We got back from the funeral in Michigan yesterday morning, and I’m slowly getting my computer functional again. I had to reinstall the OS, and I’m still not sure I’ve solved the problem that I installed a new graphics card for. I’m slowly/painfully restoring the functionality in terms of software, and configuration of things like Thunderbird, Brave, Chrome, which lost all their settings after I reinstalled the OS, despite the fact that /home is on a separate drive that I mount after boot. Anyway, I’ll start posting again soon. There’ve been a lot of interesting developments in space policy…
[Thursday-morning update]
For those curious, I was having difficulty in that the machine was running like molasses, and it was initially diagnosed by Grok as an incompatibility between Wayland and my old GPU. So I replaced the GPU (a hundred-buck RX500 card), but it didn’t solve the problem. I couldn’t get it to boot with Xorg, and long story short, I finally solved it by switching from Gnome to KDE…
[Thursday-morning update]
I’ve finally run down the issue. It wasn’t software; it was hardware. My /home drive is failing. Out to get a new one.
[Friday-the-13th update]
Drove down to Irvine to Microcenter and bought a new Seagate 4T drive. I disconnected the old one, connected it the new one, formatted it ext4, reconnected the old one, mounted both, and I’m doing rsync from the old to the new.
If cars operated like computers we’d all be dead…
I have a Tesla, the car IS a computer!
Well in a somewhat amusing turn of events here’s what Grok has to say in answer to the question: “Has the computer in a Tesla cause any fatalities?”
Yes, Tesla’s advanced driver assistance systems—primarily Autopilot and to a lesser extent Full Self-Driving (FSD)—which rely on the vehicle’s onboard computer for perception, decision-making, and control, have been involved in multiple fatal crashes.
The “computer” in a Tesla refers to its hardware and software stack that powers these features (e.g., processing camera inputs, executing neural networks for object detection, and commanding vehicle actions like braking or steering). Official investigations and reports attribute some fatalities to limitations or failures in this system, often combined with driver misuse (e.g., inattention).
Key findings from reliable sources include:
The National Highway Traffic Safety Administration (NHTSA) concluded in a 2024 investigation that Tesla’s Autopilot contributed to at least 13 fatal crashes (resulting in 14 deaths) and hundreds of collisions overall. These often involved the system failing to detect or respond to obstacles in the vehicle’s path (e.g., vehicles, pedestrians, or emergency vehicles), where an attentive human driver could have intervened. NHTSA described a “critical safety gap” in the system’s design that allowed foreseeable driver disengagement.
As of late 2025, independent trackers (cross-referenced with NHTSA data) report around 65 fatalities in crashes where Autopilot was engaged, with 54 verified by NHTSA or expert analysis. At least 2 involved FSD after 2022.
Examples include:
Crashes where the system failed to brake for crossing trucks, stopped emergency vehicles, or pedestrians visible for several seconds.
A 2023 pedestrian fatality linked to FSD, prompting a federal probe.
Cases where reliance on camera-based computer vision (Tesla’s approach, without lidar) contributed to failures in certain conditions, per investigations like a Wall Street Journal analysis.
Tesla maintains that its systems improve safety overall (citing lower crash rates per mile when features are active compared to human driving) and attributes most incidents to driver inattention rather than sole system fault. However, regulators and courts have found partial liability in some cases (e.g., a 2025 jury verdict holding Tesla partly responsible for a 2019 fatal crash, awarding significant damages).
In short, while no crash is attributed to the computer alone without human factors, the system’s performance has been directly linked to fatalities in official reports.
In short, some of Tesla autopilot’s victims have shared a common fate with Wylie E. Coyote. There were no tire marks leading up to the collision site.
Now compare fatal crashes per mile driven by Tesla autopilot vs same for humans
I think a lot of what Grok is claiming is instances of people hacking the attention monitors for Autopilot, not FSD. Full Self Driving is completely different and has improved by an order of magnitude in the past year alone.
OK, here is one for you.
Tesla fsd has different driving modes. There is a hurry mode where the care will break the speed limit and match or exceed the speed of the flow of traffic and engage in more aggressive lane changes. There is another mode called, Mad Max, where it will do even more aggressive lane changes, accelerate faster, and drive higher over the speed limit.
What happens when someone gets pulled over for speeding?
What happens if someone is drunk?
I would say that the driver is still ultimately responsible, as it is the drivers decision whether or not to use those high speed modes. The car I drive daily has the ability to break all posted speed limits in the US, but will only do so if I make it do so. Same with Tesla FSD, just a little difference in the details.
Do you have multiple .config directories?
No, just one. I don’t know why it’s confused.
Maybe you have a phantom user, say someone named Raand, or Ra[invisible character]nd? 🙂
As a joke, I just tried to create a user under Ubuntu with the full name of Fu D Bar and a user name of “.fubar”.
