Well, one option, which can be pushing the boundaries of selfhosted for some, would be to use a hosted k8s service for your public-facing stuff and then a home real k8s cluster for the rest of it.

A few years ago now I was thinking that it was about time for me to upgrade my desktop (with a case that dates back to 2000 or so, I guess they call them “sleepers” these days?) because some of my usual computer things were taking too long.

And I realized that Intel was selling the 12th generation of the Core at that point, which means the next one was a 13th generation and I dono, I’m not superstitious but I figured if anything went wrong I’d feel pretty darn silly. So I pulled the trigger and got a 12th gen core processor and motherboard and a few other bits.

This is quite amusing in retrospect.

From the article: “Tesla began delivering the Blade Runner-inspired truck in November 2023”

Me: Fuck you. That is an insult to Syd Mead’s legacy.

It’s the Pravda of the VC-centric tech scene and has been for a very very long time.

(I am referencing the Soviet Union implementation thereof, for clarity)

It’s never going to bite the hand that feeds it, where people will voting-ring or the owners will just force-edit it to prevent that from happening. Outside of that, sometimes it might say something useful. The problem is that today’s problems are not because of a lack of advanced mathematics understanding or new programming languages.

I mean, I think he’s a textbook example of why not to do drugs and why we need to eat the rich, but I can understand the logic here.

When you navigate a car as a human, you are using vision, not LIDAR. Outside of a few edge cases, you aren’t even using parallax to judge distances. Ergo, a LIDAR is not going to see the text on a sign, the reflective stripes on a truck, etc. And it gets confused differently than the eye, absorbed by different wavelengths, etc. And you can jam LIDAR if you want. Thus, if we were content to wait until the self-driving-car is actually safe before throwing it out into the world, we’d probably want the standard to be that it navigates as well as a human in all situations using only visual sensors.

Except, there’s some huge problems that the human visual cortex makes look real easy. Because “all situations” means “understanding that there’s a kid playing in the street from visual cues so I’m going to assume they are going to do something dumb” or “some guy put a warning sign on the road and it’s got really bad handwriting”

Thus, the real problem is that he’s not using LIDAR as harm reduction for a patently unsafe product, where the various failure modes of the LIDAR-equipped self-driving cars show that those aren’t safe either.

It’s important to realize that the nerd you saw on the news has always been someone wearing nerd as a costume and the entire history of technology is loaded with examples of the real nerd being marginalized. It’s just that in ages past the VC’s would give a smaller amount of money and require the startup to go through concrete milestones to unlock all of it so there was more of a chance for the founder’s dreams to smack up against reality before they were $230m in the hole with no product worth selling.

While there is arguably a larger pool of people who you can reach by not having open racism and the CEO whipping his dick out (and mysteriously not slamming it into his Tesla door, even if it is a masterful gambit) you can still get a lot of white men of privilege who are smart and hardworking who don’t nominally worry about being on the receiving end of most of the harassment so it’s OK as long as they end up part of the winning team because they’ll get mega stock bucks at the end. And this does extend to the factory floor, at least people’s impressions while joining the factory floor. They wouldn’t be an engineer but they’ll be a supervisor or something?

It’s kinda un-earned? Like, there’s stories that people tell each other of questionable veracity? Some set of startups in the days of yore gave their cleaning staff or whatnot options so I think it’s become part of the cultural mythos now even if the reality is that the cleaning staff these days is contractors who are mistreated so even if it did actually happen then, it won’t happen now.

And, dono, once you’ve solved the hard problems early on, there’s less of that drive to do the truly novel things and so you get more of the people who want to be part of a company that’s going to the top and wouldn’t mind if they could coast and/or fail upwards along the way.

The problem is that employers tend to presume that they can continue to abuse people going forwards into the future because they’ve gotten away with it so far. Until they do things like yank offers from new college grads or laying off too many of the professional staff, at which point you’ve shattered the illusion.

tl;dr: Elon sowing: Haha fuck yeah!!! Yes!!

Elon reaping: Well this fucking sucks. What the fuck.

All things being equal, however, I’d rather they do the version more likely to remove themselves from the gene pool.

Masterful gambit, sir.

Okay, so understanding that Boeing has been shipping airliners that boing instead of fly or have some bolts missing…

My dad was a frustratingly retired aerospace engineer because there was this period of the 90s where we actually did shrink the defense industry until 9/11 and the contractors started figuring out exactly how to “bribe” people. And one of his side complaints is that any aerospace engineer is probably actually good at being a general-purpose mechanical engineer, except that they’ve generally made the hard stuff actually safe earlier along, but nobody will hire them. His example being fully-automated-digital-engine-controls and fly-by-wire and having three redundant chains.

