!!c1QfXUgcGY0

[highways-are-liminal-spaces]

[femmenietzsche]

On March 5, 1979, a few months after the successful dropping of landers into the atmosphere of Venus, the two uncrewed Soviet spaceprobes Venera 11 and 12, then in heliocentric orbit, were hit by a blast of gamma radiation at approximately 10:51 EST. This contact raised the radiation readings on both the probes from a normal 100 counts per second to over 200,000 counts a second in only a fraction of a millisecond.[5] Eleven seconds later, Helios 2, a NASA probe, itself in orbit around the Sun, was saturated by the blast of radiation. It soon hit Venus, where the Pioneer Venus Orbiter's detectors were overcome by the wave. Shortly thereafter the gamma rays inundated the detectors of three U.S. Department of Defense Vela satellites, the Soviet Prognoz 7 satellite, and the Einstein Observatory, all orbiting Earth. Before exiting the Solar System the radiation was detected by the International Sun–Earth Explorer in halo orbit.[5] At the time, this was the strongest wave of extra-solar gamma rays ever detected at over 100 times as intense as any previously known burst. Given the speed of light and its detection by several widely dispersed spacecraft, the source of the gamma radiation could be triangulated to within an accuracy of approximately 2 arcseconds.[22] The direction of the source corresponded with SGR 0525−66, the remnant of a star that had exploded as a supernova around 3000 BCE.[11] It was in the Large Magellanic Cloud and the event was named GRB 790305b, the first-observed SGR megaflare.

First discovery of a magnetar (neutron star with an especially powerful magnetic field), reads like the start of an Arthur C. Clarke novel

[c1qfxugcgy0]

I just watched Avatar: Fire and Ash, which means I had to complain about it in fake script form.

[furryhatersigma]

[mariacallous]

Inside the Messy, Accidental Kryptos Reveal

After 35 years, the secretive CIA sculpture finally gave up its mystery, thanks to a novelist, a playwright, and some misplaced documents. B

Jim Sanborn couldn’t believe it. He was weeks away from auctioning off the answer to Kryptos, the sculpture he created for the CIA that had defied solution for 35 years. As always, wannabe solvers kept on paying him a $50 fee to offer their guesses to the remaining unsolved portion of the 1,800-character encrypted message, known as K4—wrong without exception. Then, on September 3, he opened an email from the latest applicant, Jarett Kobek, which started, “I believe the text of K4 is as follows …” He’d seen words like this thousands of times before. But this time, the text was correct.

“I was in shock,” Sanborn tells me. “Real serious shock.” The timing was awful. Sanborn, who turns 80 this year, saw the auction as a way for someone to continue his work of vetting potential solutions while maintaining the mystery of Kryptos. He’d also been looking forward to getting compensated for his work. What came next was even more shattering. He quickly got on the phone with Kobek and his friend Richard Byrne, who gobsmacked him by reporting they did not find the solution by codebreaking. Instead, Kobek had learned from the auction notice that some Kryptos materials were held at the Smithsonian’s Archives of American Art in Washington, DC. Kobek, a California novelist (one of his books is called I Hate the Internet), got his friend, the playwright and journalist Byrne, to photograph some of the holdings. To Kobek’s astonishment, two of the images contained a 97-character passage with words that Sanborn had previously dropped as clues. He was staring at the full unencrypted text that CIA and NSA codebreakers, along with countless academics and hobbyists, had sought for decades.

The secret of Kryptos was out of the artist’s hands, in the most humiliating way imaginable—Sanborn himself had mistakenly submitted it in readable form to the museum. For 35 years the Kryptos plaintext had been a summit that none had reached. Suddenly some had attained it—not by climbing to the peak but by hitching a ride to the top. Sanborn’s grand vision for a piece of art that illuminated the idea of secrecy itself was imperiled—as was the auction. Now he had to figure out what to do about it.

Enter: The Media

The initial phone call had been friendly. Kobek and Byrne insisted that they did not want to mess up the auction. After he hung up, Sanborn called the auction house. That’s when things started going sideways. As Sanborn tells me, “They said, ‘Listen, see if the guys will sign NDAs, and see if they'll take a portion of the proceeds.’ And I said, ‘Oh geez, man, I don't know about that. But I offered it.'”

