Hacker News Reader

Comment by helloworld | Hacker News Reader

This is certainly an interesting case study in the pitfalls of behavioral design, but one important detail that the story glosses over is how rare these incidents are:

More than two dozen people [have been] killed by carbon monoxide nationwide since 2006 after a keyless-ignition vehicle was inadvertently left running in a garage.

On average, that's two deaths a year in a country of more than 260 million passenger vehicles.[1] These accidental deaths are tragic, but it's actually impressive how infrequently they occur.

[1] https://en.wikipedia.org/wiki/Passenger_vehicles_in_the_Unit...

View on HN

The sentence following the one you quote says:

"Dozens of others have been injured, some left with brain damage."

So deaths are not the whole story.

While one is certainly far more likely to die in a automobile collision, of heart disease, or cancer, I don't think it helps anyone to ignore the risks of carbon monoxide poisoning.

Incidentally, other sources of carbon monoxide can also exist in the home, so it's a good idea to have a carbon monoxide detector there. Perhaps cars too can be outfitted with them.

You make a reasonable point, and I don't mean to trivialize a serious issue -- carbon monoxide poisoning caused by keyless cars -- but in comparison, bathtub drownings are far more common: about 335 a year.[1]

[1] https://www.seattlepi.com/national/article/Someone-drowns-in...

Ah but what's the ownership rate of bathtubs compared to keyless start cars?

Ok, but what’s the ‘deep enough water to drown in’ usage rate of bathtubs compared to usage of keyless entry cars by their owners? These days bathtubs seem to be inconvenient showers...

You’d also have to consider bathtub usage (I shower daily, but only take a bath every couple of months)

Yeah, but the anti skid pad in my bathtub (which might prevent drownings) is slippery when damp, so I fell in the bath while taking a shower, and could have been killed.

I guess that sort of death goes under the “drowning” category, since the bath mitigation can kill shower-takers.

> I guess that sort of death goes under the “drowning” category

How could it be? Do you plug your drain when you shower? Are the investigators in your area too lazy to find head trauma?

if after falling the shower head is aimed towards your face you can probably drown. The fall only knocked you out, the water caused death.

I wonder if there is a forensic term for a "one-two punch" of this sort.

In my shower/tub combo, I sometimes accidentally plug the drain by bumping the attached drain cover with my foot, causing it to slightly turn and fall into place. This could easily happen if I fell.

Sometimes I have a slow drain and I don't fix the until it becomes slow enough to fill up substantially during my shower. If I were unconscious in the sub with the shower running, it could have enough time to fill up. This would be especially true if my fall caused the water to switch to tub mode, which sends a lot more water into the tub.

I had a friend in college who liked to take showers while pass-out drunk. He flooded the adjoining rooms several times when his unconscious body blocked the drain. Luckily this shower was not a tub or he may not have survived these incidents.

That was very informative, decent reasons not to let your drain maintenance get out of hand, or leave the stopper next to the drain.

> I had a friend in college who liked to take showers while pass-out drunk. He flooded the adjoining rooms several times when his unconscious body blocked the drain. Luckily this shower was not a tub or he may not have survived these incidents.

Jesus Christ, I wouldn't know about college, but that seem crazy. I hope your friend got the help he needed.

If you're considering bathtub usage, you'd also have to consider how many people park their cars in enclosed garages and how many of these cars lack safety features to prevent you from leaving it on.

Unfortunately this will have to entries into the Darwin awards. If you are careless enough not to double check, there are consequences. I drive an electric car, so this will never happen, I however always make sure everything is locked, all lights are off, and plugin every night.

Nice to be so perfect (or is it overchecking via OCD?) to never have forgotten your keys/specs/coat, ever left a light on or been distracted by an interruption.

Just like OPSec it’s only perfect until it isn't.

Perhaps I'm being oversensitive, but this seems a vile comment.

The Darwin award implies people "eliminate themselves in an extraordinarily idiotic manner". I don't think that doing what you were doing for years without any danger is now considered idiocy.

It's like in 40 years a toaster or an oven keeps running if you don't disable it through your "home controlling mental device" and burn your house to the ground while you are sleeping. Oh, and you cannot opt-out of the "mental device", you have to use it like it or not.

