meandering, the payoffs need to be set up better, and it needs an editing pass. Still a fun rampage, happy to have paid for it.
Reaper's Lottery and Executioner's Gambit: furry scifi, with very heavy raygun gothic and gumshoe inspiration. It's intensely furry, like the rest of the Hayven Celestia works, enough to probably be offputting to anyone else, but like Skinchange (cw: featureless furry nudity, scifi violence) it's got a pretty strong core underneath it.
On December 5th, three weeks after the government shutdown ended:
District Judge Telephone Conference set for 12/16/2025 2:00 PM before Judge Robert R Summerhays. Call-in instructions will be sent to counsel via electronic mail prior to the conference. Issued by Judge Robert R Summerhays on 12/5/2025. (crt,Taylor, L) (Entered: 12/05/2025)
But, hey, Broussard's still only 20, maybe the court will manage to find its ass with both hands and an atlas before literally no one can benefit from the final judgement... if the court's actually making a decision. Anyone want to bet whether a final judgement comes out before Christmas?
Possible, but I was direct-linking hardcore stuff under that, so dunno how effect it'd be.
It definitely works for some people, so don't read that as a complete nonstarter for everyone. It just doesn't work for me, even if the fantasy was still interesting, either being self-conscious while waking up or getting an elbow to the eyeball are kinda moodkillers.
Hm... Can you check your default sound device? Windows will normally try to only output to one audio port at a time. In Win10, left-click the Sound Volume control from the System Tray, click the name of the currently-selected audio output, and you should get a list of all available outputs. Win11 hides the interface a bit more, but it still exists. Look for one that says Speakers or AC97. You may have to restart applications or refresh web pages to have them recognize the new intended audio output, although modern apps are usually pretty good about responding near-instantly.
Very common problem where it defaults to HDMI or DisplayPort, especially when going to monitors without speakers or where the speakers default to no volume.
It's possible to get multiple outputs working simultaneously, but it's kinda jank and requires going pretty deep into the interface weeds.
[cw: almost certainly TMI oversharing. And most of the names here are either gay or male-leaning-bisexual]
How good is your imagination in this arena? Do you have a "mental spank bank" that surpasses the one on your hard drive?
I write smut, if not necessarily good smut, so my imagination's doing fine. There's a lot of options that aren't really featured in conventional porn at all, and despite the best efforts of the furry fandom to explore new domains of indescribably bizarre smut, there's a lot of stuff that either isn't available or only has a tiny number of not-great examples. Forget weird kinks: "Woman in jockstraps, het sex, go" seems vastly underserved and trivially easy to execute. I've got a handful of story irons in the fire, and while most of them are probably not desirable to the average poster here (
In the bedroom, I'm a lot more interested in pleasing others. That's fun and sometimes a good start for a stroll down memory lane, but those experiences don't always or even often transfer over well if I'm just trying to shut down for the night, so to speak.
On the flip side, there are things other people have produced that I either could or would not consider, or have done anywhere near the same quality, even solely within my imagination. There's a really good short form animation of people competing in a video game while also abusing the settings on remote vibrators, doesn't even show a hint of a genital, would never have considered it as a concept beforehand, really gets my motor running. Fek's Spellbound (cw: technically there's a pair of boobs, but it's almost all gay) and a lot of Ruaidri's animations should be comedic, but it's just so well-executed that it works. A lot of LawyerDog/Cantio, Braeburned, and Rick Griffin's stuff is both comedic and horny. Even for conventional human-on-human camera porn, there's a lot of imagery that's not something I'd have considered.
On the gripping hand (hurr hurr), a lot of scenarios presented by my own imagination or presented by others, aren't really practical or desirable; sometimes they're not considered because they're bad ideas. Sex while gaming is actually pretty unpleasant, quel surprise. Wake-up sex is an interesting fantasy, but no level of carefully-circumscribed consent beforehand can overcome a startle reflex. A lot of rough sex or degradation is just unpleasant, rather than kinky in a fun way. Ethical exhibitionism or enhibitionism isn't as time-consuming to set up as it sounds, but it's still ultimately very chilly and I don't have the physique for even ENM. Orientation play just doesn’t really work in situ; fantasies that depend on magic or superscience or both tautologically can't actually happen. Haven't tried the Braeburned orgies, but don't really want to, either.
