whbboyd's recent activity
-
Comment on Waymo pauses Atlanta service as its robotaxis keep driving into floods in ~transport
-
Comment on Bolt CEO says he let go of his entire HR team for creating problems that didn’t exist: ‘Those problems disappeared when I let them go’ in ~tech
whbboyd Link ParentYeah, I've said it before: it's appropriate as a line employee to be skeptical and critical of HR, but the alternative to HR performing HR functions is for management to perform HR functions,...Yeah, I've said it before: it's appropriate as a line employee to be skeptical and critical of HR, but the alternative to HR performing HR functions is for management to perform HR functions, which is much, much, much worse.
-
Comment on GitHub confirms breach of 3,800 repos via malicious VSCode extension in ~tech
whbboyd Link ParentThis old joke seems apropos.This old joke seems apropos.
-
Comment on Tildes Survey #5: Pineapple on pizza? (Results) in ~talk
whbboyd LinkOtto Pizza in Cambridge, MA used to serve a pulled pork and mango pizza, sort of an uber-classed-up Hawaiian. It wasn't my favorite of their flavors, but I liked it. I'm curious what the pineapple...Otto Pizza in Cambridge, MA used to serve a pulled pork and mango pizza, sort of an uber-classed-up Hawaiian. It wasn't my favorite of their flavors, but I liked it. I'm curious what the pineapple haters think of that topping combo.
Personally, I think pineapple is a fine topping. It's not one of my go-tos, but I don't mind it. I admit I don't get the memetic hate for pineapple in particular, of all the weird things people put on top of pizza. Like, why pineapple and not fresh tomato (which has the same textural problems—you gotta pre-cook your tomatoes—as well as being gross and tasteless at anything less than the peak of the local season) or anchovies (just seems weird to me tbh) or, like, baked ziti (this is actually delicious, don't knock it 'til you've tried it)? People can obviously have their individual preferences, which I'm not judging, but the whole cultural thing about pineapple on pizza is strange to me.
-
Comment on Medium term cold storage options? in ~comp
whbboyd LinkWhen I was looking into this a few years ago, the consensus answer seemed to be that there's not a great option, but archival-quality optical media is probably the best one. (Flash doesn't have...When I was looking into this a few years ago, the consensus answer seemed to be that there's not a great option, but archival-quality optical media is probably the best one. (Flash doesn't have great longevity offline, and hard disks are something of a question mark—and the typical failure mode isn't "some degradation", it's "this complicated mechanical device has broken and does not work at all".) Consensus at the time seemed to be that archival-quality media was expensive out of proportion to the increase in quality, and there was a whole lot of discussion of (mostly non-actionable) concerns like the original manufacturer of a piece of media. My conclusion ended up being:
- Optical media is cheap enough that just buying the "expensive" stuff is probably worth it.
- Burn multiple copies, confirm that they are readable (coasters are pretty uncommon these days, but not unheard-of), and distribute them geographically.
- Try to make sure they're stored appropriate, i.e. in a case, in the dark, not too humid.
My schema is mostly to have encrypted backups stored online (in Backblaze B2, in my case), with encryption keys, a copy of my password vault, and a handful of other useful things on the backup disks. In the end, I've got a few dozen megabytes on my backup discs (on DVD media, because that's all that was available, lol).
For long-term storage, the answer is, oddly enough, to keep it online and monitored. Individual units of storage media are pretty fragile, but a NAS with a handful of drives and someone checking up on it regularly will keep data stored more-or-less indefinitely. (Online storage is much easier to accidentally delete stuff off of, of course, so it's not a panacea.)
-
Comment on From $250 million megadeal to empty offices: the unraveling of Bad Robot in ~movies
whbboyd Link ParentBut not Abrams Trek, obviously. ;)Very Trek.
But not Abrams Trek, obviously. ;)
-
Comment on Interesting material types for fantasy resources/macguffins other than crystals or metals? in ~creative
whbboyd LinkSome inspiration, maybe, from The Elder Scrolls III: Morrowind (a great source of inspiration for all manner of Weird Fantasy, tbh): Chitin: insects have armor. If you have giant insects in your...Some inspiration, maybe, from The Elder Scrolls III: Morrowind (a great source of inspiration for all manner of Weird Fantasy, tbh):
- Chitin: insects have armor. If you have giant insects in your setting, then it's an obvious choice for your characters to, uh, "debone" them and use the segments of exoskeleton in protective clothing.
