Live data from Hacker News

Software is eating the car

spectrum.ieee.org

421–430 of 549 posts

Re: Software is eating the car

#421

Short answer to the lede: no, the industry cannot cope. Or rather, it will limp along with bloatware, bugs, and malware exactly the same way we see desktop OSes bloat, or the way we see routers and set-top boxes hacked to become botnets. In my 40+ years in the industry I've yet to see code get SMALLER. With the exception of Linux kernel 1.0 in the 90's which was a step backwards into smaller, more compact code, code…

>"I just want a car with as FEW knobs/buttons/levers as necessary."

I also want "dumb" car. Not completely dumb but say the equivalent of ones released before 2010.

Re: Software is eating the car

#422

Short answer to the lede: no, the industry cannot cope. Or rather, it will limp along with bloatware, bugs, and malware exactly the same way we see desktop OSes bloat, or the way we see routers and set-top boxes hacked to become botnets. In my 40+ years in the industry I've yet to see code get SMALLER. With the exception of Linux kernel 1.0 in the 90's which was a step backwards into smaller, more compact code, code…

My dad bought one new car in the 90's.

It was a Dodge Dakota. He demanded a free Service manual from the salesman.

Conversation with my dad in my car after the sale.

Son---I demanded a Service manual because that truck has a computer, and I don't like computers. Always ask for a Service manual.

(The salesman would probally laugh if you requested a Service manual today.)

Son---I'll put in my own stereo. That overpriced factory cd player is a waste of money.

Son---power windows are just something that will break at the wrong time. If I ever get to the age I can't physically roll down a window; shot me.

Different time?

Re: Software is eating the car

#423

Earlier quoted context omitted.

> "Modern cars are absolutely terrible in the UX department, but they are a hell of a lot safer, so there’s that." tangentially, "safety" is highly cargo-culted. things that seem so obviously safer are taken without question as better, but in many cases, such features really only provide a false sense of security along with substantive unintended consequences. most safety features in cars (e.g., lane-keeping) allow p…

Modern cars have more design features that improve crash survivability (better airbags, better crumple zones, tested with more realistic crash tests, etc). Those seem like a pretty unalloyed improvement to me. I'm not saying you're wrong about the things you mentioned, but cars really have got safer, in important ways.

Unless you crash into something going highway speed a car from 1990 or 1980 is fine as long as you're wearing a seatbelt.

The states look good because safety tech that only matters at the extreme end has improved so the drunks and distracted teenager who would have dies had they gone off a cliff in a 20-30yo car are now surviving.

Basically the medium to low speed impacts that people not behaving particularly poorly get into have always been highly survivable.

Re: Software is eating the car

#424

Earlier quoted context omitted.

if you’re prone to being overwhelmed by stress while driving a machine that can potentially kill you, that’s a sign to get more training, not to mollify oneself with an illusion of safety. ignorance is not bliss in this case.

if you’re prone to stress while driving a machine that can potentially kill you, that’s a sign you are aware of your predicament

Have you ever seen how stressed out drivers behave?

We've all seen that person in a Honda fit who's so terrified of being on the freeway during rush hour they're driving like an overloaded scrap hauler and generally causing a problem in whatever lane they're in.

That doesn't make the roads safer.

Re: Software is eating the car

#425
post #33

Earlier quoted context omitted.

> If electric cars ever become widespread I have a strong suspicion that soon enough after that we will be able to build electric cars from kits, cheap and with no frills. We are on a cusp of a new technological revolution.

> soon enough after that we will be able to build electric cars from kits I'm skeptical. An internal combustion engine is a complicated, engineering marvel. But a complete engine, as a unit, isn't difficult to remove/insert. It's big, heavy and awkward, sure (so are electric motors and batteries), but an experienced person can do an engine swap in a couple of hours. And yet there aren't many people building ICE cars…

>And yet there aren't many people building ICE cars from kits. Why not?

There is basically no market for kit cars because if people want a particular kind of vehicle it probably already exists. If they want a particular kind of powertrain it probably already exists. So it's just a case of getting them and combining them. And there are kits that make the most popular combinations a bolt in deal.

Re: Software is eating the car

#426
post #280

Short answer to the lede: no, the industry cannot cope. Or rather, it will limp along with bloatware, bugs, and malware exactly the same way we see desktop OSes bloat, or the way we see routers and set-top boxes hacked to become botnets. In my 40+ years in the industry I've yet to see code get SMALLER. With the exception of Linux kernel 1.0 in the 90's which was a step backwards into smaller, more compact code, code…

Cars are legally required to have infotainment systems (for the backup camera) by NHTSA safety regulations. We’re fucked.

This is GOOD. Drivers run over children so frequently that they need this system. We should not be putting drivers in a machine they can not safely operate.