*I don't know why she felt it necessary to give me special notice in this manner. I don't think we were very close, though we were both acclaimed by others as highly effective employees.
I wonder if she left because she saw herself at an ethical or liability crux, and thought you were competent such that you'd end up next in line after she retired, and didn't want you to feel like she was abandoned/shoving it on you or for you to not be aware of the scope of the problem. But I dunno the domain-specific context.
To be specific, Nicholas Decker is the one proving himself a real libertarian when it comes to ai-generated photos of children, or fucking almost-18 teenagers and being sucked by animals. To be fair to him, he's just being an edgelord; to be less naive, he's doing it in very stupid ways.
The headline writers are just not very good.
Hanania's just an obnoxious putz in other ways. He's pretty heavily abandoned the classical racism that the headline focuses on, but that's mostly by pivoting into a broader anti-lower-classism that's more tolerated so long as no one pins him down and makes him comment on lower-class <insert race here>. That doesn't make him wrong, but it does mean he's quite willing to play with news memeplexes so long as it's convenient.
But it's like the Newsom-Kirk interview, a lot of this stuff is just wind. Polis is more deft about it, because Hanania's whole 'but the experts' schtick is going to have more Democratic-moderate appeal, but it's not like he's actually going to care what Hanania or Decker think, just that it's useful to pretend he does. My guess is that Polis is trying to pivot to the national level, but maybe he'll retire to be a private advisor or focus on business.
I'll caveat that :
- in someone who had some heterosexual arousal beforehand (reading between the lines, possibly to the point of what we call schizoid sexuality today), was a strict top, already reported weird sex burnout (and was a piece of work otherwise: don't do nutmeg)
- extremely prone to addiction
- seemingly ended up bisexual-ish by modern standards, although that doesn't really get in the way of the wife-and-kids bit as much as the druggie bit
- in a study by a guy who was a bit of a crank even by the standards of his own time, depending on his reporting.
It's a pity that a) no credible research org is willing to try anything along these lines today, and b) the places that would want to try it are so sketchy, because it seems like tRMS should be a good deal more ethical and ... well, if not reversible, at least not as heavy on long-term infections and seizures. But I've got a kink for orientation play, followed a lot of bihackers in the tumblr ratsphere (and unintentional bihackers in the furry fandom), and I know more people who've ended up in relationships they can jerk off over but not consistently consummate than who've gotten it to work out well. Maybe they're just missing something -- I'm convinced that a lot of the 'physical' problems are downstream of scent and texture, which neither the Tulane study nor modern efforts generally train around -- but it might well be something deeper that only a small fraction of the populace can train.
On the gripping hand, if you just want a wife and kids, a gay guy doesn't really have to go that far. Beards are not new technology; post-nut clarity isn't gonna make a vial of your swimmers stop working; fujoshi are not unobtanium. Which points to the broader issue. Despite the perceptions, gay guys are looking for more than a hole (or pole) to pump and ignore until the next time they get horny.
The US says it had a "seizure warrant" for it. What does that mean? Was the vessel subject to US laws at any point, violated them, and this is the result? The US says that the tanker is sanctioned. What does this mean?
The United States issued a number of sanctions in 2022 related to the transfer of Iranian-originated oil in violation of US law. The Skipper, then operating as the Adisa, was one of the ships in question. See here for another example.
The seizure warrant is sealed, unsurprisingly, but it's almost certainly issued based on further evidence that the ship was being used to transport oil to sanctioned countries.
There's some messy legal philosophy stuff involved, here; the actual behaviors are not new.
... I'm going to be blunt, and this is separate from Motte rules since my expectations don't matter for that, but I don't think you gained much from the AI-gen, here.
There's a couple fixes ("Dixon" vs "Dimon"), but there's also spots where the LLM is introducing changes in meaning that either aren't correct or aren't what you were going for ("attempts" to "attempt", "can’t build themselves" to "no longer capable of building themselves", the introduction of "importing their safetyism to our shores").