- Bonemold: in-setting, this is essentially a composite whose matrix is bone. Metal AF. There's a relevant in-game book series (CW: body horror).
- Ebony: neither a particularly dark-colored wood nor, as most in-universe characters believe, a volcanic compound (it's often described as a "glass", but it clearly behaves more like a metal, and there's another material which is actually called "glass" and is non-ductile); rather, it is the congealed blood of a god, left in deposits where it fell to the ground when his heart was ripped from his chest and thrown across the continent.
-
Comment on Does anyone want to buy an unused Pixel 10? in ~tech
whbboyd Link ParentIt definitely is not, necessarily, though it's possible it varies on the model of phone. Xperia devices can definitely be carrier-unlocked but bootloader-locked (which makes them useless for my...It definitely is not, necessarily, though it's possible it varies on the model of phone. Xperia devices can definitely be carrier-unlocked but bootloader-locked (which makes them useless for my purposes, and yeah, confuses a ton of people). Pixels may conflate those settings, but they don't need to.
-
Comment on Why are we still doing this? in ~tech
whbboyd Link ParentTime to go back to calling compute "flops"? ;)Time to go back to calling compute "flops"? ;)
-
Comment on I tried ranking my albums out of five stars - I think I've gotten it wrong. Thoughts? in ~music
whbboyd LinkThe lack of five-star albums is because as you average more things, by definition they tend to the mean. If you averaged your entire music library, would you be surprised for it to come out at a...The lack of five-star albums is because as you average more things, by definition they tend to the mean. If you averaged your entire music library, would you be surprised for it to come out at a 3?
Now, of course, albums are not formed by taking a random sampling of songs with normally-distributed ratings. There probably is a signal here; do you generally prefer your 3.5-star albums to the 2.5-star ones? The challenge is that averaging the song ratings weakens this signal. Mathematically, there are a few approaches you could take:
- Renormalize. Compute your album scores, then scale them so the top score is 5 and bottom is 1.
- You could flatten out the normal distribution while you were at it here if you really wanted to. That would have the effect of shoving a bunch of high 3s or low 2s into the 4 and 1 buckets, respectively.
- Pick a different aggregation function than "mean". "Max" would be one obvious choice; you'd then have 5-star albums, but probably no 1-star and few 2-star albums.
- You could do something complicated here, like weighting ratings differently in the average; count 5-star ratings much higher than 3-star, for instance.
…But the mathematical "fix the stats" approach is probably not what you want regardless. As you note, an album with all 3-star songs and one with half 2 and half 4 are quite different in subjective feel. (And also, people tell me that albums can be constructed as a cohesive more-than-the-sum-of-its-parts whole rather than just a collection of independent songs, though my listening habits don't allow me to observe that.) What you might want—and you kind of hint at this already—is to give albums their own rating, rather than trying to derive it from the component song ratings.
- Renormalize. Compute your album scores, then scale them so the top score is 5 and bottom is 1.
-
Comment on What are some bands you regret not seeing live (or, just never had the chance to see in the first place)? in ~music
whbboyd LinkI'm not particularly into rock/pop concerts, so I don't have any particular regrets on that front. However, my cousin (once removed) was old enough to have attended Woodstock. He did in fact go,...I'm not particularly into rock/pop concerts, so I don't have any particular regrets on that front. However, my cousin (once removed) was old enough to have attended Woodstock. He did in fact go, and had a good time, but by Monday morning he and his buddies were worn out, the weather sucked, and they were worried about the epic traffic jam that was sure to ensue as tens of thousands of spectators tried to depart from a field in the absolute middle of nowhere in upstate New York. So they left before the last act.
The last act at Woodstock was Jimi Hendrix.
He still speaks regretfully of it more than fifty years later.