So, in the aftermath of the whole Toyota throttle-by-wire thing that really didn’t go a whole lot of useful ways, I decided to check out his observation and I did some googling to discover a page where some big company was advertising to the auto industry at large their throttle controllers. And they talk about how they were built with “aerospace technologies” to be reliable and safe. And, looking farther along, it seems like that was not actually three redundant chains, just three threads of execution on the same processor.

Oh yeah, and generally any airplane that does have fly-by-wire and FADEC there is going to always be a set of reversion modes and people have to know about them. Obviously some aspects of this are far stricter because a car can just pull over to the side of the road… but also it needs to do that safely. Witness poor Anton Yelchin dying ignominiously because of the digital gear shifter thing on his Jeep.

But, yeah, the underlying problem is that the cultural expectation is to make cars that will go most places containing capabilities that a vehicle might never actually use in its entire service life and require the minimum amount of knowledge and basically zero knowledge above the collective cultural understanding of a car that’s only mildly changed since a fully-mechanically-linked control system.

As best I can tell, the touchscreen is added at the concept phase by folks who mostly know what’s going to make people look at the car and want to buy it, several years before the car hits the market and well before the actual car electronics teams are involved.

So, yeah, car UI/UX sucks right now because we’re seeing all of the things added to cars a few years ago in response to Tesla and implemented by people who think that just because they programmed a random car-focused microcontroller back in the day that this means that they understand all of the layers involved in a modern Linux or Android or Windows embedded car electronics unit including layer 8 of the OSI stack (meaning: interfacing with humans)

But, yah, dono. I don’t actually have my own car. My spouse got a Mazda a bunch of years ago now and it has actually a pretty good touchscreen interface with physical controls such that if you want to dig into stuff, you can touchscreen but all of the common stuff is switches and knobs. The generation before that had way way too many buttons and it was just gag-me-with-a-spoon. The generation after that removed the touchscreen because the leadership at Mazda decided people were just not to be trusted with a touchscreen and I feel like they went a little too far in the wrong direction. Meanwhile, in airplane cockpit design, they put great pains into having you be able to navigate by touch where necessary such that all of the knobs are differently textured or shaped. And, as I said, I don’t actually have my own car, but I have to say that if I did have a car, I’d want it to be designed like that.

My suspicion is that it’s because the shots are called by people who worked their way up doing automotive electronics. As in the microcontrollers inside of engine control units. So UX is kinda foreign.

I mean, even there, the marketplace of electric cars has expanded. For a while, it was the only car, then the only good car, now I see a lot of people buying fancy brand new not-Tesla cars that don’t seem to have the drawbacks of the earlier not-Tesla electric cars.

One of the problems with the automaking industry is that there’s a lot of stuff where you buy a very expensive machine and have a very expensive setup for said expensive machine (molds, templates, whatnot) and then the cost for almost everything else (materials, person-hours, etc) is lost in the noise.

Scaling up and down a car assembly line is hard so an automaker really wants to plan this out really really well such that the design is done at exactly the right time for the big switchover, you have a constant staffing level for the line, all of the machines are working roughly at their optimum capacity, and you’ve got enough spare cars so everybody can go to the lot and get a car of their choice fast enough to satisfy whatever impulse caused them to want a new one.

So I guess mayyyyybe possibly they are working to get ahead of the seasonality of car buying but I’m betting that what really happened is that they screwed up their planning, possibly because some chunk of the potential buying population doesn’t want a Tesla anymore.

We, the people who all roughly simultaneously chose the same name at roughly the same time only to engage in endless wars over who gets it on a new site, actually exist in a diverse multi-gendered animal-loving book-reading population, of which I am only one example, merely the member of our tribe who happened to nab it here. I’ve actually talked with other Wireheads and the similarities are interesting.

All I know is that it somehow appeared in my brain before I read the Niven book that introduced it to me. And, also, at this point in history, I find Niven and many people who operated in his orbit deeply disgusting and disturbing examples of humanity, so it’s good that I came up with it on my own.

The web is the way it is because TBL put out a feeler and looked at the sorts of things people were playing with around him and managed to assemble enough of them into the solid core that everybody else glommed stuff onto, and then was able to serve as a reasonable steward for enough of it for a long enough time period that it solidified into something reasonable useful.

But not perfect. [shrugs in fediverse]

Part of why it’s not perfect is because TBL was an academic person of privilege and so he’s never had to actually viscerally feel the sorts of struggle that most of the world feels. And he’s spent the past 35 years being fawned over.

For whatever reason, TBL has smacked face-first into “linked data” formerly known as RDF repeatedly for decades now. I mostly understand what’s going on there and it kinda makes sense and I’ve even written some code that uses linked data and I’ve been generally following along as it goes.

There was RDF and a version of RSS that was based on it, but then a more approachable thing ended up actually being the real RSS and instead of RDF-styled tagging, the formerly influential startup Technorati and the still vaguely popular photo site Flickr got everybody tagging because that was something people could grasp.