Asking the AI for information or directly for grammar checks tends to be more productive than just asking it to rewrite text, though you should still review any recommendations and especially anything involving math before acting on or implementing it, and be aware that it's going to guide you toward toward least-common-denominator answers.
There's a difficult question in the difference between "your product" and "my thinking or perspective".
For a toy problem, I threw together a short story today, about 1200 words in the original draft, zero AI. Wrote it in FeatherPad (a notepad-like), so not even spellcheck.
Except no, that's a bit of a lie. I Googled a well-known phrase about birds in gilded cages, because I wanted to play on the original text, and Google's awful built-in LLM did give me some interpretations, even if not the one I actually went with. Still, I can't say I completely avoided influence from it. In this case, I am doing the thing and the thinking, the AI's just helping save time doing it. Am I outsourcing my product to Google? To a song from the 1900s?
After finishing the draft, I shoved the full text as a file upload into Grok with a request to check for spelling errors, overall coherence, typographical errors, and clarity. It caught a dropped fragment I didn't, and had a few suggestions. I took some, and didn't take others, but every keystroke going into the final product I wrote by keyboard, no copy-paste. In this case, these fixes are something I could do, but some parts probably not as reliably as the AI, while other parts, the AI is doing stuff I don't want.
Okay, what if I wrote the original really badly: a rough outline that hit all the same story beats and general concepts, but without any of the stylistic techniques, writing quality, or many specific segments I wanted in the story. That'd clearly be against the rules for where I submitted the story, and probably not result in anything nearly as good. And since I'm not a good writer, that's really damning it with faint praise.
Actually, let's try it: Grok (story) and ChatGPT (story).
For a tl;dr, not great, not terrible. Definitely didn't get the themes down, and the humor's not great, but ChatGPT's tone is mostly on the right tack, and I could probably inject a lot more dry comedy if I pushed it harder or feed it back in on itself with instructions to crank those aspects up. I'm not exactly a prompt expert. They did a good enough job on the 'enrichment' lists that I'm pretty glad I didn't try them before writing my own version, because I'd have probably gotten stuck on them and not moved to the ones I did use. And even if those were 'worse' from a realism perspective they were imo 'better' from a thematic writing one.
These LLMs clearly produce a better product than the outlines I put into the prompts. Whether it's better as a product than the short story I actually wrote is pretty dependent on what you're measuring; I'd argue I went quality over quantity, but writers get paid on the latter. The only real ideas the LLMs shove out are the enrichment concepts, and I could demonstrably come up with different ones. Indeed, a lot of the token cost for the LLMs is poking it into even moderately-good ideas. Does that outsource my ability to do things that way?
... except ironically, a guest-mode story prompt came up with an additional good theme to introduce: spelling out the main character's lack of trust in himself. It's a theme that works great for most horror fans that this story archetype is built around, but it's not really one I could see without being prompted. I don't think it's necessary for the product, but it's definitely a different idea.
I dunno where on that line Ademonera's post falls. I'd be less happy with it if they were using ChatGPT to source names and events, and then just using the names and events without reflecting on what they actually meant (or if they even meant anything). I wouldn't really care if it were glorified spellcheck, since at least it's not another Grammerly user. And if it's that messy spot where they had an outline of all the material they wanted to hit, but let the LLM reshuffle it into a narrative... well, I guess I wouldn't really see the point.
But the problems are separate ones.
In the spirit of hydroacetelyne's and Terracotta's conversation here, I've gotten trapped in car repair again. My daily driver's an older one -- and I strongly prefer decade-old used cars to start with -- so there's been some fun and !!fun!! involved.
This weekend had two targets.
I'd had a couple stalls on idle pop up recently, and that was the most critical; so far they'd only happened in parking lots while Christmas shopping and the car had restarted after seven or eight cranks, but 'tis the season for late construction and bad traffic and neither combine well with a dead car on a road. Nothing on an OBD-II reader, no check engine light, battery is older but still testing good, alternator power is good, wire harness is fine, spark plugs are less than six months old and were pregapped to this car (and had been fine for those six months), no obvious loose connections, issues do not seem to be tied to fuel level.