-
Comment on Why are American passenger trains slow? in ~transport
whbboyd Link ParentThe European coast is also all connected to itself, while taking a boat from New York to LA requires crossing the Arctic Circle, the Panama Canal, rounding Cape Horn (infamously hazardous to...The European coast is also all connected to itself, while taking a boat from New York to LA requires crossing the Arctic Circle, the Panama Canal, rounding Cape Horn (infamously hazardous to shipping), or circumnavigating the globe.
-
Comment on Ladybird chooses Rust as its successor language to C++, with help from AI in ~comp
whbboyd Link ParentOh, don't get me wrong, there is ample evidence that humans cannot write correct C or C++, that the errors they tend to write are catastrophic (arbitrary memory exposure or code execution), and...Oh, don't get me wrong, there is ample evidence that humans cannot write correct C or C++, that the errors they tend to write are catastrophic (arbitrary memory exposure or code execution), and that Rust practically prevents those errors. Continuing to use C/C++ for security-sensitive projects is indefensible. I'm just pushing back against the idea that Rust prevents all bugs¹.
If the correctness of your program doesn't really matter², then I guess an LLM is more likely to produce output which produces wrong output but doesn't have surprise RCEs in it using Rust than using C. But for a web browser, the correctness of your program is a security issue. And so the fact that Rust effectively saves you from one particular class of very, very bad issues doesn't make it suitable as a target for LLM output.
¹ In my experience, this idea is not very prevalent among Rust users, but does seem to be quite prevalent among the hypothetical Rust users who exists in the minds of people who use phrases like "Rust Evangelism Strike Force".
² Though of course, if the correctness of your program doesn't really matter, surely
exit(1)is simpler, clearer, and more robust. -
Comment on Ladybird chooses Rust as its successor language to C++, with help from AI in ~comp
whbboyd Link ParentNah, not really. It won't output trivial buffer overflows and RCEs like it would in C, but the entire rich universe of logic errors remains open to it, and in a web browser, nearly every logic...it's entirely possible [safe Rust] is well-suited to AI code generation
Nah, not really. It won't output trivial buffer overflows and RCEs like it would in C, but the entire rich universe of logic errors remains open to it, and in a web browser, nearly every logic error is a serious security issue.
-
Comment on I hacked ChatGPT and Google's AI – and it only took twenty minutes in ~tech
whbboyd Link ParentThe model—the ginormous matrix that encodes the training data—is extremely expensive to generate, and has a cut-off as of the point at which the organization generating it stopped scraping...The model—the ginormous matrix that encodes the training data—is extremely expensive to generate, and has a cut-off as of the point at which the organization generating it stopped scraping training data and started training the model, something which they do, idk, every few months at most (because of the extreme expense). With no additional "context" (i.e. token string prefix), this is all the model "knows".
However, most model operators provide mechanisms for their models to trigger web requests and inline the responses into their token prefix. This is what happened here: the models don't "know" anything about hot-dog eating tech reporters, but when asked, they search for it, inline the author's blog post into their token prefix, and then repeat it with minor paraphrasing.
-
Comment on Tesla 'Robotaxi' status check eight months in: a complete joke in ~transport
whbboyd Link ParentLemme preface this by stating that Tesla is slimy, Musk is a monster, and there are fundamental problems with self-driving tech and policy we don't have solutions for. This is not advocating for...Lemme preface this by stating that Tesla is slimy, Musk is a monster, and there are fundamental problems with self-driving tech and policy we don't have solutions for. This is not advocating for the devil; I'm trying to strengthen the offensive against him.
Nine crashes in approximately 500,000 miles: one crash every 55,000 miles. Human drivers average one police-reported crash every 500,000 miles.
This comparison isn't apples-to-apples; human drivers have loads of minor collisions they don't report to the police (or any other authority, e.g. their insurance company). I'd assume animal collisions almost never result in a police report (I've certainly never heard of someone alerting the police that they ran over a deer). In my 50-100k miles of lifetime driving, I've had two vehicle-vehicle contacts, neither of which resulted in a police report. (Obviously, both were minor enough not to require any repair of either vehicle, and lest you consider me reckless, neither was my fault. One did result in an insurance claim.)