Kobek and Byrne were uncomfortable with that arrangement and refused to sign. (RR Auction executive vice president Bobby Livingston didn’t comment on the legal issue but says of an NDA, “It’s something that would be comforting to our clients.”) Sanborn told them his intent was to get the Smithsonian to freeze the archives—which it did. He assumed Kobek and Byrne would stay silent. “If you don't release it, you're heroes to me,” Sanborn told them.

“I thought everything was OK,” he says, “And then all of a sudden [the journalist] John Schwartz calls me and says these guys want to publish it in The New York Times.” Kobek explains to me that they contacted Schwartz in part to relieve some legal pressure. “There was threat after threat being sent to us from the auction house's lawyers, threatening to sue us for a multitude of things,” he says. (When I ask Livingston if his lawyers have been contacting Kobek, he says, “There’s lawyers talking to each other,” and adds that there may well be copyright concerns if Kobek and Byrne published the plaintext.) On October 16, Schwartz published his scoop, informing the world that the plaintext was out.

Sanborn tells me that Kobek shared the plaintext with Schwartz over the phone. When asked about this, Kobek says, “I cannot speak about that…I am under significant legal peril.” Schwartz says. “Once my editors decided it would not be revealed in the story, I deleted the text from my interviews file. I don’t know it.” (So don’t bug him.)

RR Auction now has a disclaimer on the description of the Kryptos lot, noting the Smithsonian discovery. “The researchers have stated they do not plan to release the solution,” it reads, though RR also acknowledges that there is no assurance that the solution won’t spill at some point. “It certainly has an effect on the auction,” says Livingston. “But we're moving forward with it, with the disclosure of what happened.”

Livingston points out that the firm is auctioning off not just the K4 plaintext, but a set of objects that have worth on their own, including the coding charts, the original handwritten plaintext, a unique “proof of concept” copper plate, and Sanborn’s CIA visitor badge. For collectors of cryptographic history, or even art fanatics, those goodies might be irresistible. Sanborn has also pledged to spend an afternoon walking the winner through the Kryptos story, with guidance on how to keep it going.

With the news that at least two people have the K4 plaintext, RR and Sanborn have shifted their focus to a different aspect of the Kryptos secrecy. In recent months, Sanborn has been talking more about the fact that the ultimate solution to Kryptos is not the decrypted plaintext but another mystery that will emerge when the plaintext is available. Kind of like finishing off the big boss in a video game and then finding out there’s another villain to conquer. “I said years ago, there's something after Kryptos,” he now says. “ It's a riddle. It means something. It leads somewhere else. And then all of a sudden it started getting named K5.” Sanborn is vague on the nature of K5. Is it one of the test passages Sanborn made in 1988 when he was trying to see what would fit on the face of the sculpture? Something that involves other parts of the CIA installation? Or a different Sanborn puzzle altogether? In any case, Sanborn and his auction house are now touting the K5 solution as part of the prize.

That’s not to belittle the fact that, despite Kobek’s discovery, K4 is in no way solved. The Kryptos mystique has endured largely because for 35 years the ciphertext has confounded the world’s best codebreakers. Elonka Dunin, a leader of the community devoted to endless speculation about the Kryptos solution, says she hasn’t detected a slackening of interest after hearing about the Smithsonian discovery. She says that Kryptos-world is more upset about the fact that the solution will be auctioned off—adding uncertainty to the 35-year-old chase—than at the fact that two random researchers found a copy of the plaintext. (Pretty much everyone thinks Sanborn deserves a better retirement nest egg.) To her, the real quest is to actually crack the code, a sentiment that Sanborn avidly shares. “They only discovered it, they did not decipher it,” he tells me.

Right now, the parties have reached an uneasy stasis. The auction is now open; as I write this, the high bid has topped $43,000. The auction house hasn’t changed its estimated $300,000 closing bid. Kobek says he wishes he had never sent Byrne to the Smithsonian in the first place. “It’s a sad story, in a way,” he tells me. Sanborn is in agony.

Still, I don’t think the Kryptos saga has been irrevocably tarnished. I always saw Sanborn’s artwork as a brilliant commentary on the shadow world of spookdom, where nothing is as it seems and outcomes often come without clarity. To date, the machinations of solving Kryptos—or even figuring out if it could be solved—have been on a high level, a complex web of shadows that John le Carré might have spun. The anticlimactic discovery of the plaintext, due to Sanborn’s slip, has shifted the narrative to the tragicomic purgatory of Mick Herron’s Slough House series.

