roystgnr
1d ago
How likely would that have been? I know international relations are fickle, but they usually only turn on a dime in cases where an alliance of convenience is papering over underlying hostility or where one party's government is utterly replaced by hostile opposition.
roystgnr
1d ago
Should the C compiler let you declare a function that returns a value and then let you omit the return statement? Is that mistake your fault or the language's fault? Formally doing this is undefined behavior but that does not always mean crash!
It's the language's fault (that probably should never have been allowed by the standard, and if it wasn't then the compiler could catch it by default) and it's your fault (you shouldn't have written that), and it's other language users' fault.
That third one might take a bit of explanation.
Any decent compiler these days will warn you about that error at compile time, and will stop the compilation if you use a flag like -Werror to turn warnings into compile-time errors. So just always use -Werror, right? We could all be writing a safer version of C without even having to change the C standard! Well, "look for functions that declared a return value but didn't return one" is an especially easy error for a compiler to catch, but there are others that are trickier but more subtle. Maybe you add -Wall to get another batch of warnings, and -Wextra with another batch, and you throw in -Wshadow and -Wunused-value and -Wcast-qual and -Wlogical-op and ... well, that's a great way to write your code, right up until you have to #include someone else's code. At some point your OCD attention to detail will exceed that of the third-party authors who wrote one of your libraries, and you can't always fault them for it (these warnings are often for code that looks wrong, whether or not it is wrong - even omitting a return statement could probably save one CPU cycle in cases where you knew the return value wasn't going to be used!). So, I have special headers now: one to throw a bunch of compiler pragmas before #include of certain third-party headers, to turn off my more paranoid warning settings before they can hit false positives, then another to turn all the warnings back on again for my own code, like a primitive version of "unsafe".
I was once paid to port C code from a system that allowed code to dereference null pointers (by just making the MMU allow that memory page and filling it with zeroes). And so the C code written for that system used that behavior, depending on foo = *bar; to set foo to 0 in cases where they should have written foo = bar ? *bar : 0; instead. As soon as you give people too much leeway, someone will use it, and from that point onward you're a bit stuck, unable to take back that leeway without breaking things for those users. I like the "nasal demons" joke about what a compiler is allowed to do when you write Undefined Behavior, but really the worst thing a compiler is allowed to do with UB is to do exactly what you expected it to, because then you think you're fine right up until the point where suddenly you're not.
roystgnr
1d ago
Types should be explicitly written out in code! They're a very important part of the logic!
Sometimes types shouldn't be explicitly written out in code because they're a very important part of the logic. If I write generic (templated) code that returns the heat capacity of a gas mixture at a given temperature, sometimes I just want that temperature to be a double so I can get a quick answer for a point's heat capacity, and other times I want it to be a Vector<DualNumber<double, SparseVector<int, double>>> so I can get SIMD or GPU code that gives me a hundred points' heat capacities as well as their derivatives with respect to the input data that was used to calculate temperature. There's basically no way I'm writing intermediate data types for such a calculation as anything but auto.
When designing even simpler library methods I'm also sadly kind of a fan of users writing auto out of laziness, too. If I ever accidentally expose too much of my internal data structures, use too small of a data type, etc. and have to change the API later, often I can change it in such a way that lazy auto users are still fully compatible with the upgraded version, but users who explicitly wrote foo::iterator can't compile after my switch to bar, and users who explicitly wrote int are now slicing my beautiful new size_t and are going to be unhappy years later when they run a problem big enough to overflow 2^31.
roystgnr
1d ago
I argue that this isn't a better outcome than what would have happened in C, which would also be to crash.
This is both normatively and positively wrong.
Positively: in C, Undefined Behavior often leads to a crash, but is not actually required by the C standard to lead to a crash. The outcome is literally undefined.
Normatively: If you write code that leads to Undefined Behavior, the C compiler is allowed to and often will emit code that will crash; this is the same outcome as the Rust case, but is still a worse situation because grep unwrap is a thing and grep some_regex_catching_all_C_UB is (despite linter developers trying their best) only a dream. The C compiler is allowed to emit code that will make demons fly out of your nose. The C compiler is allowed to, and often will, emit code that will hand control of your computer to the botnet of whichever attacker first discovered how to trigger the UB, at which point if you're lucky your computer is now laundering your electric bill into some mafioso's bitcoin wallet at pennies on the dollar, and if you're unlucky your computer is now an accessory to DDOS attacks or blackmail or financial scams. These are much worse outcomes. Even CloudFlare crashing is much better than CloudFlare being compromised would have been.
