kldg

I was a bit shocked by headline, given how poorly ARM and x86 compares to RISC-V in speed, cost, and efficiency ... in the MCU space where I near-exclusively live and where RISC-V has near-exclusively lived up until quite recently. RISC-V has been great for RTOS systems and Espressif in particular has pushed MCUs up to a new level where it's become viable to run a designed-from-scratch web server (you better believe we're using vector graphics) on a $5 board that sits on your thumb, but using RISC-V in SBCs and beyond as the primary CPU is a very different ballgame.

I'm kind of curious how you.... I guess, interpret the responses to when you send someone AI-assisted content. I previously thought "I don't care if it's AI or not; quality is quality", but I'm increasingly taking the position that I do care, and intentionally have started ignoring comments and especially product reviews where you get the formatted 2-4 sentence paragraphs with formal tone and rule-following. It's come to the point where as long as you don't write as poorly as Epstein, I want the errors. Actually, I'm getting so weird and romantic about it, that I think I'd argue having errors and unusual style shows an openness and vulnerability that's now a necessary gate price; like journalists have so many tools available to them, but they still make typos, factual errors in articles they have no business writing about, and fail to quote people properly -- that's great, I think.

This has been my experience, too. In dealing with hardware, I'm particularly pleased with how vision models are shaping up; it's able to identify what I've photographed, put it in a simple text list, and link me to appropriate datasheets. yday, it even figured out how I wanted to reverse engineer a remote display board for a just-released inverter and correctly identified which pin of which unfamiliar Chinese chip was spitting out the serial data I was interested in; all I actually asked for was chip IDs with a quick vague note on what I was doing. It doesn't help me solder faster, but it gets me to soldering faster.

A bit OT, but I would love to see some different methods of calculating economic productivity. After looking into how BLS calculates software productivity, I quit giving weight to the number altogether and it left me feeling a bit blue; they apply a deflator in part by considering the value of features (which they claim to be able to estimate by comparing feature sets and prices in a select basket of items of a category, applying coefficients based on differences); it'll likely never actually capture what's going on in AI unless Adobe decides to add a hundred new buttons "because it's so quick and easy to do." Their methodology requires ignoring FOSS (except for certain corporate own-account cases), too; if everyone switched from Microsoft365 to LibreOffice, US productivity as measured by BLS would crash.

BLS lays methodology out in a FAQ page on "Hedonic Quality Adjustment"[1], which covers hardware instead of software, but software becomes more reliant on these "what does the consumer pay" guesses at value (what is the value of S-Video input on your TV? significantly more than supporting picture-in-picture, at least in 2020).

[1] https://www.bls.gov/cpi/quality-adjustment/questions-and-ans...

it all takes years. it takes years for permitting to open up the power plant to run the chips. at the scale the Big 3/4 (google, amazon, microsoft, and meta-ish) are going, we don't actually have the capacity to BUILD the capacity, despite a forecast of just 1% national electricity consumption growth this year, partly because we were expecting electricity demand to slow down and for an orderly shutdown of our fossil fuel plants. we couldn't even fill >100GW of gas/coal turbine orders over the next 5 years if we had to, and we might have to, because some of our grids (notably PJM's) are forecast to be under their safety margin of over-production in the following years.

meanwhile, regional grid operators are faced with Big Tech driving tens of % of total power into private contracts where there's only one customer; they are making the decisions normally reserved for nation-states, right? reopening Three Mile Island sounded like a pipe dream a few years ago. I hear They have something like 50 more experimental, small-scale NPPs they want to fire up across the country in the next few years, too (but despite sounding like a big boon for energy, they're ~meaningless short-term in the face of how much demand we're looking at). -so this power (uh, literally) gets wrested away from the grid authorities and from what was largely the domain of government, to now be managed by techbros and a select few partners who will be reliant on their money; I'm sure that will work out fine.

anyway, part of the reason it does make some sense in the US for the government to push for more coal/LNG turbines, is because they're already there and we need them now; the permitting to un-mothball, prevent mothballing, or expand facilities, is far less arduous than what a company'd have to go through for a new facility (tho again, we don't have capacity to build all the turbines we require inside 5 years anyway). I'm not saying it's a good idea to start sending up more GHGs, but it's maybe better than pricing out electricity for residences and "real" industry. hey, who knows? maybe they'll simply build natural gas pipelines that don't leak this time.

