I actually did already know that factoid but was struggling (am still) to see how it relates to a wooden trough that merely holds cables.
Another interesting factoid about the catenary: Robert Hooke proved that it takes on the shape (though inverted) of the ideal arch, in terms of supporting loads above it. La Sagrada Familia in Barcelona is filled with them.
> but was struggling (am still) to see how it relates to a wooden trough that merely holds cables.
Overhead Catenary [1] is a standard term, for a system that has two wires overhead - one suspended from the posts (forming a series of catenary curve), the other suspended from that cable at regular intervals (and held level relative to the track). The wood in Boston's system seems to replace the catenary cable.
And efficiency of the line depends on the curvature so for a given target efficiency you can calculate how far apart the poles can be. For electrical lines I mean.
My own memory of the events (which might be very wrong) was that a new vice-president of IBM semiconductors decided to drop bulk CMOS and focus exclusively on SOI (Silicon On Insulator). That suddenly left Transmeta without chips to sell. They had to scramble to find a new supplier and design their next generation processor for it (since the Crusoe wasn't portable to any other fabs). They were able to launch their Efficeon on TSMC 130nm (with a later version on Fujitsu 90nm) but the gap in supply was far worse for a startup than it would have been for a big company.
Backwards. The incompetent Transmeta board picked a VP from NVIDIA to be the CEO and his first action was to kill the IBM contract and move to TSMC, and forced TSMC to use a new unqualified process. This left us without chips to sell for over a year and notebook venders were furious and never returned.
This is what killed Transmeta, not all the technical details.
Cisco and Cray used IBM fabs for multiple generations in the aughts but they weren't startups. Before the rise of TSMC it was a weird situation where fabless companies were kind of picking up extra capacity from IDMs.
Yes but there are habits that are especially important for preventing stroke, such as getting 7-9 hours of sleep, monitoring and controlling risk factors related to blood vessel health that affect the brain uniquely, such as preventing irregular heart rhythms (atrial fibrillation), anti-inflammatory diet choices focused on brain health, and so on.
> Although I’ve never had a boss that needed to find an error.
I think that is key. A great mentor early in my career pointed out to me: "A" rated people need to work for "A" rated bosses. It's possible to have a "B" or "C" person work for an "A" boss, but when you put "A" people under "B" or (god forbid) "C" bosses, all kinds of problems ensue.
[I've personally experienced that situation only once, and swore never again.]
I’m not sure what maps to “A,” “B,” “C” here. My gut says: “B” is the kind of person who you’d use this trick on, “C” might be too lazy and just not bother, and “A” might be confident and respected enough to say (and have everybody believe) that they checked and didn’t have any issues. Only “B” has that mix of insecurity and some ability…
Actually, I bet you could have an ok workplace with “A” workers under “C” management. Or maybe the “C” turns into an “A” if they manage to hire good people and get out of their way…
I guess it depends on what "A", "B" and "C" means exactly.
But the problem with "C" managers is that they won't judge "A" work as "A" work, won't understand why some of the "A" work is important, and will get in the way of the "A" engineers, making them go down "C" paths.
A "C" level manager brings the whole team down to "C" level and destroys the morale of "A" and "B" workers while they're at it.
An "A" level manager can guide everybody towards "A" level work.
Fedora makes major upgrades pretty easy - you can even do it via the GUI Software Center, then reboot.
Personally I'm using Kinoite[1], an "immutable" version of Fedora that has an immutable base image, which makes it nearly impossible to break things during updates (even major upgrades).
Like many of you, I serve as IT support for family. Some of those family are beginning to slip cognitively, so I'd like to say: Fk google for doing this. You are confusing my relatives who cannot tell the difference between your ad-spam and actual links, and it is not an exaggeration to say that you are now taking advantage of old people.
I'm trying to install adblockers (uBlock) and move them over to chatgpt when possible. If anyone has better ideas, I'm all ears.
OpenAI has even more VC money to pay back than Google ever did.
Practically all the large tech companies so far have turned to ads and monetizing users rather than charging enough to remain more neutral.
I suspect one day, when you ask ChatGPT "Can you give me a link to mid journey", you'll instead probabilistically get a link to whichever competitor paid OpenAI for the best placement.
