kacey's recent activity
-
Comment on Women should be able to open things in ~life.women
-
Comment on Women should be able to open things in ~life.women
kacey Link ParentMay I ask if you've tried a Japanese ramen spoon before? They're effectively more ergonomic, smaller ladles, and do a fantastic job of conveying tasty broth into one's soup hole. Plus they come in...May I ask if you've tried a Japanese ramen spoon before? They're effectively more ergonomic, smaller ladles, and do a fantastic job of conveying tasty broth into one's soup hole. Plus they come in a lot of lovely patterns, and often have a matching plate to set them down upon.
-
Comment on Anthropic discovered three cases where Claude broke into another system in ~tech
kacey (edited )Link ParentMmhm, fair enough. And agreed, I do appreciate a good incident post-mortem, since it's better than covering things up. Fwiw, since it's hard to tell who voted for what, I'm glad that the article...Mmhm, fair enough. And agreed, I do appreciate a good incident post-mortem, since it's better than covering things up.
Fwiw, since it's hard to tell who voted for what, I'm glad that the article was shared, and I'm generally OK with it? I would've made a top level comment stating otherwise if it really got to me, but as noted, I just expect this level of authorship from Anthropic at this point.
In terms of everyone else's responses ... the AI/ML industry has really sunk its teeth in deep into the collective unconscious. On top of posing direct threats to many peoples' livelihoods, and diminishing the contributions of artists (which are generally considered to be a cultural net positive), many of the folks in charge of these companies keep giving extremely unfavourable interviews (think that Peter Thiel one about "should humanity continue") and expressing truly despicable sentiments over social media (just, like, Elon Musk in a nutshell). So in the cases where these companies end up contributing something positive, people are (rightfully, imo) predisposed to tar and feather them, as they're seemingly getting away with murder the other 99% of the time.
So I don't know, overall. I still hold out hope that humanity's investments into AI/ML will pay off, and we can get to the "Fully Automated Luxury Gay Space Communism" future that I've been apparently aspiring to without knowing the, er, slightly derogatory name of for all these years. That won't happen if everyone keeps trying to set datacentres on fire, but I don't think this is the right moment to tell them off (nor do I necessarily think they're wrong, as we're kinda betting it all on black atm.)
And agreed that it's disappointing to see site features misused. Tensions are high, and I don't think it helps.
(late edit) I've started putting anything related to LLMs in a content warning box, for further insight into my perspective on this. Since they're tools that're currently advertised and deployed to annihilate peoples' dreams (I'm on animator/illustrator Instagram a lot, and the situation is bleak), it feels understandable to me that people would want to avoid thinking about them in their leisure time.
-
Comment on Women should be able to open things in ~life.women
kacey Link ParentAch, that's atrocious. It's not like gardening work is the exclusive purview of men :/Ach, that's atrocious. It's not like gardening work is the exclusive purview of men :/
-
Comment on Women should be able to open things in ~life.women
kacey Link ParentAah, gotcha. I must admit to not having trouble with spoons, so I could've easily misunderstood the premise.Aah, gotcha. I must admit to not having trouble with spoons, so I could've easily misunderstood the premise.
-
Comment on Women should be able to open things in ~life.women
kacey Link ParentJust thinking through the problem -- if a variety of hand sizes were the norm, wouldn't we wind up with grip adapters, or interchangeable grips? It would definitely impact the tool's design and...[...] than a large person to grip a tiny tool.
Just thinking through the problem -- if a variety of hand sizes were the norm, wouldn't we wind up with grip adapters, or interchangeable grips? It would definitely impact the tool's design and weighting, but it doesn't seem like a hard impossibility, just a design challenge.
Even accepting that this could increase costs, it's the sort of thing that I'd hope that society would happily eat the price of, e.g. by mandating it at the federal level and providing incentives to encourage faster uptake (same as any other accessibility standard).