-oh, and then there's the problem with these new datacenters disrupting the traditional power demand curve, because they don't really do as much peak draw anymore; their peak draw is approaching base load, as LLM batching (when a company has a bunch of stuff they want processed and can wait a day for it to run in "off-hours") is sold, and if unsold, that time can be used as training time; so the modern datacenter is a 24/7/365 organ; the heart, powering our society, Moltbook. the importance of this is it makes solar less financially attractive, because now we need to be able to bank more energy since more demand's shifting to overnight. we might also want to consider just getting the moon really, really hot? then we can get a truly substantial haul of lunar light for our panels. you know, we decided against nuking hurricanes again recently; maybe we could build some new ones and nuke the moon, a lot.

The LLM is probably also not going to launch into a rant about how they incorporate religious and racial beliefs into their life when asked about current heads of state. You ask the LLM about a solar configuration, and I think it must be exceptionally rare to have it instead tell you about its feelings on politics.

We had a big winter storm a few weeks ago, right when I received a large solar panel to review. I sent my grandpa a picture of the solar panel on its ground mount, covered in snow, noting I just got it today and it wasn't working well (he's very MAGA-y, so I figured the joke would land well). I received a straight-faced reply on how PV panels work, noting they require direct sunlight and that direct sunlight through heavy snow doesn't count; they don't tell you this when they sell these things, he says. I decided to chalk this up to being out-deadpanned and did not reply "thanks, ChatGPT."

I clicked expecting to see someone with a huge, very long hose extension and was disappointed.

yeah, I can say that except for elder areas (not necessarily dedicated facilities, but there are things like "RV parks" which cater mostly to older folks but also families; they usually have 10mph speed limits), I've never seen someone driving a golf cart around town while I've lived in MI, OH, or PA.

I do see people driving horse-drawn carriages, ATVs (probably illegally), snowmobiles (legally in some parts of MI during Winter or condition-dependent), and riding mowers (probably illegally) in and around towns, though. Very rarely, I see someone driving an e-bike; this rareness is mostly because they aren't allowed on the sidewalks here and there's no bike lane, so you need to drive and signal like a car, which is pretty awkward given how many e-bikes don't even come with real brake lights (though many falsely advertise red rear running lights as a brake light, which'd be illegal to drive unless you hand-signal whenever you brake).

While I'm sympathetic to this argument, I should point out patent time to expiration for medicine in the US is pretty inoffensive (relative to how bad it could be, like software patents), and we already have plenty of drugs for excreting excess. We get a big basket of drugs into public domain each year, and government would be wise to publicly celebrate this, I think; would help with the general sense of impending doom citizens feel.

Semaglutide molecule patent will expire in 2031 here (many caveats to this). For the most part, you can get any pill ~15+ years old for ~nothing without insurance, but associated devices like auto-injectors can extend this due to goofy rules; I expect execs thoughtfully considered medical patent law when deciding to initially trial and release GLP-1s as an injection.

it depends a lot on the application, I think, though certainly you can point to cloud services like Cloudflare or whatever Burger King was using to track how many times a clerk said "You Rule" (while capturing all customer audio data, which was then stolen by low-effort attackers) as high-risk; just because you don't feel the safety risks of outsourcing data to a black box on the cloud doesn't mean they don't exist, it just means you get to neglect them. when I headed IT at an SMB, I was given a lot of leeway, and our department had its own budget, so cutting out SaaS was a high priority so we could do more. if I were heading today with LLMs' present competency, I would have replaced much more, up to and including Salesforce which was draining the heck out of our budget despite us not doing anything technically interesting with it.