Re: Software is eating the car

#427
post #162

Earlier quoted context omitted.

I recently purchased a Kia Telluride. Perhaps I'm now biased because I love the car, but I think it has an excellent combination of tech and usability. 1. It still has buttons and knobs for the things that should be buttons and knobs (e.g. climate control, volume, etc). 2. The heads up display is the killer feature that should be standard on all cars (as is only available on the top speced Telluride). It displays spe…

Regarding HUDs... it's the one feature I desperately want for my next car. I've been looking into aftermarket options, but all of them are ridiculously cheap and bad or vaporware.

Lot's of Mazda's have HUD. And generally good usability.

Re: Software is eating the car

#428
post #267

Earlier quoted context omitted.

> If I'm the dark pattern developer I'm not logging when the EULA is accepted or how many times they declined it This cuts both ways. If the user can show one case of their rejecting the EULA without it being logged, they can then make the claim--correctly or not--that they repeatedly rejected it. If the car refuses to work without the EULA being accepted or rejected, proof of its movement would be sufficient to show…

There is hiding information and not collecting it to begin with. void handleEula() { auto eulaAccepted = readEulaAcceptedFile(); if (!eulaAccepted) eulaAccepted = promptForEulaAcceptance(); writeEulaFile(eulaAccepted); } This function reads a file from disk and writes back to it every time. If filesystem write audits aren't enabled there is no way to determine if the most recent write of "true" was the only write. >…

> ... falls apart for me when I consider that the user would have to prove they never did something, except that one time.

It falls apart for you, since you are the person trying to trick every single driver. It won't necessarily fall apart for a judge.

Re: Software is eating the car

#429

Earlier quoted context omitted.

This is missing the point - Person A wants their next car to have certain features, but person B also wants their next car to have certain features which person A doesn't want. Car companies aren't going to make 200 car variants with different features combinations, they're just going to group all the features people want into new cars and ship it to everyone. To stay competitive they just put out whatever it going t…

I was going to say something like this, that products (like a lot of things) are a combination of multiple users' wants, but I'd say that the main problem here is that the implementations themselves are becoming more monolithic. The extreme case (as you mention) is the moment you stick in a touch screen and a fair amount of intelligence. You might as well put a jillion features in there plus you save money on physica…

Automation drives down costs for standardized, mass produced items, but it may also remove the economies of scale that made it possible for more variety to exist due to decreased demand for those varieties.

Re: Software is eating the car

#430

Short answer to the lede: no, the industry cannot cope. Or rather, it will limp along with bloatware, bugs, and malware exactly the same way we see desktop OSes bloat, or the way we see routers and set-top boxes hacked to become botnets. In my 40+ years in the industry I've yet to see code get SMALLER. With the exception of Linux kernel 1.0 in the 90's which was a step backwards into smaller, more compact code, code…

> Damn. I just want a car with as FEW knobs/buttons/levers as necessary.

In 2005, there was a huge publicity campaign to make cameras mandatory in cars on the basis motivated by some parents driving over their toddlers playing in the driveway.

Of course, there are many rather more straightforward solutions to that problem. Given the fact that such parents must be in a rather select group, I doubted having cameras would prevent other sources of injury to their children even if they prevented parents from moving them down.

Now, auto manufacturers had to put a camera in every car[1]. Yes, the requirement took effect in 2018, but the manufacturers knew it was coming.

So, you have a screen in the car. In addition, there is the temptation to monitor and collect information. You can sell that to insurers as "anonymized" data. Etc etc. You can enhance the tracking dimensions a lot if you can also get to track all the other stuff people do. Offer them the honey of pairing their phones with the entertainment system, and, boom.

> 221 people were killed by non-traffic (not on public roads) backover crashes in 2007, and 14,000 people were injured.[1]

Given that the value of a statistical life is about $10 million, if the backup camera eliminates all such deaths and injuries, its benefit is about $5 billion. Conveniently, we do not know how many fatalities and injuries have been caused by trying to change stations or select tracks on a touch screen. So, on the cost side, we are stuck with the cost of all the equipment, software development, bug tracking, and, above it all, the annoyance of having to live with all this additional equipment one cannot avoid for the foreseeable future.

Here are sample stories:

* https://www.houstonchronicle.com/news/over-in-an-instant/art...

* https://www.nytimes.com/2005/03/12/nyregion/2yearold-run-ove...

* https://www.kidsandcars.org/how-kids-get-hurt/backovers/

* ...

Ah, the Houston Chronicle stories has this nugget:

> A cost-benefit analysis, required as part of the rule-making process, put the cost of each life saved at up to $19.7 million.

It's all for the children. How there you object?

[1]: https://www.thedrive.com/news/20612/rearview-cameras-are-now...

Post reply on HN