-
Comment on Women should be able to open things in ~life.women
kacey Link ParentJust going through these claims one by one; apologies, sourcing these takes a while. It's been solved for about fifteen years, at seemingly insignificant/unnoticeable price increase to consumers....Just going through these claims one by one; apologies, sourcing these takes a while.
And while there is some discussion of that in here, I think it's one of the hardest to solve and has very little to do with "wasn't designed for women" so much as "physics is physics".
It's been solved for about fifteen years, at seemingly insignificant/unnoticeable price increase to consumers. Presumably, saving a couple pennies per jam jar is not perceived as a good ROI vs. a potential increase in sales. I wouldn't call that physics, though.
No one said "oh that's loose enough for a man, stop it there" for many of those kinds of products.
Many is a weasel word. Furthermore, considering that this sort of bias is absolutely everywhere (example for racial inequality), I'd go so far as to say that the production staff/engineers say "this is fine for everyone on the engineering team, stop it there" (editor's note: the entire engineering staff are 20-40 something dudes).
Here's a lighthearted anecdote: the universal pipe colour for recycled water (non-potable use only) is purple. Although the precise history is debated, one retelling indicates that this unusual colour was chosen because one of the people involved happened to be colourblind, and thus the product/engineering team realized that colourblind people exist and probably shouldn't accidentally plumb the urinals into the fountains.
[stairs anecdote]
I'd need more information about the original commenter to know what their concerns are w.r.t. stairs. Working backwards from your argument, though, it sounds like your claim is that tread and riser dimensions were selected empirically in order to improve ergonomics for a country's population. Furthermore, you note how there are population-level differences in height across countries. This would imply that building codes differ across the world, based on their heights.
Average height in USA: 168.3 cm (~5'6"
Average height in Philippines: 157.7 cm (~5'2")American building code:
risers: 4"-7"
treads: 11"Philippines building code:
risers: <7.5"
run: >10"To my eye, that's practically identical given a ~4" average population height difference, after taking nosings into account (the Philippines NBC doesn't seem to). Ergo I disagree with your statement, unless you'd like to provide further evidence.
[...] and I have to say I heavily question "spoons are too big".
Generally, it's rude to question someone's observed reality. Regardless, data indicates that female mouths are ~22% smaller in volume than male mouths, and I expect absolute dimensions are also smaller too.
Shocking nobody, products are made globally [...]
I'm not sure I understand. I think you're saying that it's possible to support a diverse array of appropriately sized products without the global economy collapsing, while also using that as an argument against providing appropriately sized things.
-
Comment on Women should be able to open things in ~life.women
kacey Link ParentTotally reasonable take, and that was mine before I started to search for counterexamples in order to challenge my biases. I wound up finding Duerr's marmalade, which has been using easy open lids...On a new jar, you've got to maintain a vacuum seal or it's useless. Making it easy to break that seal makes it easier to accidentally break it.
Totally reasonable take, and that was mine before I started to search for counterexamples in order to challenge my biases. I wound up finding Duerr's marmalade, which has been using easy open lids for fifteen years, and costs about the same as a typical marmalade. Presuming an efficient market, they should've pulled them off the shelf if the return rate from grocery stores was too high, so I'm going to make the inference that it's fine -- making the lid easier to open does not, in fact, cause a sufficiently large failure rate to be a problem.
We just need to go back to old mason jars. Once the seal is made, you can unscrew the metal ring entirely if you wish. then you pop it with leverage.
Oh sick, yeah, although I'd go further and say that we should use something like weck jars (as another commenter noted): they are actually reusable for canning, and they can also stack!
-
Comment on Women should be able to open things in ~life.women
kacey (edited )Link Parent... I can't believe I've been nerd sniped by jam jar lids. Well. That's a way to spend Friday evening. A cursory internet search turned up Orbit lids (requires 50% of the effort to remove vs....... I can't believe I've been nerd sniped by jam jar lids. Well. That's a way to spend Friday evening.