There’s still a chance for a satisfying ending. Sanborn’s hope in auctioning off the solution has always been that the winning bidder will continue the process of vetting guesses until someone finally arrives at the answer via math and grit. To help his successor, Sanborn’s been working with technologist and entrepreneur Igor Jablokov to build an AI system that can autonomously sift through the guesses. Jablokov described it to me as a virtual Sanborn.

Jablokov is a rabid Kryptos fan; he even owns a unique replica of the sculpture that was created for an episode of the TV show Alias. He’s also wealthy, having sold his first company to Amazon. He tells me that earlier this year when he heard of Sanborn’s financial situation, he offered to buy the solution and take over the vetting. Sanborn could not explore that option because he’d already agreed to do the auction.

Hearing this, it dawns on me that I might be talking to the eventual winning bidder for the Kryptos secret—and the person who will continue to protect its remaining secrets. I ask Jablokov whether he is going to do just that. The pause after my question indicates I might have hit paydirt. “I’ll leave that to ambiguity,” he says. “But obviously I'm an actor, an interested party.” We’ll see what happens on November 20 when the auction closes. I suspect that no matter the outcome, ambiguity and mystery will still pervade one of the greatest art projects of our time.

[official-kircheis]

I mean it’s very fitting that the plaintext was discovered because of a human slip-up, sidechannel attack, protocol violation. that’s how it usually goes! 5 dollar wrench attack and all that

[unichrome]

I’m slowly beginning to accept the reality that 2007 was not last year but in fact almost four years ago

[c1qfxugcgy0]

What’s your favorite color? You seem like the type of guy who would either have a very specific color you LOVE or be have very vague and broad favorite colors…

Normally I would say the infrared color of 9250 nanometers, the backbody peak of a human body, the invisible light that reveals the presence of a person. Pretty eldritch and definitely esoteric.

But I've got to say it's ultraviolet, instead. A terrible, silent radiation shed by the celestial body we orbit that kills bacteria and roasts our skin? A blue bluer than blue, light so hot that it ignores lenses and can only be focused by mirrors, and destroys anything it touches? Now that's a color!

[alienestrangement]

[c1qfxugcgy0]

Tools of Large Aircraft Manufacturer

(previous post in this series)

One peculiarity of LAM is that the vast majority of the hand tools are pneumatic, air powered. The reason for this is obvious: the company is more than a century old! The factory I work in was constructed in the 1960s! Electric tools existed at the time, but they sucked. And, of course, battery powered tools didn't exist at all. In contrast, air tools were light, handy, easily throttled (small cheap AC motor controllers wouldn't exist for another 40 years) and easily repaired. So, naturally, you run air lines to every work area in the factory, and once they're installed, you keep using them.

First of all is the humble rivet gun, the right hand of the aircraft mechanic. There's no practical way to electrify this tool. LAM mechanics could be building vertibirds in 2070 with microfusion cell powered drills, but the rivet guns will still run on compressed air.

(Something I didn't know before joining LAM is that the gun is applied to the head of a solid rivet, with a guy on the other side holding a bucking bar against the tail of the rivet. The gun rapidly bounces the rivet inside its drilled hole against the bucking bar, mushrooming the tail of the soft aluminum rivet against the steel anvil.

Again, I want to emphasize, surprisingly large amounts of this is done by hand. Maybe 80% of the little bumps you see along each rib and stringer are solid rivets manually plinked in and hammered down by meat humans. There's a reason these planes cost $100 million each.)

And of course there's bizarre, alien looking pneumatic drills and screwguns. Without the need for batteries and electric motors, pneumatic tools shrink down to just the basics, the smallest possible chucks and air motors attached to a vestigial grip. But that's not weird. You know what's weird?

Did you know you can make a vacuum pump using a stream of compressed air? It's the easiest thing in the world, just two cones, no moving parts. LAM uses it anywhere you need a spot of vacuum, like if you need to apply some perfectly uniform clamping force across a composite repair:

And it's used for even more prosaic things. Need to clean up some drill shavings?