> Is there controversy over the true inventor of the telephone?Yes, there is controversy over the true inventor of the telephone. While Alexander Graham Bell is widely credited, several inventors and researchers argue for recognition based on their contributions:
> Antonio Meucci: An Italian inventor who filed a patent caveat for a "voice communication apparatus" in 1871, five years before Bell's patent. Meucci's device, the "teletrofono," could transmit voice over a wire. Due to financial hardship, Meucci couldn't renew his caveat, and Bell was granted the patent in 1876. In 2002, the U.S. House of Representatives passed a resolution (H.Res. 269) recognizing Meucci's contributions, stating he demonstrated a working device earlier, though it didn't officially credit him as the inventor. Some still argue Meucci deserves primary credit.
> Elisha Gray: An American engineer who filed a patent caveat for a telephone-like device on the same day as Bell, February 14, 1876. Bell's patent was filed hours earlier, leading to disputes. Some claim Bell may have had access to Gray’s ideas through patent office connections, though no definitive evidence supports this. Gray later challenged Bell’s patent but lost in court.
> Philipp Reis: A German inventor who developed a device called the "Reis telephone" in 1861, capable of transmitting music and some speech. While it was less practical for clear voice communication, some argue it was a precursor to the telephone.
Antonio Meucci invented the Teletrofono around 1849 and filed a patent for it in 1871. I know this mostly because it was a big deal in a Soprano's episode.
Wait long enough and it seems like almost any company tries anything to increase its bottom line, but the main difference between ChatGPT and Google is at least ChatGPT attempts to give a paid option. Again, I don't think that'll stop them from ever getting to that point... but it'll go farther than "here's search, we pay for it via adtech".
Kagi is a similar boat - the product is what you pay for, not what they can get users to put up with.
> at least ChatGPT attempts to give a paid option. Again, I don't think that'll stop them...
Netflix also attempted to give a paid option, but now we have an "ad-supported" plan. I think that same logic of maximizing profit means that even if there are some people paying for ChatGPT, the amount of free money that is sitting on the table means that we will see "ad-supported" ChatGPT pretty soon once the low-hanging fruit for quality enhancement start to dry up, which is kind of already happening.
I think the coexistence of ad supported plans is orthogonal to the above. E.g. Netflix still has an ad free plan, regardless of the other plans, but Google gives you no option.
"And now let's introduce this video's Sponsor, SpywareVPN"
Yeah, sure, "ad-free plan". As long as you don't watch (what feels like) the majority of videos on the platform.
I pay for premium, but I'd gladly pay 4x as much if Youtube also required creators to mark sponsored segments and let them all get skipped automatically if you paid for youtube "double premium double ad free" or whatever.
The premium plan actually gives you a little button to fast forward sponsor segments. (Not sure, if that's also on the free plan?)
You are right, that you still need to hit that button. It would be need to trigger it automatically. As far as I can tell, creators already tag the relevant segments as sponsored for other legal reasons.
1. The reason why ChatGPT is free despite being honestly very advanced, is that they want the general public to have an association of ChatGPT being "the default AI", just like Google is the default search engine and YouTube is the default video platform. Once they have this position they can throw as much garbage at the users as they want and nobody will care. This is why it doesn't really matter how much it costs now to capture the market, if the potential benefits are huge.
2. Once the market is captured and solidified, ads and enshittification ensue. If you're willing to put on your tin foil hat for a second, I'd tell you that as a matter of fact the technologies to integrate different services with ChatGPT are being developed right now, and once they're ready it's just a small step to make sure that ChatGPT prioritizes answers mentioning those integrated partners, which can easily be justified to users as legit quality-of-life improvements.
Maybe the answer is indeed to just buy a book and go touch some literal grass, and let the civilization drown in the sewer of disinformation it produces.
How's that different to all the time and effort spent on making television shows so that they can direct your attention to the next beer commercial, which also took lots of time and effort to make?
I'd say it's practically guaranteed. It would be wildly unprecedented to not follow up the amount of hype and fundraising in the LLM AI industry without a massive amount of enshittification following it.
Even if improvements continue for years we might already be near the peak of LLM usefulness because all of greedy and abusive dark patterns are sure to follow once the manic land grab settles down.
This is one of the reasons why I’m getting familiar with self-hosting. Local models are improving shockingly fast. I use Gemma3 27B for generating summaries of podcast transcripts, for instance.