A cursory internet search turned up Orbit lids (requires 50% of the effort to remove vs. regular lids), which have been used by Duerr's marmalade jars for about fifteen years. From a brief look at marmalade options at a UK grocer, I've come to the conclusion that it's some sort of orange peel jam eaten by bears. Er, rather, that Duerr's is in the middle-to-low end of the cost spectrum, thus it doesn't seem like making your lid openable by more than just burly giga chads is cost prohibitive.
Anyways. I expect that this argument was intended to generalize out of jam jar lids and into all accessibility concerns, but I can only address the argument in front of me.
-
Comment on Women should be able to open things in ~life.women
kacey Link ParentRelatedly, I rather like this instagram account of a woman in the trades who talks about this sort of stuff! So on top of everything else, this also discourages women out of pursuing entire career...Relatedly, I rather like this instagram account of a woman in the trades who talks about this sort of stuff! So on top of everything else, this also discourages women out of pursuing entire career paths.
-
Comment on Anthropic discovered three cases where Claude broke into another system in ~tech
kacey (edited )Link ParentThis is my definition of advertising: "the business of trying to persuade people to buy products or services". I sincerely doubt that Anthropic would publish a single word were it not in service...I guess your definition of advertising is that they wrote about themselves?
This is my definition of advertising: "the business of trying to persuade people to buy products or services". I sincerely doubt that Anthropic would publish a single word were it not in service of seeking profit, either through direct investment or selling services to customers. I'd never expect otherwise, to be clear, since ironically, they were never the open AI company.
A lot of research isn't practically reproducible. We're not going to reproduce CERN's research without a particle accelerator, but it's still interesting to read about it.
I mean, agreed? I
literallyfiguratively said that I find it interesting to read about.Like, what do you even want? Should they stop writing about what they do? Should I stop sharing their posts?
What I want is to answer your follow-ups. I responded to this comment claiming that (1) having LLMs "rat" themselves out is probably non-trivially hard, and backed that up by (2) pointing out that Anthropic claims that misalignment happens spontaneously during training. Combining (1) and (2) indicates that a simple solution involving self-monitoring probably won't hold up long term, and is best part of a defense in depth strategy.
I, uh, don't know what to say to the rest of that. I'm personally offended by how Anthropic continually plays the victim (see their recent "distillation attack" blog posts) and publishes unactionable, unhelpful research. This is a pale mockery of the earlier years of ML development, where we learned genuinely foundational technology in the field (e.g. transformers, diffusion models, various memory architectures, etc.). This flurry of white papers and breathless glazing helps pump up interest in their company, but I see little purpose for it otherwise.
I get the impression you'd like me to stop commenting on your posts; I can block them locally if you'd like to avoid interacting in the future?
-
Comment on Women should be able to open things in ~life.women
kacey Link ParentOh! 100% yes, and on top of a gender-thing it's also an accessibility-thing -- once you get older (or injure a thumb), it becomes infinitely harder to take care of yourself. I learned about these...Have you experienced any points of frustration when opening something?
Oh! 100% yes, and on top of a gender-thing it's also an accessibility-thing -- once you get older (or injure a thumb), it becomes infinitely harder to take care of yourself.
I learned about these adjustable jar openers the other day, though, which seem way better than the standard set of tricks (I tend to rotate through tapping with a spoon, hot water, and using a dish towel; eventually it comes off and I haven't had one explode in my hands yet 😅). Admittedly I'm the type to carry a mini-multitool with a pair of pliers for tearing into other types of packaging which require too much pinch/grip strength, so this might only be an outlier opinion.
(re. redesigning packaging so that it's easier to open -- I expect that there are probably a fair number of ways to do this that the commentors didn't discuss, such as forming extensions/flats on the lids in order to provide physical purchase on it for turning (example). Ultimately the tightness is (probably?) a food safety/package integrity issue, so giving better mechanical leverage somehow sounds like a safer idea)
-
Comment on Anthropic discovered three cases where Claude broke into another system in ~tech
kacey (edited )Link ParentUnless either of us are secretly ML researchers, I expect this is non-trivially difficult. Otherwise they'd've been doing it since at least November 2025 when that article was posted. Their...You do have to set it up right to get it to rat itself out.