Bailey: "C is unsafe, because of all the memory unsafe code people have written, and we should rewrite everything in Rust to fix all of it!"
The second clause here is false IMHO (though bias makes MO very H: I've been writing a little C and a lot of C++ for 3 decades and have no current plans to stop), but the first clause is simply theoretically and empirically true and belongs in the motte.
I do wish the second clause was true, for some language if not necessarily Rust, because I have about a hundred other gripes with C/C++ that can probably only be fixed by someone starting from scratch ... but whenever I investigate a new language that I'm excited to see fixes flaw X, they seem to do it at the same time as they omit all possible support for features Y and Z and end up with something worse (for some of my purposes; there are three other languages I write in for different use cases) overall.
roystgnr
3d ago
I never knew Marley was a Jewish name.
One of Bob Marley's kids married a Jewish-descended woman; do she and their kids count?
"Marley" usually comes from Old English, occasionally as an Anglicization of Irish.
"Jacob" is Biblical, so it is a common Jewish name, but it's also the middle name of one of the most famous Protestant Christians of Dickens' time, John Jacob Astor, possibly the wealthiest businessmen in the world when A Christmas Carol was written.
or maybe you might be wrong.
Yup. The most relevant etymology here is that of "Secure Signals". A Holocaust denier who named himself after the SS might not be the best source for objective discussion of antisemitism.
roystgnr
8d ago
Was this some "fruit of the poisoned tree" thing? I'm reading that they literally found his DNA on the knife sheath left at the crime scene.
roystgnr
9d ago
there are a lot of irregularities in his behavior
Norm MacDonald voice: "Irregularity Number One: he assassinated a guy!"
I'm not sure "I should wander around rural PA with my manifesto and weapon" is any crazier than "This shooting should definitely fix American health care, or at least impress Jodie Foster or something!"
roystgnr
10d ago
they were all oddly okay with this?
"All" was two people, one of whose first reaction was to beat the other to the ground after finding out about that last twist, so I wouldn't say he was 100% okay.
The operative who orchestrated it was 100% okay with it, sure: "And if your conscience is bothering you, you should soothe it with the knowledge that you may have just saved the entire Alpha Quadrant, and all it cost was the life of one Romulan senator, one criminal... and the self-respect of one Starfleet officer. I don't know about you, but I'd call that a bargain." Best quote in the show.
But the full reason it was one of the best episodes ever was watching that furious Starfleet officer struggle to become okay enough with it: "So, this is a huge victory for the good guys! This may even be the turning point of the entire war! There's even a "Welcome to the Fight" party tonight in the wardroom!... So... I lied. I cheated. I bribed men to cover up the crimes of other men. I am an accessory to murder. But the most damning thing of all... I think I can live with it... And if I had to do it all over again... I would. Garak was right about one thing – a guilty conscience is a small price to pay for the safety of the Alpha Quadrant. So, I will learn to live with it...Because I can live with it...I can live with it. Computer – erase that entire personal log."
The thing about the Trolley Problem is that in its classic formulation it feels simple enough: you pull the lever and the trolley kills one person, but thanks to you the trolley didn't kill five. 5>1, yay you! But what if we replace the inanimate trolley with a conniving human being, and your push enables cold-blooded murder? What if you're only in a position to give and take lives because you took oaths you've now spat upon? What if there was nothing but the fog of war to inform you about how many people were really going to be hurt on each track, or about how badly? What if you can't even ask another trustworthy soul about whether you did the right thing, because the utilitarian decision's effectiveness relies on your continuing ability and willingness to lie about it? Is this still a situation where cold utilitarian calculation trumps virtue ethics? Would you go full Kant and immediately reveal the truth to everyone, damn the consequences?
I guess also that episode where it cold opens to Worf having rough dry humping sex with Dax?
Okay ... I haven't actually watched all of DS9 in decades ... but when was this? How did I forget this? Was it bad enough that I've actually blocked it out?
roystgnr
10d ago
1979 was believed to be either South African with Israeli help or Israeli with South African help. The explosion appeared to be only a few kilotons-equivalent, though, so it could have been a fission bomb rather than a "hydrogen bomb", and if it was a fusion bomb then it was almost certainly a neutron bomb, designed for "low" explosive yield in favor of radiation.