$40/head/year (including employees no longer with company) for a call metrics suite is low-stakes and relatively easy to replace what we want out of it, and this is an example of something we did replace with a $0 solution with my own abysmal-at-the-time coding skills. ~nobody's about to replace Microsoft suite, though (a couple replacements before me, they earnestly tried; there were still some laptops with OpenOffice on it; I admire them, but I'm not dealing with our sales team trying to figure out what an ODF is).

I love this "petty kingdom" budget model, by the way, as someone whose work personality could be described as "cheap analyst." I'm paying $40/month per head for Software X in your department, and I have an inferior replacement for $0/month/head which meets specs and which you can't quantify productivity loss for (essentially, it just looks ugly and feels bad). I can therefor cut that out of my budget entirely while meeting my obligations, and if *you* really want the decadent solution, *your* department can bear that cost. Either way, I get plenty more money to basically not have to be a dick (like charging careless employees for broken/stolen equipment, or getting an above-expectations solution for ADA employees); and sometimes, maybe some antennas show up on the roof which would be difficult to justify cost for if asked, but I'm way under-budget so nobody would.

not to say this "should" be the solution, but does Proton not offer easy email filtering & automation (e.g. skip inbox, delete)?

I ask because I haven't yet bothered to implement it on a from-scratch email server I stood up a couple weeks ago (just kidding; I wanted to brag about SPF, DKIM, and DMARC test passes from Gmail with both inbound and outbound encryption). I can say from this experimentation and using Google's Postmaster tool, though, that emails being reported as spam by users is *very* serious; Google's threshold is 0.3%; if just 0.3% of users report your email as spam, it's considered a policy violation and your emails are likely to go to spam or have delivery refused outright. idk what Proton's policies are. (edit: by extension, this means enforcing authorization of users is very serious; if someone abuses the service as an open relay, your whole domain is toast)

looks like you've made considerable changes since comments; all prices I checked were accurate (while nothing I checked on diskprices was). this looks genuinely helpful as this is something I look into myself by manually looking around, and is clean/easy-to-use; bookmarked it. only thing I might like on top of this is to be able to filter by if renewed or not, though they seem to often have it in the product title.

Strongly agree with this. I actually kind of wish I could go back to SMB IT (well, I mean I can, but I'm ~retired) with the knowledge I now have; I would've done things very differently/better. If it's small enough, you'll find yourself taking on all sorts of interesting facilities work; it can be pretty fun not knowing what the work day will be like. Some days I would paint, other days I'd be cursing at Active Directory, other days I would be working on the security system, other days I would be running new cables, and sometimes I'd just sit around and play video games. I left IT to do on-call overnight/weekend work in B2B tech support with the same company, where my role upgraded to sleeping through my shift.

interesting. I assume the things I host at home (which is everything) will be recoverable by family at time of my death if they're so interested. I left some instructions in will, but not for things I ought to, on reconsider, like "what" the thermostat is and how to access it beyond the web interface...

-but I do not plan for longer-term than that; I assume if kid cares, she'll bother repackaging it. I am curious about some of the non-book solutions. I think you might be able to rig up a solar solution, burying a box underground with an SBC/mini-PC with SATA or NVMe connectors; the SSD should be fully read from at least once a year. The solar should be degraded-but-fine after 50 years (excluding animal damage, which I find is pretty substantial if you keep them close to the ground... will want redundancy), but the battery will not be, but I haven't been around for long enough to really have a "feel" for how long LiFePO4/other cells will last... I wouldn't be surprised if they could go >20 years with very shallow discharge cycles in a degraded state (perfectly usable for having an SBC read drives once a year), but I'd be very surprised if you could get a standard cell to 50 years.

I tried getting a tape drive a few years ago. I bought a used one off Amazon, but HPE refused to give me the software for it; I wound up refunding it rather grumpily. It'd be pretty surprising if the tape drive lasted for 100 years (edit: interfaceable probably the bigger issue), but under careful conditions with ~no use, the tape cartridge itself should last 30-50 years (with no guarantees; you'll want redundancy on different batches/brands).