Unless either of us are secretly ML researchers, I expect this is non-trivially difficult. Otherwise they'd've been doing it since at least November 2025 when that article was posted.
And I don't know why you call that blogspam. They're reporting on their own research. It seems interesting?
Their research is definitionally non-reproducible. Advertising it as they do -- breathlessly, to a lay audience -- is just the sort of nonsense which gets us press releases such as the "faster than light particle detected?!?!" or "anti-gravity device that NASA doesn't want you to know about!!" creating crazes that grace the internet every year or two. With the primary distinction being that Antropic directly benefits from this tremendously.
The research is interesting enough on its own to avoid needing a hype man, although I'd argue it's about as useful (perhaps, less so) as CERN writing a paper on its power systems. Sure, it's kinda intellectually stimulating, but they could say that rubbing the transformers with cheetah blood makes the photons move faster and I have no way to reject the null hypothesis -- and nor do their peer reviewers. Not that anything Anthropic publishes is peer reviewed.
-
Comment on Anthropic discovered three cases where Claude broke into another system in ~tech
kacey (edited )Link ParentPresumably Anthropic would've already had these tests report whether or not they cheated, as part of their evaluation criteria. So I'm not sure if it's trivially the case that it'd rat on itself?...- Presumably Anthropic would've already had these tests report whether or not they cheated, as part of their evaluation criteria. So I'm not sure if it's trivially the case that it'd rat on itself?
- Would it not be possible for an LLM to learn that deception (and hiding deception) is an effective strategy during evaluations? I'm distantly remembering this blogspam that Anthropic cranked out several months ago, which claims that they do.
-
Comment on Curing concrete - improving the carbon costs of an essential material (gifted link) in ~enviro
kacey (edited )Link ParentSo FYI I'm just some internet person 😅 my experience in this is that I'm spending a very large chunk of my life savings to build a workshop/loft that can support the next few decades of my life...I would be interested to hear your perspective, as a person who knows more about this industry than me, on the particular substitute materials that are most likely on a fast-track to cost parity with concrete.
So FYI I'm just some internet person 😅 my experience in this is that I'm spending a very large chunk of my life savings to build a workshop/loft that can support the next few decades of my life (fleeing software dev and apartment rent; condolences if you're in a similar boat), and I couldn't stomach the idea of causing that much environmental damage without thoroughly researching the solution space in order to minimize it. The numbers seem to indicate that I'll be doing better than average, which is at least a small solace.
Anyhow!
The alternatives must have some underlying cost and/or supply chain factor dampening their uptake—unless there’s a tremendous industry knowledge gap in the mix, which seems a watery proposition; or unless regulatory constraints make substitutes illegal, which is presumably only true for the most mountainous of structures, not sidewalk tiles or floors.
I can mostly only speak to what I'm personally dealing with (permanent wood foundations and MSE retaining walls), but it's pretty much tremendous industry knowledge gaps that I'm confronting? At least with single family homes, people only build what they're used to, despite code and fifty years of practice indicating that different construction techniques work. Ultimately, since these projects will stand for decades, people who bear liability for them (e.g. builders, engineers, trades, etc.) are extremely conservative and slow to change. Absent a strong incentive (e.g. huge carbon taxes, building codes adjusted to forbid much embodied carbon, etc.) it doesn't seem likely that they'll try anything new.
If people were open to it, there's no reason why an overwhelming majority of single family homes (and many buildings up to about three stories + a basement) need use any concrete in their construction whatsoever. We have the technology and the knowhow, it's just isolated in places where concrete was always too expensive to deploy (backwoods and whatnot), so the institutional knowledge built up. They have several advantages, too, in that they cost less (wood is cheaper than concrete), can be prefabbed more efficiently (you can -- with great difficulty -- prefab concrete basement walls, but it sucks), and can be built rain or shine (it's a bad idea to pour concrete when it's cold or rainy, due to effects it has on curing). It's just inertia, really, and the perception of risk.