The perfect tool for sucking up difficult things. An electric motor would probably have a bad day sucking up mixed carbon dust and steel shavings. But there's nothing to be damaged in a pneumatic vacuum! It'll eat anything you feed it, no problems.

But that's not the weirdest tool. Check this out.

A vortex tube is a chunk of metal that separates a stream of compressed air into hot air and cold air, using demon magic. In other industries the cold stream is typically used, but vortex tube hot air guns are used surprisingly often on the factory floor, anywhere you need to cure a paint touchup quickly. (Ideally, within a single eight hour shift, so the same mechanic can record the start and end time)

Using a vortex tube heater has advantages. Again, no moving parts. It's impossible to start a fire by using it carelessly, since the hot air stream won't ever be higher than 200°F. It outputs an absolute torrent of fresh, hot air, which is great when the goal is to drive out the solvent from a patch of fresh paint.

But it is not efficient. The Vortec 609 claims it can do 900 BTU/hr (263 watts) of heating while consuming 15 cubic feet of air at 100 psi. The IV5076055 two stage air compressor needs a 5 horsepower motor to output that much air, or 3,677 watts. 263/3677 = a fat 7.1% efficiency. A regular electric space heater, need I remind you, is 100% efficient. (And, using Alec Watson's favorite device, the vapor phase change refrigerator, you can do much better) For a company that endlessly nagged us to deposit recyclables in the correct bins and put compostable plastic utensils in the cafeteria that crumbled in your hand, it seemed an odd oversight.

[starrythroat]

i decided to google my top 1 favourite mushroom that i used to hunt as a child because it tastes heavenly but seems to be somewhat rare target and i wanted to know more about it

[violetfractal]

A circle inscribed in a square covers 79% of the square.

A ball inscribed in a cube fills 52% of the cube.

A 4-ball inscribed in a hypercube fills 31% of the hypercube.

A 5-ball inscribed in a 5-cube fills 16% of the 5-cube.

⋮

A 9-ball inscribed in a 9-cube fills it up less than 1% of the way, yet there’s no room to fit a second ball of the same size without intersecting the first.

⋮

In dimensions 23 and up, you can fit a little cube in the corner of the diagram, such that the cube has a larger volume than the ball!

That's especially disconcerting because the 23-cube has 8,388,608 corners. Even if you inscribe the little red cubes in all eight million corners, your big cube will still be 99.5% empty space.

[c1qfxugcgy0]

So now that your five year long prophecy has come true, do you think tumblr will burn out slowly after this or will it just implode?

Well, it’s certainly taking its sweet time about it.

(New followers might not be aware that I predicted Yahoo would ban NSFW content shortly after the announcement of the acquisition. Which was, my word, twelve years ago now. As far as I know this is the only post I’ve done that’s broken 10k notes.)

(My second most popular post (in my heart but not notes-wise) is my James Webb Space Telescope Quirks And Features post)

[aorish-deactivated20251222]

in plato's republic, socrates famously "goes down to piraeus" and then gets accosted on the way back for narrative reasons, which doesnt make sense unless you know about ancient athens' wildest infrastructure project, the "Long Walls" that ran from athens to piraeus

no big deal just building a giant 5km long walled road to the port so my city cant get sieged again

[c1qfxugcgy0]

I will never complain about a toolbox being too far from the jobsite again.

(previous post in this series)

Large Aircraft Manufacturer (LAM) has announced, to the surprise of nobody with a brain, that certification of our latest aircraft, Advanced Widebody Carbon Wing (AWCW), has been delayed to March 2026.

This firmly sets management on the horns of a dilemma. They have something like five thousand expensively trained employees on the AWCW production line who will not have much to do for the next year. You can continue production and clutter up the hardstand with precertification aircraft. But the process of certifying the aircraft against severe weather, bird strikes, lightning strikes, etc etc, will inevitably require serious changes to the beta aircraft. LAM must then modify every one of their backlog aircraft, ripping out the interior, replacing bond wires and ground straps, then reinstalling all those parts. Doing structural work inside a complete aircraft naturally takes much more time than doing it from scratch in the production jigs designed to accommodate such work.

(And if you don't believe me, just watch This Old House.)

Naturally, LAM tracks every minute of worker time on each aircraft. Enough rework can wipe out LAM's entire profit margin on a bird, especially given the large discounts it offers to early buyers of new model aircraft.