I can't imagine Ukraine would invite nuclear reprisal by nuking Moscow, not unless the tanks were literally rolling into Kyiv at the time, but Toretsk would be an even less likely target.
roystgnr
10d ago
First time watch, not rewatch?
In hindsight, DS9 doesn't start to get frequently great until the very end of season 2, despite having a relatively strong pilot for a Star Trek show, just to warn you. You might be tempted to just skip ahead a couple seasons, and indeed early DS9 mostly lets you get away with that (like most 90s-era TV shows, producers didn't dare risk losing an audience who missed one episode and hence they pushed for low levels of arc-plot and continuity), but definitely find some watch guide to consult if you do. There are some good early episodes that are worth watching because their character development makes later great episodes even better, there are a few great episodes ... and IMO there's one of the best episodes in all of Star Trek, just plopped into Season 1 because that's where they needed it for one character's arc.
Old news, but I just ran across it when it went viral again recently and the Bloomberg story reminds me of it:
Exploitable systems are so much easier to create than secure systems that it's hard to attribute even actual proven exploitability to malice! Aside from the software issues in that discussion, consider the hardware. Fifty years ago, if something you brought into your business had a tiny secret microphone, that would have been proof-positive that someone with major signals-intelligence chops was trying to bug you. Today, it just means that the fastest way to create a special-purpose electronic device is to just grab some general-purpose computer board and flash it with your own special-purpose software, and of course your general-purpose-computer designer threw in a 3.5-cent-each MEMS microphone because why not?
roystgnr
11d ago
I'm afraid I didn't save a source on this, just the numbers, but the last studies I saw claimed that:
- $12K/year is a 92nd percentile income for camgirls
- $50K/year is a 99th percentile income on OnlyFans
- The median OnlyFans income is around $2K/year
- The modal OnlyFans income is 0.
Yeah, there was some girl who made $1M in 24 hours. That's how dualized labor markets work. From the outside, it looks awesome to be a multimillionaire actor, ball player, musician, or whatever, because the ones who become successful are almost tautologically the ones that everybody knows about, and the ones who don't make it big become invisible.
No, "most of these girls" cannot "buy mansions, hire maids, babysitters, tutors, and never work a single day in their lives after grinding OnlyFans for a few years". The ones who metaphorically won the lottery can, sure, but every blue collar working guy can literally buy a lottery ticket if he wants.
That said, I'm not sure if there's much more downside to the metaphorical lottery ticket than there is to the literal one. You do have to choose between being open about your sexual past with future partners (and risk scaring some off) or being tight-lipped about it (hurting the level of emotional intimacy you can share), but that's been a tradeoff that everybody's had to face since the Sexual Revolution, whether there were cameras involved or not. "I only did it for the money" might even inspire less jealousy than the standard "I loved my exes just as much, but don't let that make you pessimistic about our future; tenth time's the charm!" Worst case scenario: at some point somebody's going to set up a Reverse Image Search with much more advanced AI than Google originally had and much fewer scruples than Google currently has, and they're going to suck in every random archived torrent they can find ... but under what circumstances is even that really going to backfire on you? The modern liberal consensus is "that's your own business", and the most common conservative consensus is "that's awful, but you can repent and be forgiven".
housing has unaffordable while boomers could get a house on a single blue collar salary, etc., despite every single official statistic contradicting them.
It was the generation before the boomers who really had cheap houses. Fun fact: the 1940s jump from 75 to 110 on the inflation-adjusted Case-Shiller index was called a "housing shortage", and people back then expected to see prices eventually brought down again, in a decade or two tops, not further doubled.
To be fair, houses have also skyrocketed in average size (50% IIRC) and quality (part of that 1940s price increase was that luxuries like "indoor plumbing" were becoming universal) since that time. We can also naturally afford to spend more of our income on houses than we could during the Great Depression or WW2, and we tend to still have more disposable income left over.
I'd still cut the kids some slack on this one. We're still (hopefully!) at a housing price peak today, despite mortgage rates more than doubling a few years ago. Double the cost of their houses, then double the cost of borrowing the cost of their houses, and pretty soon we're talking about serious money!
roystgnr
15d ago
I think you would have a very hard time locating someone who scores in the 90th percentile of numerical reasoning but the 10th percentile of verbal reasoning.