AFAICT we don't have a rigorous metric we can use here (and, of course, we don't know what Tesla isn't reporting to the NHTSA). I'd assume insurance claims would be at least more comprehensive; attempting to cobble together stats from a few different sources suggests roughly one collision insurance claim per 50k miles driven, though of course a significant portion of those will be for minor collisions with no injuries or functional damage to any property.
In any case, whether Tesla's automated accident rate is on par with or much worse than that of human drivers, it immediately makes two very strong points against them:
- The common self-driving booster point of "it just has to be meaningfully better than human drivers" is moot. All the other problems with that point aside, Tesla's automated driving isn't meaningfully better than human drivers.
- They have absolutely no business doing testing and R&D of demonstrably-unsafe tech on public roads. In a just world, the company officers would be prosecuted for reckless operation (plus property damage, injuries, etc. from all the accidents) and the company barred from selling or operating automated driving tech. We'll see what we end up getting.
-
Comment on In a blind test, audiophiles couldn't tell the difference between audio signals sent through copper wire, a banana, or wet mud in ~tech
whbboyd Link ParentCRT displays are silly for a variety of reasons, but nobody with even half-functioning eyes will ever confuse the image from one with an LCD (no matter how sophisticated a CRT simulation shader...CRT displays are silly for a variety of reasons, but nobody with even half-functioning eyes will ever confuse the image from one with an LCD (no matter how sophisticated a CRT simulation shader you're running). I would say gaming on a CRT is more akin to vinyl—maybe dumb, but not literally undetectable the way a lot of audiophile equipment is.
-
Comment on Hair loss open discussion in ~talk
whbboyd (edited )Link ParentI know you didn't ask for advice, but as someone who also had a side part veeeeeeery gradually turn into a combover over about a decade of balding, I'd say: switch to a short haircut sooner rather...I know you didn't ask for advice, but as someone who also had a side part veeeeeeery gradually turn into a combover over about a decade of balding, I'd say: switch to a short haircut sooner rather than later. I look way better bald with my hair all trimmed to a centimeter or so than I ever did balding with longer hair.
I sympathize a lot with not feeling like you can get a straight answer about thinning hair from your barber, though. I suspect a lot of their male customers are very very sensitive about it and would react poorly to honest feedback even if it were solicited.
-
Comment on Hair loss open discussion in ~talk
whbboyd LinkMy dad went bald in his early twenties. I went bald in my late twenties. I never stressed about it because I could see it coming from a mile away. (There was an awkward period where I was bald-ing...My dad went bald in his early twenties. I went bald in my late twenties. I never stressed about it because I could see it coming from a mile away.
(There was an awkward period where I was bald-ing but not yet bald, where my pre-balding hair style very gradually morphed into a combover. These days I get my hair cut short, but making that transition felt awkward, and I probably should have done it earlier than I did.)
I do have extremely bodacious facial hair, which as a cis man more than makes up for any self-image issues my baldness could otherwise cause. =)
I'll second @DefinitelyNotAFae's note about sunscreen or a hat. I wear a hat outdoors religiously. Cancer risk aside, sunburning your scalp is no fun at all.
-
Comment on 'Right-to-compute' laws may be coming to your state this year in ~comp
whbboyd Link ParentBut not for either of its two representatives in the House (of Representatives). They're "Montana Representatives". ;)use "a Montana senator" for one of the two representatives Montana has in the US Senate
But not for either of its two representatives in the House (of Representatives). They're "Montana Representatives". ;)
Stopping erroneously is, while much safer than driving into a flood, not ideal. And the cars can't stop unconditionally for puddles, or any schmuck with a garden hose can DoS them. (Furthermore: and make it look like an accident.)
To be fair here, "don't drive into floods" is unironically challenging for human drivers, as well. It's tough to judge the depth of a body of water, so a great deal of the "is this a dangerous flood or a shallow puddle" decision comes down to context and heuristics. What's your extrapolated topography of the street? If the water is flowing, what does it look like? Have any other cars gone through it? Is there a schmuck with a garden hose guffawing next to the road? Etc.
There is definitely a reason these trials have heretofore been in socal or the southwest, though.