This is not idle supposition. LAM was hauled through an identical hall of thorns when Advanced Midbody Carbon Aircraft (AMCA) was delayed in certification a decade ago. Fifty aircraft required expensive rework, putting the entire program in the red for years afterward. The scars are fresh, and LAM is not eager to repeat the experience. Thus, AWCW production rate has been cut to zero point zero.

But what to do with the workers?

Airplane factories are always attached to an airport. [citation needed]

Everything outside the factory is the flightline. Flightline is where all the problems with an aircraft catch up with it, and occasion screaming matches between facility managers (who are desperate to clear their patch of concrete and get the plane in the sky) and production managers (who will have the rare pleasure of seeing their face on the nightly news when that plane kills three hundred people).

Airplanes require a really incredible amount of maintenance. If production delays mean the plane doesn't get delivered to the customer on time, scheduled maintenance can happen while the airplane is being made. These are not problems that happen when you build cars, I can tell you. This is the shop I, along with 20 of my coworkers, have been loaned out to.

There are lots and lots and lots and lots and lots of moving parts on an airliner. Every LAM aircraft has a design life of 30 years. They cost hundreds of millions of dollars each. Because they are so damn expensive, our customers want to fly them as close to 24 hours a day as possible, in rain, snow, sleet, from Kabul to Kathmandu, from sea level to eight miles above ground. Sealed bearings, so beloved by the automotive industry, are simply not an option across aerospace's range of temperature, pressure, salt spray, and total joint lifespan requirements.

As a result, every single metal on metal joint on the airplane has a grease fitting, and a prescribed grease type for each fitting. In just the photo above there are seven fittings visible. The document that lists every fitting on the plane is eight hundred pages long.

But greasing the points is, honestly, not that hard. You've got eight hours to finish any given IP, and in a storage IP the greasing will take, at most, 30 minutes. Greasing is not the problem. The problem is the fucking skin panels.

The exterior surface of a wing is, uh, important. It carries the weight of the aircraft, it has to be aerodynamically smooth to a frankly annoying degree, each carbon fiber wing skin panel has to be as light as absolutely possible, the insulative carbon fiber composite must be coated with an outer antistatic conductive layer to bleed off static charge, but at the same time the inner layer needs a more conductive aluminum foil layer to conduct the powerfully destructive lightning strike energies each plane will experience, oh, about thirty times over its rated lifespan.

On that list of priorities, "making it easy for ground personnel to take a panel off" is low on the list of the priorities. Very low. Real damn low. Put on your SCUBA gear and investigate the pelagic depths kinda low.

You take off the panels. Maybe ten percent of the screws will strip when you apply force, which means you get to carefully, slowly drill out the titanium fasteners while standing at the top of a scissor lift in the rain.

(There is an art to drilling out a Phillips head titanium screw. Ordinarily, you want to use carbide tooling, which is sharp, but brittle. But even after stripping the hell out of a screw there will still be some remnants of the head, which the cutting edge of the carbide drill will catch on and break. So when your crew is assigned to a new plane, the first thing you do right away is rush to the tool room to get drill bits before your oafish coworkers clean them out, and get both HSS and carbide bits-- tough and ductile steel to knock down the remnants of the screw head and then carbide to do the bulk of the drilling. And once you're into the bulk of the screw, you do peck drilling-- three or four seconds of drilling, then pull the bit out and apply lube. This isn't for the benefit of the drill-- it can handle high temps just fine. What you absolutely, must not do, is let the screw get too hot. Because when titanium gets hot and then cools down, it hardens, and you just turned a ten minute job into a four hour one. Because after you finish drilling the hole you follow it with a steel screw extractor, and there's no extractor on Earth that's going to bite into hardened titanium.)

You apply Aeroshell 33 to the bushings on the slats torque tube and carefully brush on Cor-Ban 27L to the specified exposed metal surfaces. You call QA out, who bitches and moans the entire time for being rousted out of their crew shelter to get rained on to witness that you greased the things that needed to be greased.