I would also add software solutions like PaperBack to the pile of paper storage ideas. Essentially, you create a giant compressed QR code printed to a sheet of paper, and this is like going from storing audio as uncompressed audio on CDs/DVDs to mp3 (DVD-MP3 was underrated!); instead of storing 500 words, you can store an entire book. There's not terrible complexity here, so it should be trivial to reverse-engineer, especially if you include a couple page of notes on how the algorithm works. -or heck, in 100 years, maybe a 50-year-old LLM can simply look at it and read it like you wrote words.

also essentially how many large news organizations have pivoted. $520/year for WSJ, $400/year for Bloomberg (excluding terminal-only news and other extras, of course), $390/year for NYT, $120/year for WaPo (with exclusions). For only $2,500 or so a year, you can have a balanced stream of news and journalism. -But not your household; you need to pay extra for family plans.

(or you can do what most people do)

it strikes me as a dangerous time to try going to court over this as politics are currently aligned to fight back with new laws overriding court interpretations, at least in the US. God knows what's happening in China; afaik, it's a free-for-all outside requirements to avoid "sensitive topics". Between US and China, you have nearly all of the "top 100" LLMs.

Very neat! I've thought about this with frontier models because they're ignorant of recent events, though it's too bad old frontier models just kind of disappear into the aether when a company moves on to the next iteration. Every company's frontier model today is a time capsule for the future. There should probably be some kind of preservation attempts made early so they don't wind up simply deleted; once we're in Internet time, sifting through the data to ensure scrapes are accurately dated becomes a nightmare unless you're doing your own regular Internet scrapes over a long time.

It would be nice to go back substantially further, though it's not too far back that the commoner becomes voiceless in history and we just get a bunch of politics and academia. Great job; look forward to testing it out.

Can confirm no-BYOD policy is typical. I had to whine directly and without invitation to school principal to get an exemption for daughter. The trouble with no-BYOD is the kid must bring the school-controlled Chromebook home and connect to the home network for homework (which often requires Chromebook). Many US middle and high schools have an IT department of 1 or 2 people; it introduces abuse risk I think schools in general are not appreciating.

I see the problem with Chromebooks and cloud stuff more generally as being that it's difficult to see the productive use-case of programming outside just shuffling a bunch of data around. If your program's not actually doing something useful, it seems like it'd be difficult to imagine a career in it. -But if a kid can get a relay to trigger via button and then maybe via web interface and then maybe automate it, I think that opens the world of hacking up to them. You know, for $10, they can have a fully-solar (w/battery) thermometer or whatever they want -- the thermometer can feed a thermostat to energize a relay coil to start a heater or whatever.

-But I might be outlier, because in school we had robotics class a lot of kids were pumped for, but I was confused because we never did anything useful with them; it was more like an art class, except at least in art class we baked ashtrays for our parents. -But what am I supposed to do with a 5-watt robot that follows yellow tape?

yes, this is also largely how I use it; suiting workflows to my taste in a sort of "artisanal" way. most recent tool is a Windows "omniterminal" for Serial/SSH/SFTP, but there is a handful of bugs and missing wiring left (serial in particular is still in poor shape). 4MB .7z archive; portable binary; Rust source included: https://archivalcopy.com/content/category-utilities/category...

I also agree with comment that it seems silly to try widely releasing something you could just ask an LLM to make. Instead of something taking a team months or years, you can knock out most things in a weekend. I think vibe-coded stuff is best for hobbyists and experimentation.

Bless you and your family for all time and beyond. Having to talk to someone before I even get a price to compare, or a demo, drives me mad, and then a week later you get their contract and find they claim ownership of everything your company uploads to them -- all that time down the drain, and the salesperson never read the contract so they don't know what to say. Then there are the smaller companies with unwritten policies -- we used to get call metric software from a small Swiss outfit, but I discovered we were billed based on how many employees we've ever had, not based on current employees, with no method to delete terminated employees from the database -- on what planet do you expect someone to pay a recurring expense in perpetuity for someone who showed up for training one day 5 years ago and was never heard from again? I was so mad when they gave us the renewal price, we made our own replacement software for it.