I recently indulged in a little smack-talk about how Math majors have higher SAT verbal scores than English majors, but now it's bothering me that I don't even know what the correlation is between SAT math and verbal scores ... and I'm not even sure how to find out! Google searches and AI summaries seem to be so polluted with stat questions about correlations in hypothetical SAT results that I can't quickly find anything with results for the correlation in actual SAT results. Even the College Board's annual report, which is at least statistically literate enough to define and report standard deviations for each subtest, doesn't report the correlation between subtests. They have other reports of correlations between paper and digital SATs, between SAT results and future college grades... They'll even report separate correlations of subtest scores with other tests and with HS GPA without mentioning the subtests' correlation with each other!
Well wait I do find this: A report from Connecticut estimates a 0.89-0.9 correlation based on an observed 0.82-0.85 correlation. N=1,343 but I'll take it. Then with the bivariate normal distribution CDF from Octave/Matlab, I ask for 0.1-bvncdf([norminv(0.1, 0, 1) -norminv(0.1, 0, 1)], [0 0], [1 0.9; 0.9 1]), which is ... 1.5e-10? About one person in seven billion? (as opposed to the one person in a hundred we'd get if there was no correlation). If I go with that 0.82 correlation I still only get one person in 2 million, so that kind of test score would probably be a thing that's happened before, but only because the kid was having a lucky day with math and bad luck with reading simultaneously, not because you'd expect to see a score like that again on a retest.
Of course SAT scores aren't actually a Gaussian distribution, but I think that thought experiment still strongly suggests you're right, and anyway there's no way we're finding data with higher moments. Even if we got our hands on raw data, I'd bet that a kid with 650+ math and sub-380 reading is much more likely to be a recent immigrant who's still struggling with English, not someone who's actually got poor verbal reasoning skills in general.
roystgnr
15d ago
Oh, no, you can get less charitable while being more correct. An old Eliezer Yudkowsky quote I saved:
Education is not secretly intended to turn adolescents into conforming factory workers. If 'they' were trying to produce factory workers 'they' would take advantage of elementary modern research on conditioning, reinforcement, and shaping to produce much better factory workers, rather than making conformity so unpleasant and unrewarding. If schools were actually trying to teach things they would take advantage of modern research on spaced repetition. If grade schools and high schools were secretly babysitting institutions, they would offer more flexible hours. If colleges had been designed by employers to weed out anyone with trouble submitting to authority, then there would be harsher enforcement of drug rules or more prohibitive sexual regulations as tests; as it stands, many students who gain a college credential will still have trouble submitting in a workplace. The educational process has no agenda, hidden or otherwise. The overall process of going to college might have the intention of gaining a piece of paper, but the actual day-to-day activities of college are not being optimized for any intended consequence, by anyone.
There are a lot of relevant social roles played by various educational institutions. There are no relevant social roles being played as well as they could be by these institutions. It's not exactly that nobody is trying to optimize for any particular consequence, though, it's just that handling any one of their roles perfectly well would conflict with a different role being played for a different interest group, and trying to optimize for both at once would interfere with the desires of a third interest group, and so on.
roystgnr
16d ago
"If I had more time, I would have written a shorter letter."
If I'd expected a nomination I'd have tried to be a little less rambly. That was supposed to be a "I've been a bit too pessimistic about Blue Origin, here's a correction, my bad", but I kept seeing more to add and I couldn't think of anything to cut.
So, screw it, here's more to add, for anyone who was interested in the prior post:
The first LandSpace Zhuque-3 launch just reached orbit, then failed its booster recovery. For them this is excellent news on the whole.
Great news: new rocket designs tend to fail their first flight as often as not, and it's always a relief to see "not", especially with a design that's trying to be cutting edge with stainless steel tankage and methane engines. For comparison, their Zhuque-2 failed on both its first launch and on the latest of its 5 launches since.
Bad news: they could have been the third institution ever to manage a powered orbital booster landing, mere weeks after Blue Origin became the second; now there's a decent chance of them being beaten to it.