Now it's time to put the panel back on. First, you throw away all the used fasteners and order new ones from Logistics. Any screw that touches a flight component is used once, and only once. Try not to think about the dollar value of the two pounds or so of aerospace titanium screws you just shitcanned. Be careful when reordering, though-- across the five or six panels you're pulling off you'll have two different types of screws of differing surface finishes, (structural screws vs. antistatic electrical bonding screws) different diameters and different screw lengths. Why? Because fuck you, you stupid mechanic. You deserve to suffer. Your life should be only pain.

(If you screw up on this step and can't button up a panel before end of shift you need to "short stamp" the IP saying what you did and did not do, check the panel into the WIP cage (remember to label it with the part number, IP number, and your employee number!) and then "maintain closure" by covering the empty spot with a sheet of plastic taped down along its entire perimeter with 3M 8979 duct tape. It is, of course, still raining while you're doing all this, because some fucking idiot decided to build an aircraft factory in the Pacific Northwet. Does duct tape stick particularly well to sodden wing panels? No, it does not.)

The one advantage of going to work at 5 am is that you never miss a sunrise

Assume you have all the screws you need and you haven't dropped any of the panels and damaged them while bumbling around. Apply Braycote 248 to the threads and start banging them home with a torque-limited screwgun.

Once installed, there are those two important electrical bonds mentioned above. LAM does not take your word that you've correctly installed the panel, of course, they want you to measure it. Getting the antistatic value is easy enough-- one probe on the head of the fastener, the other to the surface of the panel, value in the hundreds of kiloohms. Impossible to screw up.

What's harder is the lightning conduction path bond. That's measured in single digit milliohms, and it's from the foil lining of the panel to the structure of the wing. The foil is hard to access, since it's on the other side of the goddamn panel you just expensively installed.

Well, in some cases, you can just reach from an adjacent open panel. (The IP notes which panel does not require a lightning bond reading, and you are supposed to infer that this is the last panel to install.) But LAM defines "adjacent" somewhat loosely. By the time you are on the final panel, you are measuring bonds by duct taping one probe of the M1 meter to the end of a broomstick, crawling up the asshole of the plane, and jamming it against the back of a panel six feet away. This is as stupid as it sounds, and it takes several tries and quite a lot of fumbling around to get a good reading. If you don't get a good reading, then you will have the experience of taking the panel off, cleaning it real good, and then trying again, while your team lead breathes down your neck.

But if the readings are good, you unthread yourself from the guts of the wing, pound in the last panel, plug in your scissor lift, dump your cleaning materials contaminated with various exotic aerospace greases and weirdo solvents into the hazmat bin, return your tools to the tool room, and clock the fuck out. You've got a different airplane to grease tomorrow!

[c1qfxugcgy0]

Electric car chargers at the grocery store were busy today

[c1qfxugcgy0]

Let's check in on this grocery store, eleven years later.

Sadly enough, both chargers are completely defunct. Victims of a format war, both SAE J1772 and CHAdeMO are used by essentially no new cars. A hundred feet away, we see why.

A row of heavily used NACS chargers.

In the early years of the electric car buildout, it was thought that high power DC chargers would be expensive, rare and unusual, located far from population centers, intended for users on road trips. In cities, chargers would be cheap and simple level 2 AC chargers, fed from the distribution panel of their hosting building.

This plan, uh, sucks. It totally blows, in a few somewhat counterintuitive ways.

First: Level 2 chargers are intended to be, above all, compatible with existing infrastructure and to be cheap. 32 amps at 208 volts is the most you can easily draw from a branch circuit in a North American commercial building, like a grocery store. That's 6.6 kilowatts (just multiply them together) which is both a lot and not nearly enough.

Ten cars charging at the same time will double the power consumption of an average retail establishment, which means they just can't install all that many car chargers without requiring eye-wateringly expensive infrastructure upgrades. In the eleven years since Fred installed these first few level 2 chargers they added... no more. This is almost certainly because it would require upgrading their transformer, digging up the existing service drop, and installing new electrical panels. Tens of thousands of dollars of upgrades just to add a few more chargers? Not going to happen! And it didn't happen.

And the flip side: 6.6 kilowatts is a lot for a business, but not a lot for an electric vehicle. The first Tesla Model S, released fifteen years ago, had an 85 kilowatthour battery. Which means (Just divide one number by the other) it would take a level 2 charger twelve and a half hours to charge it from empty.