Anyway, long story short: I now require the price and details before I'll even consider talking to a salesperson, not the other way around. Might actually be a good job for an AI agent; they can talk to these sales bozos (respectfully) for me.

Rust works very well for vibe-coding because the compiler is typically very specific and a bit picky; helps keep it out of traps. I like using WASM with Rust as originator when vibe-coding because AI models are still very "sloppy/forgetful". There is one thing you should still definitely do beforehand, and that's pick out which crates you want and write up the initial Cargo.toml for it, because frontier models have a lot of trouble with fast-updated libraries.

What frontier models also excel at is writing their own libraries and skipping third-party dependencies. It's very easy for a human to just pick up a bloated 750kb library they're only going to actually use 15kb worth of its code for, BUT that library can serve as a valuable implementation model for someone very patient and willing to "reinvent the wheel" a little bit, which is definitely going to be AI and not me, because I just want to point to a black box and tell it what to do. For big things like web server, I'm still defaulting to Axum, but for things like making a JS soundbank parser/player or a WebGL2 mp4 & webm parser/demuxer & player, these are tasks frontier models are good for with the right prompting.

To an extent, maybe counter-intuitively, I think the big thing we'll see out of AI is an explosion of artisanship -- with humanoid robots in 2040, perhaps, our households may be making their own butter again, for example.

This website has guidelines? I have a friend who vomits out what an AI's response to something is to current events; drives me mad, but I resist being a dick about it (to their face).

I don't recall any instances where I've run into the problem here, maybe because I tend to arrive to threads as a result of them being popular (listed on Google News) which means I'm only going to read the top 10-50 posts. I read human responses for a bit before deciding if I should continue reading, and that's the same system I use for LLMs because sometimes I can't tell just by the formatting; if it's good, it's good - if it's bad, it's bad -- I don't care if a chicken with syphilis wrote it.

yes; I think it depends a lot on how accepted the icon is. Every few months when I have to open ST's CubeProgrammer, my brain substantially deteriorates because they don't use text and the icons for their main tabs are not always understandable to me. (and the thing's otherwise a relic from 20+ years ago)

X, volume & mute icon, disk icon, upload/download icon -- these are all fine and good; you don't need to spell those out for me because their use is so widespread that even if I didn't know what they meant, it'd be very useful to learn. I have no idea what a generic "integrated circuit" icon means, though, and I doubt anyone will use it elsewhere to mean the same thing, so I just click around randomly until I find what I'm looking for, like I'd do with the previously-unlabeled 6-switch panel in my living room where the positioning of the switches have no obvious relation to the physical placement of the ceiling lights.

I think Apple's menu actually shows exactly what I want and expect; show icons if it'll help me, don't show them if they wouldn't; though in come cases, I think Apple could apply some more icons (like for "Stop", there are a few good choices for that).

Good luck, bud. I found I wasted a lot of time in my youth and only stopped as I aged; took time to find purpose, I think. That purpose also evolves and becomes both "wider" and "kinder" with age, I think; to move from zealotry into a kind of zen progression.

def not my wheelhouse, but I assume easiest way to keep bad guys out is to use copyleft license and only enforce against bad guys. despite what some say in defense of billionaires, you don't actually need to enforce every violation of your rights; rights don't disappear under law just because you don't use them.

I guess it depends on what you do. I do python, rust, and web frontend in Windows. I have a personal bias against Docker, which'd otherwise be the primary WSL draw for me since if I want/need Linux, I can SSH into the majority of the machines in my house.

I'll throw out my unpopular opinion/experience here, too: I haven't liked any "desktop experience" I've seen or used for a Linux distro, and they all look and feel very similar to me: foreign, basic, and difficult for me to tweak and produce with. I greatly dislike the React stuff both on the web and in Windows, and use Classic Shell, which I'm satisfied with. Windows is easy to customize and almost everything can be tailored without even needing a reboot, many even with registry options already made and just waiting for a bit to be flipped.