Crankshaft position sensor replacement. Bizarrely, it's actually easier to get to in this make and model than the oil filter, and while that's damning with faint praise (how do you get the oil filter wrench in there if when some hurried Jiffy Luber overtightens it?) , it's still a nice change from some other similar cars.
Still managed to cut my hands working on it. You'd think I'd listen to my own advice.
((To nerd out, both Grok and ChatGPT were able to diagnose this specific issue from ""I had a <Year> <Make> <Model> which had the engine stop suddenly while in a parking lot. Trying to restart the engine took several attempts. A tester gave good battery, alternator, and starter electrical values. What causes are likely?" On the other hand, neither of them gave complete and correct advise on replacement, both warning that I'd need to get under the car to fix it and not warning about how important proper fit and tightness were.))
Other one was a stereo head unit. I'd bought the kit (and wiring harness, and a 'one-size-fits-none' adapter from asian to DIN rails) a couple months ago with the dual goals of some alternative to country music, and a working backup camera, then left them on a shelf. Already too late to work on it before the weather gets chilly, but I was already cold, so here we go. The wiring harness was easy enough, though I'm glad I bought a new soldering kit -- absolutely love these things compared to the typical Weller. Asian-to-DIN adapter was absolute trash: not only did it require random parts of the plastic be snipped off in the instructions, the instructions weren't even close to right, so I ended up spending an hour in the cold with an X-Acto doing fit tests. Backup camera was found hard and left untried for now: finding a path to get from the head unit to the license plate on a hatchback that doesn't risk wires getting pinched every time I load up on groceries or Lowes detritus is gonna be some doing.
Separately, also helped a fellow robotics mentor/friend out with his recent purchase. He bought what he accurately calls a 'clapped out' sedan with the intent to get it into normal working order and flip it, and oh my god there was so much spray paint. On the upside, whoever started working it probably wouldn't know hot to prep cardboard for a can of rustoleum, but on the downside there's enough plastic on this thing that it didn't always care and what the actual fuck. Light blue spray paint. I hate those walmart-grade LED strips on cars, the previous owner used sheetmetal screws to hold them on, and they still seem sane compared to the spray paint.
DS9 has its stinkers, but at its heights it's easily the best Star Trek by a wide margin. I'd still put Babylon 5 a little higher, but it had steep competition from DS9 in a way that no other Trek of that generation, or the current gen, could or was even trying to supply.
That's an impressive amount of effort for a Halloween costume, and some remarkably unobservant partygoers.
... I'm almost afraid to ask, but did you have the pig's head facing in, or out?
Yeah, that's probably a more honest engagement with the events. I've just seen a lot of people say it was a big important deal that tells us about falling modern standards, so it really bugs me that it's just such a mess.
Is that…Trek?
Nextwave (cw: sound, mild profanity)! Only Warren Ellis comic I can unconditionally recommend. Very short series, sadly.
Not exactly unusual—didn’t Scott write about ADA enforcement in these terms? The main limiting factor is the difficulty of bringing a case. Technology has to have reduced that cost, so a given org can target smaller companies.
I think so, but I can't find it now. The ADA (and the Texas abortion pill thing) are laws or at least regulations on the books, even if they're probably getting stretched to their breaking points. A lot of this stuff isn't a strict rule, or even necessarily written down anywhere, so much as it's just Understood at ultimately one-on-one scales. Sometimes that's unavoidable: in aviation, I think you could get five opinions from three DERs if you bring up flutter analysis. But it ends up in a world where a lot of things are theoretically allowed, and you can even find people doing them quite publicly, but also prohibited.
Over five percent of China's power comes from fission plants, and that's underrating it since they've got very high uptime compared to on-demand plants. As for why it hasn't scale up faster, China's political classes had very obvious mixed feelings about dependence on foreign-produced infrastructure for a long time, which only went harder as western regulatory overhead killed western nuclear power. While they've theoretically had 'domestic' production of nuclear plants since the mid-90s, they didn't actually manage serious production of the CNP-600s until 2010-2012... at which point the Fukushima disaster and its political fallout lead them to go back to the drawing board and start the production cycle again.