That's a lot of damn time for a car to be parked at a retail establishment! In practice, it was never worth it for customers to plug into a level 2 charger unless they were planning on being there for all day.

Secondly, since these chargers were fed from the main panel inside the building, the business owners were heavily incentivized (heavy gauge copper is expensive!) to put the chargers as close as possible the building. Like say, right next to the building. You know who parks right next to the entrance? Fucking everybody! Level 2 chargers were heavily prone to being ICEd out, making them unavailable for their intended users

[c1qfxugcgy0]

At my last job, we sold lots of hobbyist electronics stuff, including microcontrollers.

This turned out to be a little more complicated than selling, like, light bulbs. Oh how I yearned for the simplicity of a product you could plug in and have work.

Background: A microcontroller is the smallest useful computer. An ATtiny10 has a kilobyte of program memory. If you buy a thousand at a time, they cost 44 cents each.

As you'd imagine, the smallest computer has not great specs. The RAM is 32 bytes. Not gigabytes, not megabytes, not kilobytes. Individual bytes. Microcontrollers have the absolute minimum amount of hardware needed to accomplish their task, and nothing more.

This includes programming the thing. Any given MCU is programmed once, at the start of its life, and then spends the next 30 years blinking an LED on a refrigerator. Since they aren’t meant to be reflashed in the field, and modern PCs no longer expose the fast, bit-bangable ports hobbyists once used, MCUs usually need a third-party programming tool.

But you could just use that tool to install a bootloader, which then listens for a magic number on the serial bus. Then you can reprogram the chip as many times as you want without the expensive programming hardware.

There is an immediate bifurcation here. Only hobbyists will use the bootloader version. With 1024 bytes of program memory, there is, even more than usual, nothing to spare.

Consumer electronics development is a funny gig. It, more than many other businesses, requires you to be good at everything. A startup making the next Furby requires a rare omniexpertise. Your company has to write software, design hardware, create a production plan, craft a marketing scheme, and still do the boring logistics tasks of putting products in boxes and mailing them out. If you want to turn a profit, you do this the absolute minimum number of people. Ideally, one.

Proving out a brand new product requires cutting corners. You make the prototype using off the shelf hobbyist electronics. You make the next ten units with the same stuff, because there's no point in rewriting the entire codebase just for low rate initial production. You use the legacy code for the next thousand units because you're desperately busy putting out a hundred fires and hiring dozens of people to handle the tsunami of new customers. For the next ten thousand customers...

Rather by accident, my former employer found itself fulfilling the needs of the missing middle. We were an official distributor of PICAXE chips for North America. Our target market was schools, but as a sideline, we sold individual PICAXE chips, which were literally PIC chips flashed with a bootloader and a BASIC interpreter at a 200% markup. As a gag, we offered volume discounts on the chips up to a thousand units. Shortly after, we found ourselves filling multi-thousand unit orders.

We had blundered into a market niche too stupid for anyone else to fill. Our customers were tiny companies who sold prototypes hacked together from dev boards. And every time I cashed a ten thousand dollar check from these guys, I was consumed with guilt. We were selling to willing buyers at the current fair market price, but they shouldn't have been buying these products at all! Since they were using bootloaders, they had to hand program each chip individually, all while PIC would sell you programmed chips at the volume we were selling them for just ten cents extra per unit! We shouldn't have been involved at all!

But they were stuck. Translating a program from the soft and cuddly memory-managed education-oriented languages to the hardcore embedded byte counting low level languages was a rather esoteric skill. If everyone in-house is just barely keeping their heads above water responding to customer emails, and there's no budget to spend $50,000 on a consultant to rewrite your program, what do you do? Well, you keep buying hobbyist chips, that's what you do.

And I talked to these guys. All the time! They were real, functional, profitable businesses, who were giving thousands of dollars to us for no real reason. And the worst thing. The worst thing was... they didn't really care? Once every few months they would talk to their chip guy, who would make vague noises about "bootloaders" and "programming services", while they were busy solving actual problems. (How to more accurately detect deer using a trail camera with 44 cents of onboard compute) What I considered the scandal of the century was barely even perceived by my customers.

In the end my employer was killed by the pandemic, and my customers seamlessly switched to buying overpriced chips straight from the source. The end! No moral.

[heysawbones]

these are murdering me

[heysawbones]

just,,,