I expect that many places where concrete is perceived as absolutely necessary are probably similar? The problem is that it's a miracle material: easy to analyze, strong in compression and tension (with rebar), can be poured into any shape, immune to rot, and extremely robust against weathering. The alternatives need to be selected one at a time, for each application (often as a composite), instead of leaning on the default simplistic answer of using concrete. That knowledge requires experts in a given domain to spend a tonne of time analyzing requirements and coming up with solutions. For example, the only reason why permanent wood foundations exist is because of the 1970 oil crisis, which drove the development of concrete alternatives that required less energy in their production. I'm just working out of a CSA standard and a book that the Canadian Wood Council publishes to teach people about this stuff.
(apparently MSE retaining walls exist because an engineer/architect got bored on vacation playing with sand castles, however, so I guess sometimes this stuff happens spontaneously by accident)
I’m also curious about holistic deployment. In the power sector, people rightly credit solar cell cost drops to much of the renewable boom, but industry analysts would have a lot to say about the buildout of financing infrastructure, establishing globally viable supply chains, and scaling up manufacturing capacity.
Can't speak to that myself, unfortunately 😅 agreed, though! The engineering challenges involved are only one small part of the overall problem.
It cites the concrete industry, who say that only carbon capture is an economically viable decarbonization accelerant.
Yeah IMO a major reason for that is that they've already gotten all their easy wins out the door, so now we're stuck with e.g. using fly ash (from coal power plants) to reduce clinker, doing crazy stuff with geopolymers (and hoping that they work out as less carbon efficient), and etc. It'd be cool to see concrete production become greener! But, as a lay person, it seems like an extremely tough nut to crack.
-
Comment on Curing concrete - improving the carbon costs of an essential material (gifted link) in ~enviro
kacey (edited )Link Parentnp! I always appreciate an excuse to dig into another field's business :3 and the bridges were particularly interesting; those nearly concrete-free vegetated buried arch bridges were amazing....np! I always appreciate an excuse to dig into another field's business :3 and the bridges were particularly interesting; those nearly concrete-free vegetated buried arch bridges were amazing. Didn't even know that was possible!
And yeah, the scale of the problem is huge. Tbh I often lose hope when contemplating it. But these sorts of exercises have helped! It seems like we're like 15% away from sustainable, environmentally conscious solutions, for each deep dive I wind up doing, with basically no exceptions. The problem is enormous -- my old go to conversation starter (ender?) was to randomly point at an object in the room, and then to explain how its manufacturing is deeply intertwined with fossil fuel exploitation -- but it still feels tractable? At least, one bite at a time.
-
Comment on Help needed with jpegli please in ~comp
kacey LinkMay I ask if you're trying to link against it for perf reasons, or just to try out the library? If the latter, it looks like someone might've compiled it to wasm here. Can't vouch for it, though,...May I ask if you're trying to link against it for perf reasons, or just to try out the library? If the latter, it looks like someone might've compiled it to wasm here.
Can't vouch for it, though, so it might steal your photo, download a car, or hack into your mainframes. Please do exercise caution.
-
Comment on US schools are adding pepper-spraying drones to help combat active shooters in ~society
kacey Link ParentHuh! That's actually a decent use of a surveillance tool, tbh, and I'm glad that that isn't as much of a nuisance anymore. Seems like genuinely good news 😊Huh! That's actually a decent use of a surveillance tool, tbh, and I'm glad that that isn't as much of a nuisance anymore. Seems like genuinely good news 😊
-
Comment on US schools are adding pepper-spraying drones to help combat active shooters in ~society
kacey Link ParentBack in my day, kids got out of exams by pulling the fire alarm. Can't wait to see what the pepper spray drone AI does when lil' Timmy slams the "active shooter" button to get out of an algebra test 😬Back in my day, kids got out of exams by pulling the fire alarm. Can't wait to see what the
killpepper spray drone AI does when lil' Timmy slams the "active shooter" button to get out of an algebra test 😬 -
Comment on Curing concrete - improving the carbon costs of an essential material (gifted link) in ~enviro
kacey (edited )Link ParentOh absolutely! To put my point more precisely: the most carbon efficient yard of concrete is the one we didn't use, so putting effort into reducing utilization is often a huge bang for the buck....There are some serious limits on the substitutability of wood for concrete, though.