It helps my puny, smooth brain, too, to just think of Windows being graphical and Linux being text-based; helps me remember what I'm doing.

yeah, I'd guess the person who made this decision just wanted to cover their butt. it got a good laugh out of me reading the caption, though. "hmmm. who could use this distinct branding so many people are familiar with? it looks a bit orange; maybe it's Home Depot."

I lean toward the opinion there are certain things people (especially young people) should be steered away from because they tend to snowball in ways people may not anticipate, like drug abuse and suicide; situations where they wind up much more miserable than they realize, not understanding the various crutches they've adopted to hide from pain/anxiety have kept them from happiness (this is simplistic, though; many introverts are happy and fine).

I don't think I have a clear-enough vision on how AI will evolve to say we should do something about it, though, and few jurisdictions do anything about minors on social media, which we do have a big pile of data on, so I'm not sure it's worth thinking/talking about AI too much yet, at least as it relates to regulating for minors. Unlike social media, too, the general trajectory for AI is hazy. In the meantime, I won't be swayed much by anecdotes in the news.

Regardless, if I were hosting an LLM, I would certainly be cutting off service to any edgy/sexy/philosophy/religious services to minimize risk and culpability. I was reading a few weeks ago on Axios of actual churches offering chatbots. Some were actually neat; I hit up an Episcopalian one to figure out what their deal was and now know just enough to think of them as different-Lutherans. Then there are some where the chatbot is prompted to be Jesus or even Satan. Which, again, could actually be fine and healthy, but if I'm OpenAI or whoever, you could not pay me enough.

in defense of this comment, you do need to do a heck of a lot of preparation (including psychologically) to do anything publicly anymore. wild west days are long gone, at least for US-based servers. I spend a lot of time thinking about how to stop users from interacting too freely, to censor and moderate them so I don't wind up on some news site in 20 years being accused of hosting a site *Widely Used* by pedophilic narcoterror jihadists; I would like to not, but user content (and especially their information) is a huge liability to host... unless you're Equifax or Facebook or Google or some other large corporation -- then you can accidentally dump out everyone's sensitive financial information and only pay them $9 in compensation (or whatever the amount was; I keep throwing the cards they send me in the trash).

(yes I'm salty about that still)

in defense of 2.5 (Pro, at least), it was able to generate for me a metric UNIX clock as a webpage which I was amused by. it uses kiloseconds/megaseconds/etc. there are 86.4ks/day. The "seconds" hand goes around 1000 seconds, which ticks over the "hour" hand. Instead of saying 4am, you'd say it's 14.

as a calendar or "date" system, we start at UNIX time's creation, so it's currently 1.76 gigaseconds AUNIX. You might use megaseconds as the "week" and gigaseconds more like an era, e.g. Queen Elizabeth III's reign, persisting through the entire fourth gigasecond and into the fifth. The clock also displays teraseconds, though this is just a little purple speck atm. of course, this can work off-Earth where you would simply use 88.775ks as the "day"; the "dates" a Martian and Earthling share with each other would be interchangeable.

I can't seem to get anyone interested in this very serious venture, though... I guess I'll have to wait until the 50th or so iteration of Figure, whenever it becomes useful, to be able to build a 20-foot-tall physical metric UNIX clock in my front yard.

agreed, though I think the issue more is that these systems, deployed at scale, may result in widespread/consistent unexpected behavior if deployed in higher-stakes environments.

an earlier commenter mentioned a self-driving car perhaps refusing to use a road with a slur on it (perhaps it is graffiti'd on the sign, perhaps it is a historical name which meant something different at the time). perhaps the models will refuse to talk about products with names it finds offensive if "over-aligned," problematic as AI is eating search traffic. perhaps a model will strongly prefer to say the US civil war was fought over states' rights so it doesn't have to provide the perspective of justifying slavery (or perhaps it will stick to talking about the heroic white race of abolitionists and not mention the enemy).

bias when talking to a wide variety of people is fine and good; you get a lot of inputs, you can sort through these and have thoughts which wouldn't have occurred to you otherwise. it's much less fine when you talk to only one model which has specific "pain topics", or one model is deciding everything; or even multiple model in case of a consensus/single way to train models for brand/whatever safety.