But they've put >3 GW of fission power online just in the last year. As bad as their political situation is for power construction, it's still beating the west's.
Google's been trying something similar with Sima 1 and the more relevant Sima 2, though I'm not seeing anywhere near as much information about what the model parameters and configuration were for that one. Qwen-2VL-7B seems, intuitively, way too small to make this sort of deep analysis and decision making, and it's kinda weird that a lab environment didn't go to something like Qwen-2.5VL-32B. But 7B was also obscenely good at captioning videos and making problem solving analysis from it, and people had gotten some results, if not great ones before.
Unfortunately, a lot of the value in the study is going to depend on exactly what and how they tested the model, and there's really not enough detail here. An hour-long autonomous play session of 'finish this mission' is the big selling point, but I don't know Genshin well enough to say whether a) that mission was nontrivially different from training data or b) that it involved more than 'follow quest marker, spam A at enemies when lock-on-button does anything.
It'd be interesting to see more information about how well these models handle completely out-of-training problems, though. I've talked about using a MineCraft mod to see how well a model can create a 'new' solution, but these sort of games are trivially easy to present completely out-of-training problems, ranging from stuff as trivial as an enemy or attack that's changed color, all the way up to completely novel gameplay mechanics (eg, FFXIV threw in a "change color to reflect attacks" mechanic several years after initial release). I wouldn't expect an LLM to possibly one-shot every version of this, and some probably aren't possible for architectural reasons (eg, even if a model could go from vanilla minecraft to GTNH from, no plausible memory-constrained implementation would have the context window for even some mid-game recipes), but I think it'd say some interesting things regardless.
To some extent, it does happen: one of the many swerves of the Peanut The Squirrel saga from last year was that the squirrel's owner was doing very gay4pay-looking onlyfans while having a wife. Which could be bisexual, but you can be bisexual and work blue collar, and a lot of the framing was more 'what straight guys think gay guys want' than what even the bottomiest gay guys actually want.
Which points to a part of the problem, if you're a straight guy trying to sell to gay ones. Look at fantasy (art or written) gay4pay or orientation play, and there's a lot of stuff that's not just going to be uncomfortable for a straight guy to try (though there's definitely a lot of that: forget taking a dildo, who wants to wax their chest), but will also just be very hard to credibly recognize or sell.
On the other hand, SquirrelDaddy was an OnlyFans making in the top 1% and maybe top 0.1% of male earners, so maybe my tastes are just weird.
That said, the available business strategies are less viable. The nice thing about gay courting's that there's somebody into everyone, and sometimes the breakdowns can come in surprising ways (eg, one of the top 0.001% male earners on OF is outright obese)... but there's not necessarily many people into anyone. The entire male side of onlyfans pulls in about a fourth of the revenue that the women's side does, and the top (hurr hurr) is saturated with tops that are, to be blunt, not possible for the average man even if he wanted to. Any of the highest-end creators in any space tend to have people who are doing that career as a full time job and a half, but to be blunt, almost all of the top male creators are genetic freaks. Ain't no amount of zinc and pineapple that's doing that.
I wouldn't recommend a gay (or 'I'm just posing naked for my fans, which I totally assume are women') actor go to Dubai for sex work, though. The UAE's more extreme punishments are theoretical even for their own inhabitants, but they have made life miserable for tourists in the past. While those punishments theoretically include M/F situations, they're much more likely to hit man-on-man ones.
China actually started up a molten salt 'thorium' (eg, starting with uranium, then moving to thorium) reactor last year, with the first full thorium cycle this November. I'm not optimistic about its effectiveness, but that's more because it's a lot more complicated than it needs to be, rather than net energy or net cost problems.
A lot of the various cost problems with nuclear plants reflect political willpower, rather than actual material costs. That's most serious in the United States where we've intentionally made them several times harder to produce at the same time that the control and construction technology has gotten much much better, but most western governments have done something similar. (or just had politicians launch rockets directly at the construction sites.)