It took it no problem. But since this isn’t a toy system, but one I use at work for real work I canceled it. But it would be curious to know if it created a /home/.fubar which would be invisible, without an “ls -a”.
…and yes, in case you’re curious I used to make quite a good living doing things like this. Primarily to hardware not software. At least with hardware the possibilities are somewhat bounded. Software Q/A gets my enormous respect for the endless possibilities. The worst part of writing CLI apps is dealing with user input….
This reminds me of trying to make control software fool proof in the lab. I found out that graduate students are absolute geniuses when it comes to breaking fool proof systems. Sigh.
I can get a cube computer so cheap anymore, I quite fooling with separate cards. I have a Geekom with an AMD processor that has integrated Raedon graphics.
Did you also switch back to the new graphics card?
What Grok has to say about KDE/Wayland:
Yes, KDE (specifically KDE Plasma) uses Wayland—and has for several years in a major way.
Here’s the current status as of March 2026:
Since KDE Plasma 6.0 (released early 2024), Wayland has been the default session for KDE Plasma on most distributions.
Plasma 6 strongly emphasizes and defaults to Wayland, with many improvements in smoothness, HDR support, color management, explicit sync, and multi-monitor handling compared to the older X11 session.
The vast majority of KDE Plasma users are already running the Wayland session.
Looking ahead:
KDE has announced that Plasma 6.8 (expected around early 2027) will become Wayland-exclusive — the native X11 session will be dropped entirely.
After that point, X11-only applications will continue to work via the Xwayland compatibility layer, but the core Plasma desktop itself will no longer offer a pure X11 session.
In short:
If you’re using a recent version of KDE Plasma (6.x series), you’re very likely already running on Wayland by default unless you deliberately selected/logged into the X11 session at login (via SDDM or similar display manager).
Wayland is now the modern, recommended, and future-proof choice for KDE.
I’ve finally run down the issue. It wasn’t software; it was hardware. My /home drive is failing. Out to get a new one.
DO NOT ATTEMPT TO FORMAT THE NEW DRIVE!
Like last time a typo here can be fatal. Just do the “dd” from one to the other… Promise?
Also strongly suggest you CAREFULLY check the arguments you give to the “dd” command before you issue it.
I’ll probably disconnect the old drive, format the new one, then just manually copy folders over. It’s just data.
Can you make a Clonezilla image of the old drive? If so, it might be a good idea. this trick has saved my bacon more than once.
At the risk of being shunned for being off-topic, a computer in my household was having trouble booting and I spent hours following the suggestions of the A.I. until on boot it gave a message “unable to set time and date.”
That one I knew to be a bad CMOS battery, which was almost fully drained when I tested it, and replacing it perked that computer up.
Reminds me of the time we had a Mac Quadra 710 (don’t quote me on the model number, this is late 90s wer’e talking about), that eventually failed boot up with a “sad Mac”.
Not wanting to bother with it, I took it to the local Apple retailer (now defunct, you’ll see why in a minute) and told them to take a look and give me a call with their diagnosis FIRST.
Eventually a week passed. Then another. After the third week I decided to give them a call. “Definitely on the bench. Will get back to you” was my answer. After two more days passed I called again. “Oh yeah, I think “joe” (don’t remember the name) took a look at your machine. Let me ask him when he gets here and call you back”. This time they actually did. And the answer was not a good. “Yeah, ‘Joe’ looked at your machine and it’s definitely the motherboard. It’ll cost you $700+ to replace.” After which, I thanked them for their time and effort but NO THANKS on that repair. I think we may have paid around $800 for it new. I paid their small two-digit “diagnosis fee” and picked it up.
I then went on-line (had another working computer plus the ones at work ahem) and found a retailer that was selling populated MAC motherboards (populated with memory sticks that is) for around $128. So I ordered one. It arrived in the mail, swapped it out and voila, the machine booted right back up.
Then I got to looking at the two. Hmm. Put the old motherboard back in. Confirmed the “sad Mac” again. Then swapped the first DIMM stick. No joy. Then swapped the 2nd DIMM stick, still no joy. Swapped both DIMM together (only two DIMMs on this motherboard were populated as I recall). Still no joy. Then I got to looking again. Only one component left that I could swap without a soldering iron. A small battery. I presumed ran the Time-Of-Year (TOY) clock. VOILA!, booted right up on the old board. Called Radio Shack (still existed back then) to see if they had that particular 12V lithium battery (back when they were expensive and rare). They did! $12. I sent then motherboard back and had to pay a stiff restocking fee of about half the cost of the new board. I consider it to this day a lesson learned….
Interesting.
Troubleshooting a computer with the A.I. is tedious, but your experience with Apple Service was worse.
I wasn’t sure you were going to mount both drives to do the copy or just use “dd”. Since you did and are using rsync, I would have suggested just for safety you mount the old drive as read-only.
Can’t remember if that causes trouble with rsync. But I don’t think so.