Good news: they came really close to being second, with the returning booster crashing right next to the landing pad, despite Blue Origin having roughly a 150% head start. At this rate they'll almost certainly manage a landing faster (relative to company founding) than either Blue Origin or SpaceX did. Older Chinese launch vehicle designs were infamous for dropping expended boosters near populated villages with little control while spraying carcinogenic orange smoke; even if LandSpace takes a while to perfect their landings, a clean methane fire right over a dedicated landing zone is a perfectly reasonable outcome.
Bad news: the landing burn appears to have turned the entire booster into a fireball from one failed engine ignition. Half the point of clustering several engines on one booster is supposed to be that you can recover from a single failure, as happened to a SpaceX Dragon 1 cargo mission back in 2012.
Great news: even if they have a little work on robustness and landing to do, this rocket may be economical for them already; if they've hit their payload targets (hard to be sure yet; their first flight was a "mass simulator") then they've already doubled the payload of the Zhuque-2 and are up to roughly half that of Falcon 9, which they're expecting the Zhuque-3E to meet (at roughly the same price, too) after engine upgrades and a vehicle stretch.
One of the possible outcomes of the "new Space Race" has always been another Pyrrhic victory, this time for China: flags and footprints, Apollo style, followed by a series of cancellations of and creations of economically unsustainable programs, Apollo/Shuttle/Constellation/SLS style. That possibility is looking more and more remote. The next attempt at a new booster design landing, from a different Chinese launch vehicle, is expected later this month. Chinese institutions may take 3rd and 4th place in the reusable booster race.
roystgnr
18d ago
Oh, hell, I forgot about the Masculine Mongoose! Yeah, those were wonderful, and more accessibly so than "Kindness to Kin"; still short stories, but I shouldn't have forgotten to mention them.
The "conspiracy world" stories were interesting but not great as stories.
I hadn't read "but hurting people is wrong"; thanks for the recommendation!
roystgnr
18d ago
deterministic at macroscopic scales, or close enough that I don't care about the difference
Nitpick, since I don't think it matters for your philosophical point, but the difference can be huge. I know the "butterfly farting in the Mezozoic" thing sounds like a silly metaphor ... but it's not silly, and it's not even a metaphor. Microscopic perturbations changing macroscopic weather (in weeks of time, not just eons) is a straightforward consequence of "chaotic dynamical systems exist" (not in a two dimensional state space or lower, but easy to prove with three dimensions) and "the weather is a chaotic dynamical system" (not actually proven, but Lorenz's first three-dimensional system demonstrating chaos was something of a simplification of the real sextillion-Avogadro's-number-dimensional system, and we're pretty confident the more complicated one actually is more complicated). A butterfly changing the course of a tornado was literally an example Lorenz used in his talks. Chaotic systems can be deterministic (except in the sense that we can never accurately determine their output because the cost of doing so increases exponentially with simulation time), but if you combine a deterministic chaotic system with a stochastic initial perturbation, that perturbation grows exponentially until it's the size of the attractor and the whole macro state depends on it.
roystgnr
19d ago
He doesn't have any other fiction on the level of HPMoR, though, does he? "Three Worlds Collide" was interesting but not great. I really like "Kindness to Kin" but it's just one short story. I'm working my way through "planecrash" right now, and so far it's a pretty good first draft of something that could have become a good novel series after it got a ton of editing that it didn't get.
On the other hand, HPMoR is roughly four long novels put together, and I watched it get produced in real time and saw how little editing it got, and although I find that intensely annoying (I'm trying to avoid getting too spoilery, but at at least one point there's an explicit moral that the protagonist has been an idiot by neglecting to consider advice from others, and the irony just hurts), it's quite amazing for someone to make it from beginning to end of that much writing, juggling a coherent arc-plot through multiple major tonal shifts, without ever seriously dropping the ball. In "planecrash" he's at least capable of co-writing and adapting to others' ideas, so he could have handled working with an editor if he'd ever actually made a career of writing and gotten one.
roystgnr
23d ago
Microsoft have embraced H1-B visas and Infinity Indians with open arms... and look at what's happened to their products.
Windows 95 and Windows 98, a product line with over a hundred million users, had a bug that caused them to invariably crash after 49.7 days, when a 32-bit millisecond uptime counter would roll over.
This bug went undiscovered for nearly 4 years, because so few of those hundreds of millions of computers made it 7 weeks without crashing that nobody noticed the final hard cutoff of any survivors.
roystgnr
24d ago
Technically the FCC rejection of Starlink's rural broadband subsidy application was after Musk made the offer to buy Twitter, but before the transaction took place.