There's a revealed preferences sense where, if you can't solve those political problems, you can't produce power at price, and it's not entirely wrong. But it's misleading to treat it as a physics problem.
Coyotes are pretty well-hated and oft-culled because of their attacks on pets and livestock, though, and there's also the 'dingos ate ma baby' option of simple incompetence. That said, if you really want to go nuts on coyote conspiracies, the degree that coyote populations have exploded and the individual coyotes themselves have gotten much smarter in <10 generations is a real fun question.
For fun conspiracies I actually believe:
- Piggate wasn't real. For all it Took Down A National Government, Cameron was already a political dead man walking before the drop, and it was just a really convenient way to force him out without actually engaging with the political controversies that had undermined his party. The same behavior is totally consistent with an already-unpopular prime minister getting smeared by a schmuck he'd pissed off badly enough, and then found that none of his 'friends' were willing to pay the political capital to back him up.
- There was a coverup one direction or the other for the Bloomberg Supermicro thing: either a lot of people who could prove it were told not to do so at the risk of destabilizing international relationships, or a lot of people who could disprove it were told not to do so lest they destabilize US financial markets (and get blackballed). I'm not very confident on this one, but it's just such a weird goddamn story.
- A number of serious industrial or transportation sector accidents were really Reinvented Suicide As A Group Activity, but various incident analysis groups have instead used them for purposes ranging from getting unrelated political goals to deflecting from local political or social problems to just shaking down foreign businesses for cash. There's been a handful of these situations where jurisdiction friction has lead to them getting 'caught' -- aviation is particularly prone to it, with SilkAir 185, EgyptAir 990, and the recent Air India 171 -- but I think they're far more common than anyone wants to admit or even mention publicly, especially since there's a risk that publicizing them could incentivize further or larger attacks of the same kind. Basically, most large countries have a bunch of CEAF 5735 in a thousand different fields. The SL-1 incident is the safest one to mention, but there's some electrical and chemical processing examples from the tens to hundreds of deaths.
- A lot of 'advocacy organizations' related to industry regulations are wholly-owned government groups, and are explicitly-but-nonpublicly threatening to bring the weight of those government orgs to bear if targeted companies don't agree. Yeah, boring, almost too obvious to be worth mentioning for the obvious cases, between Ofcom and NCOSE existing, and
XTwitter's recent fine in the UK. But there's a lot of these orgs running at <100 person levels regulating through smoke-filled backroom deals; a lot of what's 'weird' about the modern era is just the ability of those orgs to impact companies with large impacts but not the large scales of pre-internet companies.
- Prev
- Next

Yeah, especially given the broader zeitgiest at the time, it was a genuinely surprising take, and the level and degree of conflict between the heroes and the government is a much more nuanced take than the "you people are young" summary he'd give in interviews. As an exploration of political philosophy or philosophy of war, it does a pretty good job, if limited by its time and its awareness.
My big complaint's just that it doesn't really feel great about its characters. It's a comic book, and a short-run series at that, so expectations are never high to begin with, but the ending is undermined not just because It's Woke, but because it doesn't really feel like a conclusion for the characters that got to it. Tom feels very Batman-inspired and Horus very Superman, and that's a classic for a reason. Do their perspectives actually say anything about Truth, Justice, and the American Way? About
assassinscriminals being a cowardly and superstitious lot? Or if they're working as alternative company counterparts to Captain America and Iron Man, anything about their political philosophy? Artemis pointedly compares the US government with the Nazis in one argument with Dominic: did he persuade her before his death, or was her violent persona and facing always an act?You don't need this sort of deeper layering. Black Summer benefits in the sense that not doing it means you can't do it poorly, like No Hero or The Boys and their utterly wretched X-Men parodies. But it's frustratingly noticeable given how little else there is to say about the characters.
That said, I do think it's one of, if not the best, Ellis short series. So part of it's probably me not clicking with him as a writer in general.
More options
Context Copy link