In hindsight, we should have known this would happen eventually. At this point, we have to be actively be against free services. Every time its just a ticking time bomb. There's literally no incentive for them to be an actual good service, just good enough that you tolerate it and not consider other options, but shit enough that they can extract value out of you.
I recently signed up for an annual subscription to Kagi on their Starter plan and I couldn't agree more. Search quality with them has been great so far, and I realize their small web search and exploration features too.
I've been slowly working to find other paid services as alternatives to the free ones that I'm currently using (next big one was shifting away from Gmail and onto a personal domain for mail using Fastmail). Migrated away from Notion and using Obsidian with Syncthing running on my unRAID server at home. Generally just trying to find alternatives that aren't in the data mining and user lock in sphere and more about maintaining a positive user experience without taking advantage of their users and their data.
This. Please can we go back to the days where I simply pay for services or items instead of being trapped inside a maze of buy now pay later, credits, coupons, bonuses, gifts, tiers, etc
I am sure there have never been such a time, but I long for it anyway.
> At this point, we have to be actively be against free services.
Nah, GCC is free, Linux is free, Debian is free. What we need to be against is free stuff provided by for-profit entities, because the love of money is the root of all evil.
Linux is free as-in freedom. Linux is not zero-cost: it has taken tens of billions of dollars of investment from thousands of organisations over three decades - and countless volunteer hours - to make it what it is today; that the wider community gets Linux security patches and feature updates for free is a side-effect of the GPL license coupled with the low marginal cost of reproducing software once-written. I’m here to remind people that the bulk of Linux’ codebase was not written for free as an act of charity.
What I’m saying is that, hypothetically, if the entire business-world suddenly ditched Linux overnight and went back to IBM and Burrows like it’s the 1960s again again (and let’s pretend Android isn’t a thing either) then no-one would be funding significant Linux dev/eng work, and as much as we value the hacker-spirit of unpaid community/volunteer projects, I feel it isn’t enough to keep Linux viable and secure (especially in high-visibility, high-exposure scenarios like desktops and internet-facing services).
Eh, I heard lots of complaints from the Xen folks that the prevalence of RedHat in the kernel development community leads to double standards that makes their favourite product (KVM) get nicer and quicker reviews than the Xen related changes.
Not Google related, but cognition and older relative relevant: The amount of predatory scamware targeted towards older adults on the app stores is infuriating. I have a family friend who is now in the early-mid stages of Alzheimer's, but is still able to live at home and enjoy her life. She gets confused and stressed out by the fake 'alert! all your photos will be deleted!!' ads that pop up when she does her adult coloring books or jigsaw puzzles on her ipad. Apple's recommended apps in this category are evil in this regard, every single one. I've had to disable $80/week 'security' subscriptions from her account more than once. It is shameful that this is allowed.
to be fair, I had a Kagi subscription for a year, but I recently cancelled it (mainly because I'm not working now and need to cut monthly expenses). I'll probably rejoin eventually, but I can understand why people cut these things out or won't do it. Recurring monthly expenses can be hard.
Creat your own family yahoo — a website you maintain that has links to the websites they commonly use like mail and bank. Set as home page and new tab page.
It’s a slight security risk since it shows where you have accounts.
If you are savvy, build your own search that just passes it to an LLM and returns as page.
This is a real risk. I know of someone who got phished with a fake number for Apple Support (the fake number was promoted and appeared at the top of the search results). Apparently they do this with banking phone numbers as well.
Network wide ad blockers like PiHole are also quite useful but they can cause some confusion from the client side because things just break for no apparent reason.
I pay for Kagi for search, my family uses Kagi.
I pay for NextDNS to block ads, all of my family's devices use NextDNS.
I pay for credits on OpenRouter and host an OpenWebUI instance, all of my family's AI is private.
I pay for the news - The Economist, the WSJ, FT, NewScientist, etc. Lies are free, the truth is behind a paywall.
The only thing money can't buy, yet, is a phone network free of robocalls.
X-ray fluorescence detects elements based on their characteristic electromagnetic spectrum when irradiated with x-rays.
Not very much like a mass-spectrometer which creates a characteristic pattern of masses resulting from the test material as it is manipulated by the electron ionization or chemical ionization process. Where ions are detected across the atomic mass range of the particular spectrometer, forming a characteristic pattern or "spectrum" across that range.
