FWIW, here are numbers:

Gasoline-powered cars: 1530 fires per 100,000 sold
Hybrid cars: 3475 fires per 100,000 sold
Electric cars: 25 fires per 100,000 sold

Source: https://www.kbb.com/car-news/s...

An interesting aside: The delivery trucks used by the US Postal Service for the last 40 years are crazy fire-prone. I learned this recently because my brother is a mail carrier and a few weeks ago he was out delivering when he saw wisps of white smoke from under the hood. He pulled over and jumped out to investigate, but by the time he got out it had turned into large black streams of smoke and by the time he walked around to the front he could see small flames. Instead of popping the hood he ran to back and started pulling the mail out. The fire truck arrived about two minutes after he pulled over (this all happened about 100 yards from the fire station) so a couple of packages in back got a little damp but he got everything else clear.

His boss brought him another truck about 10 minutes later and he loaded it up and finished the route. Neither snow, nor rain, nor external engine combustion stays these couriers, you know.

Mail truck fires are so common there's even a web site that tracks them: https://mailtruckfires.com/. This is my brother's truck: https://mailtruckfires.com/uta...

There are four or five distinct serious design flaws that make mail trucks fire prone. Mail carriers are excited about the new trucks that are coming, which should be much better. Many of them will be EVs.

I believe the plan was to give that sobriquet to Lake Michigan, in honour of its decidedly flaccid silhouette.

I had a problem where AI scrapers were absolutely DETERMINED to fish out every possible query string from a search results page. Almost all of the query strings they tried were invalid due to shitty and dysfunctional string substitution. "&page=100" wouldn't be followed by "&page=101", it would be followed by "&pag&pag=1010" or something even more insanely half-baked, until the query strings were like 100+ characters long. It was the technological equivalent of watching HIV mutate in real time.

But the insane thing was that, aside from page number, they were always requesting info about the same other criteria: filtered by the same user, the same page type, and with no text string. So I just took those particular values and started banning logged-out users who requested that combination of criteria.

I figured I'd need to change my tactics in a couple of days once the botnet got bored of that particular page and moved on to requesting bogus entries for another user.

MariaDB> select count(*) from ip_bans;
+----------+
| count(*) |
+----------+
| 671671 |
+----------+

It hasn't.

No. I refuse to allow Trump to redefine conservatism. True conservatives may be criticized for caring more about economic competitiveness than people, but Trump doesn't care about either.

I do not think that word means what you think it means.

My 2024 Kia already has a better version, so really not new. Radar warns you you're about to get "doored". I suppose I could open it anyway and lose the door, but it doesn't stop you. And it works on motion, so parked cars don't set it off.

BMW Blames Young People

That sounds intentionally deceptive / deflecting. Makes it sound like Young People broke into the R&D labs and rewrote the blueprints, and forced the manufacturers to remove physical buttons, without their consent or ability to say NO.

This isn't about them being responsible for it, it's about car marketing analysts doing research on their customers, and coming to the conclusion that this is what they will buy, while paying little to no attention to what they really want or need. Without thinking about how this will pan out down the road. And now they're seeing all the backlash and pushback and are hunting for someone to blame.

It's like the parents blaming their kid for his cavities, while ignoring their parental responsibilities. "We gave him the candy because he liked it, so the outcome is all his fault!"

Nobody wants to take responsibility for the state of modern car interfaces, because they're all doing it, and the first one or two that accept responsibility are going to take a temporary hit to their image. They all want to change back, but instead they're all standing around looking at each other, "well, c'mon, you do it first!" so they can all quietly make the same change in a year or two and escape the hole they're in. They all want to change back, but nobody wants to be first.

Reminds me a lot of headphone ports on smart phones. Pretty much at once everybody added a headphone port to their phones. It was a bit of a design nightmare, being physically tethered to your phone, but everybody went "all in" on it, and nobody could give it up because they knew their rivals would all jump on the opportunity to lambast their rival for dropping such a mainstream feature. Finally, after bluetooth was starting to mature, Apple finally bit the bullet and removed the jack. And just as expected, multiple big names dragged them through the alley in their marketing, for having gotten rid of a feature that "everybody loved". They kept up the gas-lighting while their designers quietly erased the headphone jack from their next design. Then like a switch, they quit panning Apple's lack of headphone jack as they released their next model that didn't have one either.

I expect the same thing's going to happen with car interfaces. A few manufacturers are going to bite the bullet and make the necessary changes, get slammed in the media for a year or two, and then everybody else is going to quietly make the same change.

So it's already on by default? And they are enshittifying the non-AI part of the service to push you to leave it on... Sounds great.

As hokey as something like that would be, I'd still pay to see it too.

The reality is that most of the auto manufacturers' design choices, whether it is the shape of the car, the trim, the gadgets or the controls, are driven by only a few things:
1) Regulatory requirements and fuel efficiency
2) Cost
3) The Chinese consumers.
The last one is a considerable factor.

In modern cars, real buttons and switches don't directly control the equipment they are for, they tie into a bus connected to a central controller that handles the car's functions. They can still add a subscription model on top of that.

Inference is neural network limited. On systems with entirely separated data and compute, this is a memory bus constraint. In quantized systems, ALUs and FPUs are illogical. A simple 16 cell LUT is suitable for noise free multiplication. A neural network circuit with a dedicated LUT and addressable buffers more or less eliminates bandwidth constraints. But this requires neural networks more similar to FPGA cells rather than classic compute.

That said, separating inference from data drops network depth considerably. And in transformers depth increases memory access exponentially which is why nearly identical MoE and dense models perform so differently. We of course need data centers for models with massive numbers of active parameters. But massive numbers of active parameters weaken models as that means depending on training rather than looking up facts. Huge active parameter sets will never be a good design. A real world analogy is that huge models are like Jeopardy champions who have crap loads of partial trivia facts in their heads. Smaller models are like librarians who don't have a trillions useless facts memorized but knows how to find the data for the researchers who will use the librarian repetitively to follow through to the answer. The model only needs enough training to use the data.

What really matters is, how fast it can find the data.

Inference suffers greatly when you treat weights as data. And bigger models have more trivia like facts memorized from their training. This means more active parameters and greater network depths with higher memory bandwidth needs. Even now, you see the chatbots are improving drastically because they rely substantially more on data and way less on inference.

I am not sure which part you consider gibberish, but it makes perfect sense to me. But it might sound better in my head.

P.S. I actually have considerable data to backup many of my claims. At work I'm sitting on about 50MW of compute. We build in shipping crates in a repurposed mine. There is a group of us where our goal is to avoid spinning up more compute. We need to deliver inference to 200,000 users eventually. (not customers, employees) and we could throw millions at NVidia and end up hosting some crap model, but we focus instead on cutting that memory bandwidth need. And yes, I'm far from being the smart guy on the team.

I've seen many examples of ICE vehicles catching fire. It is absolutely not true that they are "electrical systems of the cars catching fire." ICE cars catching fire are simply so commonplace that they don't even make the papers, unless some celebrity like Jay Leno is burned.
  https://normantaylor.com/blog/...
  https://thedailyautomotive.com...

Having a regulation on the books and zero enforcement is still a regulatory problem.

No. It means "correlation". If you find yourself with a thumb on the scale of causation, I think you owe it to yourself to ask why.