If you think back to the dream of a “What would a Semantic Google look like” that folks were going on about a bunch of years ago and fast-forward to today with all of the AI-generated search result abuses and whatnot, you realize that if they had succeeded at it, the web might actually been a worse place.

These days, ActivityPub is actually a proper constrained system that fits into the RDF thing that TBL has been face-smaking into. It’s actually got arcs in ways that Technorati/Flickr/et al tags don’t. So I guess we can call this progress, but I don’t know what percentage of people, even folks who care deeply about the Fediverse, actually grok the complete concept. Either way, the fediverse has pushed forward the “linked data” dream more than anything TBL has done. And part of this is that the successful fediverse apps have been grounded in being practical forward steps for people looking to jump off of a train moving in the wrong direction.

So, dono, I guess you could see a Fediverse refactor such that the data, which is already linked data anyways, ends up getting stored using Solid mechanisms with a decentralized concept of identity and Pods, et al. And if they were actually practical people, showing up with patches would be a great way to get people onboarded because there’s at least a few things there that might be useful.

But, in the end, it’s not that they are failing at being practical people who can deliver a useful solution, it’s that I don’t think they can really conceive of the important bits well enough to spend a lot of energy thinking they are making a better world when they are actually paving the way to make it even worse than had they sat around having fancy cocktails and bemoaning the state of the world and not tried to make Solid happen.

Well, the photo in question is: https://www.flickr.com/photos/cyberspace/53570370332/

Funny, just this morning I woke up to someone commenting on one of my pieces of art that I’d posted on Reddit that if I hadn’t put in the comment how I did it, they’d have thought it was an AI generated picture.

It’s super-painful to be a technologist and an artist at the same time right now because there are way too many people in tech who have no understanding of what it means to create art. There’s people in the art community who don’t really get AI either, of course, but since they are trending towards probably the right opinion based on an incomplete understanding of what the things we see as AI actually are, it’s much easier to listen to them. If anything, the artists can labor under the misapprehension that the current crop of AI tools are doing more than they actually are.

In the golden age of analog photography, people would do a print and include the raw borders of the image. So you’d see sprocket holes if it’s 35mm film or a variety of rough boundaries for other film formats. And it was a known artistic convention that you were showing exactly what you shot, no cropping, no edits, etc. The early first version of Instagram decided that those film borders meant “art” so of course they added the fake film borders and it grated on my nerves because I think it was the edges from a roll of Velvia, which is a brilliant color slide film. And then someone would have the photo with the B&W filter because that also means “art” but you would never see a B&W Velvia shot unless you were working really hard on a thing. So this is far from the first time that a bunch of clueless people on the tech side of the fence did something silly out of ego and ignorance.

The picture I posted is the result of a bunch of work on fabbing, 3D printing, FastLED programming, photographic technique, providing an interesting concept to a person and an existing body of work such that said person would want to show up to some random eccentric’s place for a shoot, et al. And, well… captions on art exist for a reason, right? It adds layers to the work to know that the artist was half-mad when they painted it and maybe you can tell by the painting’s brushwork or just know your art history really well but maybe you can’t and so a caption helps create context for people not skilled in that particular art.

And, there’s not really “secrets” in art. Lots of curators and art critics will take great pains to explain why Jackson Pollock or Mark Rothko so if you are still wandering around saying “BUT IT LOOKS LIKE GIANT SQUARES” that’s intentional ignorance.

Now, I’ve been exploring my particular weird genre of art for a while now. Before AI, Photoshop was the thing. Much in the same way as I could have thrown a long enough prompt into a spicy-autocomplete image generator, I also could have probably photoshopped it. Then again, the tutorials for the Photoshop version of the technique all refer back to the actual photographic effect.

Describing something as it’s not has long been a violation of social norms that people who are stuck in a world of intentional ignorance, ego, and disrespect for the artistic process have engaged in. In the simultaneous heyday of Second Life and Flickr, people wanting to treat their Second Life as their primary life caused Flickr to create features so people could mediate this boundary. So, on one level, this isn’t entirely new and posting AI art in the painting reddit is no different from posting filtered Second Life to the portrait group on flickr. It’s simple rudeness of the sort that the unglamorous aspects of community moderation are there to solve for.

I have gotten quizzed about how I make my art, but I’ve never seen anybody go off and then create a replica of my art, they’ve always gone off and created something new and novel and interesting and you might not even realize that what got them there was tricks I shared with them it’s so different. Artists don’t see other art in the gallery and autocomplete art that looks like what they saw, they incorporate ideas into their own work with their own flair.

Thus, there’s more going on than just mere rudeness. I’ve been doing this for a long time now and the AI companies have a habit of misrepresenting exactly what content they have stolen to train their image models. So it’s entirely likely that the cool AI picture that someone thinks my art looks like is really just autocompleted using parts of my art. Except I can’t say “no” and if there was a market for people making art that looks roughly like mine, I’d offer paid workshops or something.