And the Biden grudge against Musk was at least a year older still. Biden holding his "Electric Vehicle Summit" but then not inviting the manufacturer of a supermajority of US EVs because they weren't unionized was ... well, the phrase "crossing the Rubicon" does get overused, but that NY Post article helpfully includes the photos of Biden literally test-driving an electric Rubicon, so who am I to argue with dramatic irony?
"Hopefully their future will see a little less gradatim and a little more ferociter."
A few weeks ago, the topics of SpaceX and their competition came up, and although I was "non-ironically excited for the possibility that Blue Origin's upcoming second attempt to accomplish a booster landing is about to succeed", I might have primed myself to fear disappointment by talking about Boeing's Starliner first. A year and a half ago Boeing were poised to be the ones to break SpaceX's current monopoly on United States crew launch ... and then they launched crew on a vehicle that still hadn't fixed all the Reaction Control System thruster problems from three years earlier, and they needed SpaceX to get them safely back down again.
It's a good thing I didn't quite finish this post yesterday like I'd planned, because today it's been announced that Starliner won't even try to fly humans again until after another cargo flight, and NASA's brought their contract down to 4 missions instead of 6. I don't think either side of that contract wants any part of it anymore; it's all trying to recover a fraction of sunk costs at this point.
Blue Origin, on the other hand, appears to indeed be moving from "gradatim" mode to "ferociter" mode!
After a number of minor delays, the NG-2 mission successfully launched New Glenn's first mission out of Earth orbit: two ESCAPADE spacecraft (buses by Rocket Lab, instruments by NASA and a few University labs) are now on their way to the Earth-Sun L2 point; after an Earth flyby late next year they'll be headed to Mars next. Despite the long-term SpaceX focus on Mars colonization, the Blue Origin + Rocket Lab combination will likely beat them to putting spacecraft in Mars orbit; even the SpaceX launch of Europa Clipper only included a Mars flyby, on the way to Jupiter.
The first New Glenn launch lost its booster due to engine relight failure; in NG-2 a little of the live video was lost, but the booster itself stuck the landing perfectly, as seen more clearly from more distant footage, making Blue Origin the second institution to accomplish a powered landing of an orbital rocket booster, with the second-most powerful rocket to ever be recovered, and the most powerful to be recovered on its own landing gear.
The landing was very inefficient, aiming far from the target initially (as SpaceX does too, to ensure that if engine restart for landing fails the rocket will instead impact at a point where it can't do any damage), but then taking a very slow, almost "hovering" horizontal slide over to the precise center of the landing pad, burning much more landing fuel than SpaceX's "hover slam" landings. Unintuitively, this is probably all a good thing for Blue Origin. First, it's a demonstration of the ability to hover, which although inefficient as a nominal trajectory, adds robustness when things go wrong. SpaceX is well over 500 successful landings now, but this is after burning through a test vehicle and multiple "splashdown" landings before they felt ready to risk a barge, followed by four or five failed landing attempts, all due in great part to the difficulty they had landing a booster whose Merlin engines couldn't throttle down enough to hover if necessary. Second, this is a good bit of context for the rumors (anonymous, but via a trustworthy reporter) that the first New Glenn vehicle capabilities were well under the design's target payload numbers. It's common for a new spacecraft to gain unwanted mass and so lose performance during the design and testing process, but this can sometimes be fixed with subsequent iteration, and that big plume of burning mass ought to be a relatively easy target for them to quickly fix. They've also announced a 15% improvement to each of BE-3 and BE-4 engine thrust, to be deployed on future missions, so they'll be getting performance back from reduced gravity losses on both the way down and the way up.
Adding metaphorical weight to the performance problem rumors was the removal of literal weight from their first launch, whose "Blue Ring Pathfinder" looked like a toy compared to the full "Blue Ring" spacecraft bus+vehicle they'd been working on. There's still no semi-firm launch date for the full Blue Ring, but they've released photos of their first flight vehicle, in production now and at least intended for launch early next year.
We've also now got pictures of a more impressive flight vehicle: the Blue Moon Mk1 cargo lander, to be launched to the lunar surface early next year. It's half the height and only a fraction of the cargo capacity of their planned Mk2 crew-and-cargo lander, which is itself tiny compared to Starship HLS, but until those are launched this will be the largest craft ever to land on the Moon, roughly a third bigger than the Apollo Lunar Module.