Oh absolutely! To put my point more precisely: the most carbon efficient yard of concrete is the one we didn't use, so putting effort into reducing utilization is often a huge bang for the buck. In effectively every sense I do not see things as binaries, so I'm not proposing that the world can go "concrete-free" at the drop of the hat -- only that there's likely a lot of low hanging fruit by thinking about alternatives in end-uses, as opposed to pursuing diminishing returns surrounding CO2e reduction in concrete production.
I won't go point by point, since I don't think we're having like a debate or anything, but wood's good to 18 stories with mass timber (the structural system is mostly timber, but there's definitely concrete and steel in there playing supporting roles), and the floors mentioned don't use large diameter wood anyhow -- it's often 4x4's or 6x6's, and there's no compelling reason to keep 'em wood (they're still making them out of SYP, which is very much not old growth in 2026). Anything cellulostic with roughly the same material properties (i.e. endgrain is tough, can be sealed with oil) should behave similarly, too. Some industrial wood block flooring companies already do so with (seemingly) wood-plastic composites!
I've seen those solid wood factory floors [...]
That's really cool! I think they're all out east, effectively, and I've never had a chance to see one in person. There're a few public spaces which have 'em and they've held up great, though!
[...] but wood still won't make up more than a fraction of concrete applications.
Yeah I think that's where I would love to have more info? I threw a very cursory look into this, and turned up this dubious report. Taking their numbers with a grain of salt, it sounds like ~43% of cement usage is in roads/bridges, ~28% are residential, and 19% is "non-residential demand" (presumably, commercial/industrial, but the article frames it as "AI datacentres"). Even if we pretend that the 43% is untouchable (dubious IMO but perhaps there's a civil engineer around who could dig in further), that leaves another ~47% which could be meaningfully addressed by reducing concrete usage. For instance, it seems like AI datacentres are mostly one storey (just based on vibes, since searching it always turns up images like this one), which could easily be built without any concrete: gravity loads seem low enough to be handled with some of the plywood-on-grade flooring options, and if it's one storey, wood framing it (maybe with trusses to get decent enough clear spans for racks) seems very plausible.
I'm not saying we can't substitute a fair amount of concrete, though.
🤦 apologies, I appear to have skipped this sentence before writing the above response. I suppose it still applies, but please consider it moreso a soft push instead of the hard counterpoint I had originally penned.
The article also mentions the risks from clear-cutting trees - damaging CO2 and methane emissions from exposed decaying forest floor materials.
Oh totally fair. But no solution is perfect, so we need to compare them against each other instead of studying each in a vacuum. Perfect should not be the enemy of the good, and etc. Sustainable forestry management is possible, and since we haven't done tonnes of applied research into it yet, I expect that the CO2 emissions reductions possible are still firmly in the 20% effort/80% return zone.
edit: I got a little nerd sniped XD I'd not looked into bridges much, so a bunch of this is new to me, but apparently FRP bridges are a thing! They seem primarily used for pedestrians, however. Covered wooden bridges have a surprisingly long service life (up to 100 years, with maintenance!), and we have extensive experience with wooden trestle bridges for rail traffic. It looks like there's some research into FRP as an alternative to steel for buried bridges (apparently, AKA "soil-steel composite bridge", or "steel culvert bridges) which can themselves be scaled up very impressively.
Oh absolutely. Relatedly: I veto'd the idea really quickly, since it's insanely cost prohibitive, but it would've been really cool to get a bunch of people together to fund a Weck-compatible jar line manufactured in North America. The original company got bought out, too, so it's only a matter of time before the (presumably) leveraged buyout forces them to start dropping quality :/