Actually more jewelers and gold dealers than ever are using the x-ray guns professionally for bulk assay on an everyday basis. There are some handhelds which may be sensitive enough for trace analysis in food, but that requires a whole nother level of dedication beyond identification of metal objects, not just in technique and training but "laboratory" preparation as well.
The first obstacle would be convincing an owner of an instrument having capable specs, to embrace usage for things other than gold and silver assay. Then seriously pursue mastery of the instrument more so than ever to accomplish decent detection of low levels of lead and other metals like chromium, mercury, cadmium, etc.
[1] https://pubmed.ncbi.nlm.nih.gov/25214856/ ("Contaminated turmeric is a potential source of lead exposure for children in rural Bangladesh" / "Results: Lead concentrations in many turmeric samples were elevated, with lead concentrations as high as 483 ppm")
The higher it is, the less likely for challenges in detection, and/or interference from background.
>lead concentrations as high as 483 ppm
SSDD.
Shouldn't be that hard to detect at that level which is way above ppb. There are a number of reliable methods.
However if the Minimum Detectable Level for a particular test procedure was only 500 ppm or above, one of these samples would report just as clean as a sample having no lead whatsoever; < 500.
MDL's like this which vary among different test methods do need to be carefully compared to the toxicity levels being screened for.
That's another one of the confounding aspects to be aware of.
Depending on circumstances, I may or may not prefer a different calibration session for each of these two levels, even though they are both within the same order of magnitude.
Either way ideally I would be preparing NIST-traceable reference materials at the proper levels for comparison & confirmation. Not much differently than I would do for the benchtop models and the forklift models of x-ray units. And to really get down into the ppb levels that's when the ICP/mass-spec comes in handy, that's a benchtop unit itself, too big to fit on a regular desk though. However you don't really get the most out of the ICP without a huge cryogenic tank of liquid argon out back so you can "consume mass quantities" ;)
With a handheld x-ray unit, if you are only assaying gold & silver it may be fine to send it back for calibration once a year, if the pawn shops even do that. For food testing I would want more of a laboratory-style analytical procedure and calibration which is concurrent with materials being tested.
I agree that hand-held XRF guns should be able to detect such lead levels, and I believe that was in fact what the police used when they did the publicity stunt in the Bangladesh market. At any rate, it sounds like you know a lot more about the question than I do. I was only disputing pfdietz's comment, "The concentrations of lead being discussed here are as much as 1000 ppm or even higher."
Good call because results should be expectd to be all over the ball park, and I think even higher numbers could be found. But no amount of lead is supposed to be acceptable.
>sounds like you know a lot more about the question
SSDD says it all without explanation, but here's a little.
Until you've spent lots of time at the bench, it's not easy to understand why a 1000 and a 483 might just be the same sample tested in different labs.
Or even the same lab on different days.
If so that would look even more embarrassing when my arbitrary reporting convention < 500 is applied.
But it's actually not unheard of to get a positive and a negative on the same sample even with some of the most sophisticated equipment
Explaining the rest of the story could fill textbooks, but the operators wouldn't be reading them anyway :\
So that's the most important thing to know, besides the actual spectrums which are table stakes.
Back in graduate school, I TA'd an electrostatics course. We were going through the details of the basic parallel-plate capacitor, and so Prof. Peter Hagelstein (of the project you listed above) used the example of how much energy was stored in a football-field sized set of parallel-plate capacitors with oil as a high-breakdown dielectric.
The students were dutifully copying the lecture while I was sitting there with my mouth agape realizing that he was working through a simplified example of what energy storage was required for the X-ray laser. IIRC Those guys had their own substation, and would charge the capacitors. The switch would get thrown and the sublasers would shoot at the molybdenum target, which would laze in the X-ray spectrum (and the molybdenum would vaporize, I think.)
Afterwards, I asked him how on earth the energy was transferred from the caps to the sublasers: He just smiled and said "very carefully".
He spoke at MIT (early 90s?) and I remember him talking about making fun of PacBell colleagues in his comic: They would recognize themselves, ask him to autograph the comic for them, and then go away happy (thus making fun of them a second time.)
Maybe you all knew that factoid already, but I learned the name of shape only recently.
reply