admiralteal

We can’t claim to know it left them with “bad” employees. I think there’s vanishingly little evidence that recruiters actually go after the “good” employees effectively – I’m pretty skeptical that a pro recruiter actually gets you better employees, they just make the process of getting employees way less stressful. We also have no reason to assume that a good or bad employee is correlated in any way with caring about not returning to office – it’s possible very bad employees are just as likely to quit as very good ones. How do you even tell good from bad, anyway?

What this “return to office” stuff definitely DOES do is preferentially retain the most obedient/desperate employees. Which may be part of the goal, along with low-key downsizing.

Musk told workers that Tesla “will continue to build out some new Supercharger locations, where critical, and finish those currently under construction.”

Sounds to me like the plan is to finish what is already under contract and do no more. I sure am glad the US authorities committed to that north american charger standard… what’s even the status on getting a full specification for it including third-party development at this point anyway?

I can’t pull a quote for the new vehicle development team’s situation because Tesla basically just keeps making the Model 3 with barely even incremental improvements to it, and even that one has totally inconsistent build quality vehicle to vehicle. Unless someone thinks the Cybertruck is going to save them – hah.

Not really. With the super easy, friendly distros it basically just goes.

I switched to Linux Mint Cinnamon a while ago expecting to just fool around a bit but mostly boot back into windows to do stuff. I’ve now found that the ONLY thing I need to go back to windows for is when I’m forced by dumb policies to use an MSOffice product, which fortunately doesn’t happen to often (and no, LibreOffice is absolutely not a sub for MS Office. The spreadsheet app is worse than google docs, and I’d rather work in typst than have to deal with the libreoffice writer – especially as soon as I need to display an equation/figure/table of contents. Of course, I’d rather work in typst than deal with MSWord too…)

That said, I don’t really play games anymore. Games may still require frequent windows visits. But… I’ve been looking forward to a complete edition of horizon forbidden west and all accounts say it’s linux compatibility is near perfect, so maybe things aren’t so bad these days on the gaming front.

Out-of-date iFixit Article

Follow up where Sony made the good ones worse

But Fairphone does always give us hope

That was the point that hit my limit, now that you mention it – getting it to show up on a duckdns address on the https public internet. Not being able to make that work after fiddling with all kinds of contradictory guides nor with 2 or 3 completely different reverse proxy tools just left me mad. Especially since a regular ngix reverse proxy manager container works fine on the same computer, but for some reason was just refusing to connect to HA (SSL issues, I think).

Having HA just working locally didn’t really make it a replacement for the big tech solutions that already work fairly smoothly. I’m sure I’ll go back to get it the way I want one day, but the learning curve on any selfhosting is still pretty rough.

I had it briefly up and running and can only say… it’s a bear, at least if you are trying to use it as a drop-in replacement with existing hardware. I’m sure I’ll go back and sort it out at some point, but it left me just feeling tired and frustrated even when I had it doing most of what I wanted.

If you were thoughtful about hardware from the ground up, maybe it would be more straightforward, but I tried getting it running on just an old workstation with ubuntu installed on it that I use for very basic stuff like syncthing and it was just painful. Mix of Kasa/Wyze/Philips devices that are just what I’ve somehow collected over time.

It would be nice to see better first-class add-on support. I found myself needing to SSH into a VM to get stuff into it, and even then it was twitchy in all the wrong ways. Would also be nice to see better support for the containerized version, because that’s so much easier to distribute and execute compared to a VM. Next time I’ll probably just try to do it all with docker and see if it hurts less, since I don’t think any addons I was using were critical to begin with.

That said, if you’re doing HA, get a dedicated piece of hardware for it. I suspect it vastly simplifies things.

This technology existing would essentially be the end of all knowledge-sector jobs, instantly. It eliminates the value of your time, which means the labor market would almost definitely use it to pay wages VASTLY below minimum wage-per-perceived hour. People would take that bargain. You only have to work ONE day a week and we pay you a million dollars a year! …that one day a week will be time chambered up to 15 years, though.

Why pay one ace coder a bigshot salary when you can pay a whole village in the developing world to spend as much time as the problem could possibly need the same price and they’ll still finish by Thursday?

The economic ramifications are just beyond my fathoming, but I know it cannot possibly work in a society which has any kind of resource scarcity.

Sounds like a compelling argument for why we need better safety standards for cars and traffic engineering.

The argument for drive-by-wire in personal automobiles is basically that it’s safe enough for airplanes, so it should be safe enough for cars.

I mostly buy that. But there’s a glaring omission in the reasoning.

In airplanes, there’s a full incident investigation for EVERYTHING that goes wrong. Even near misses. It’s an industry that (mostly lol boeing) has a history of prioritizing safety. Even at its worst, the safety standards the airline industry and air transportation engineering are orders magnitude more strict than those of the automotive industry and road engineering.

In real terms, automobile incidents should be taken just as seriously. Even near misses should have reporting and analysis. Crashes should absolutely have full investigations. Nearly all automobile deaths are completely avoidable through better engineering of the road systems and cars, but there is mostly no serious culture of safety among automobiles. We chose carnage and have been so immured by it that we don’t even think it’s weird. We don’t think it’s weird that essentially everyone, at least in the US, knows someone who died or was seriously injured in a car accident.

So yeah, we should have drive-by-wire. But it should also include other aspects of that safety culture as part of the deal. “Black box” equivalents, for example, and the accompanying post-accident review process that comes with it. A process that focuses not on establishing liability, but preventing future incidents, because establishing liability is mostly a thought-killer when it comes to safety.

…of course, if we actually took road safety that seriously it’d be devastation to the entire car industrial complex. Because much of that industry is focused on design patterns that, in fact, cannot be done safely or sustainably.

Particularly goofy because ChatGPT is hardly the only bot and you can use the free version of e.g., Claude and get those better results now, for free.

I did the same thing twice.

Two different employers that really deserve to be absolutely thrashed but as soon as I got to the point where it was asking me my true identity I realized there was no hope it wouldn’t come back to bite me in the long run.

I understand why in their business model they want to be able to verify employment. I’d even say it’s reasonable. But the Privacy implications of it are just too tremendous and they I’ve never been practically or systemically trustworthy.

And knowing this about them means they aren’t a reliable place to be warned off of a bad employer either. The primary purpose of their site is completely undermined by this bad policy.

I love having mysterious cables that may or may not do things I expect them to when plugged into ports that may or may not support the features I think they do.

As I understand it, detecting an adblocker is a form of fingerprinting. Fingerprinting like this is a privacy violation unless there is first a consent process.

The outcome of this will be that consent for the detecting will be added to the TOS or as a modal and failing to consent will give up access to the service. It won’t change Youtube’s behavior, I don’t think. But it could result in users being able to opt out of the anti-adblock… just that it also might be opting out of all of YouTube when they do it.