We've even been shown a glimpse of test hardware hinting at long term design work: a deployable hypersonic aerobrake, "saving significant mass and cost and enabling heavy cargo delivery from the Moon, to Mars, and point-to-point missions on Earth."
And finally, as part of that thrust improvement announcement, we got a look at their longer-term plans: a scaled up "New Glenn 9x4", with 9 first-stage and 4 second-stage engines (as opposed to the existing newly-renamed "New Glenn 7x2"), stretched to be taller than Saturn V thanks to the additional thrust, expected to give them roughly 50% more capacity to Low Earth Orbit and nearly triple their payload to Trans-Lunar Injection. The 9x4 still won't have as much TLI payload as even the Block 1 version of the Space Launch System, but this is still likely to be a little more beyond-LEO payload than a fully-expended Falcon Heavy, but with a much roomier fairing like Starship, from a rocket in a (partially-) reusable configuration - and unlike with SLS, Blue Origin has been designing with multi-launch mission plans and orbital refueling in mind. That payload to LEO (two thirds of the Starship V3 target, and twice what V2 was reportedly capable of) may end up being more important than the TLI numbers in the long run.
None of these future numbers from either company are guaranteed, of course. The first Starship V3 booster just got wrecked in a failed pressure test, pushing the next Starship flight back from "January" to "Q1". New Glenn was originally supposed to launch those Mars probes a year ago, but juggled their schedule a bit after they lost two New Glenn stages, one also to a test failure and another to worker mishandling, last year.
The obvious thing Blue Origin really still has to work on is cadence. In Fall 2024, Blue Origin was expecting to do 8-10 New Glenn launches in 2025; they ended up managing one in January and a second in November. Rocket Lab likes to brag that their Electron was the only commercial rocket to ramp up faster than Falcon, and that's a fair brag, because cadence is much easier said than done. The Space Shuttle fleet was supposed to fly at least 24 missions per year, ideally more like 50; they ended up at 4-6 with a peak of 9, and cut some tragic corners just trying to reach that. I think Blue Origin has the right design to start with, at least. The difference post-landing between their gleaming methane-powered New Glenn booster and the soot covered kerosene-powered Falcon 9 boosters is like day and night, and hopefully that lack of coking is going to make inspection and maintenance of the rocket internals easier as well. Blue Origin is still talking about doing one or two dozen launches in 2026. I'll be very surprised if they even come near the low end of that, but I'm hoping to be eating crow in a year.
Still, seeing two successful New Glenn flights in a row, including their first successful landing, is heartening. Blue Origin not only managed to land an orbital rocket under the wire of SpaceX's landing 10-year-anniversary, this year they've already managed a couple entries on my checklist of what SpaceX has been up to since:
- Added downrange booster recovery options
- Added booster recovery from and (coming soon, in Blue Origin's case) reflight after missions beyond LEO
- Begun launching national security payloads
And they're working on more, both in the next year:
- Launching payload to the Moon
and in the longer term:
- Adding fairing recovery
- Adding extended fairing options
- Launching a multi-thousand-satellite constellation (hopefully starting with the first launch next year)
They're still behind, but for the first time in decades it feels like they're not falling further behind. Space launch in the USA may soon actually have options other than "keep praying Elon Musk doesn't go full Howard Hughes" or "just go back to paying far more money to Boeing or Lockheed for a fraction of the results".
roystgnr
28d ago
Hasn't enrollment in English programs actually dropped in the last few decades?
Apparently so! From a peak of 55K Bachelors' degrees per year in the late 2000s down to 40K in 2017-2018 (the latest data I could quickly find).
Sadly, that doesn't mean the employment prospects for English majors are actually much better.
This was sadly less surprising to double-check. Among new graduates, they were looking at 4.9% unemployment, 48.6% underemployment in 2023.
Just to be pedantically clear (I saw your comment on the chronological page and didn't realize it was correct in context until I looked at the parent), a null pointer dereference invokes well-defined behavior of bounded badness in Java. In C, a null pointer dereference is Undefined Behavior and so is still allowed to lead to arbitrary code execution, both in theory and in practice.
More options
Context Copy link