Old people have the mental muscle to think "I have the car keys with me, therefore the engine is off" built after decades. It's a serious design flaw to not offer at least an opt-out for these people (like key not IN the car, engine shuts off).

I'm surprised this is even allowed, in the UK it is illegal to leave a car unoccupied on a public road with the engine running. Obviously we're talking about private garages in this case, but the technical ability to have the car running with nobody in it seems to run counter to the spirit of the law.

As far as I'm aware, cars are modified for the UK market and they're not allowed to do this - I think normally a start/stop button is added and an alarm goes off if you get out of the car with it running.

Do you feel the same about speed limits, or minimum following distance? Because cars could enforce those now.

Where I live, the temperature regularly goes well below 0F every winter. If I start my car and get into it right away, it will be at least 10 minutes before it becomes noticeably warmer.

In colder climates (e.g., further North in this State), people may leave diesel vehicles idling for hours while at work, etc., because it can be so difficult to restart them outdoors in the cold.

There's a reason people start their cars and leave them running for a while.

A friend of mine owns a Tesla but on more than one occasion has left the engine running when he has borrowed his partner's car or had a hire car. It's amazing how quickly one gets out of the habit.

You are asking the wrong question here.

The right question to ask is how many CO deaths from accidentally leaving the car running in the garage were happening before the introduction of key fobs vs. now.

The difference amounts to the multitude of easily avoidable deaths that lie entirely on the auto manufacturer's shoulders.

A dozen people is a dozen people too many.

Stopping the engine when exiting the car with a fob should be the default, as it is consistent with the old, physical key behavior - and will prevent someone from driving off with your car while you have the key (as happened to someone in the thread).

> ...how many CO deaths from accidentally leaving the car running in the garage were happening before the introduction of key fobs vs. now.

The great thing about the free market is that information spurs innovation. News stories like these make consumers aware of risks they couldn't recognize before. Now we, as informed buyers, can seek out these safety features in the products we buy.

Yeah, no. Humans are awful at estimating any low-probability event. That's why seatbelts and air bags took so long to catch on, despite being positive changes.

Well, in fairness, another big factor in the case of air bags for sure, and seat belts to an extent, is that the auto companies fought them (https://news.ycombinator.com/item?id=10970118).

It seems that the humans in this thread are equally bad at it, considering the clamor for government force over an infinitesimally small number of admittedly tragic events.

But it's not "stop killing people" vs. "keep killing people". It's is this improvement a higher priority than any of the other changes that would also save lives?

The bottom line calculation should be total $ / death prevented. And then compare that to any of the myriad other changes or improvements possible. Like the early warning crash systems (more expensive, but probably more lives saved). Or more airbags. Or backup cameras (to prevent people getting run over). Or drowsy/alertness sensors. Or more $$ for self-driving tech that could take over if a driver becomes incapacitated (long shot and long-term but huge payoff).

One other point on the auto-shutoff is that this could lead to the deaths of children in overheated cars where the parents had no clue the car would turn off by itself. The car should either turn off immediately or not at all, imo.

Finally, why not install CO detectors on the cars and shutoff the car then? It couldn't be too sensitive, but surely could tell the difference between an enclosed space and bumper-to-bumper traffic.

Stopping the engine when one exits the car is NOT consistent with the old physical key behavior. One had to actually turn the car off. I'm not sure how that is different today. Before you removed the key, nor you press a button. But you still had to take an action. It wasn't automatic

Maybe it’s just because I’m used to physical keys but the reflex seems different. I’ve caught myself forgetting to turn off rentals that don’t have a physical key. But I’m open to chalking it up to different habits.

By forget you mean you opened the door and instantly realised your mistake because of the alarm it triggered?

I’m not sure that’s universal behavior. It’s not my recollection and I’ve seen material online suggesting only some cars do this.

Wait, you can't open your door with keyless cars? What about in winter when your car is running to heat up and you get out to clear your windscreen?

You can, it just emits a loud tone.

>Stopping the engine when one exits the car is NOT consistent ...

Attention to details: stopping the engine when one exits the car with the fob is consistent with what happens when you exit the car with the key you used to start it.

Before, you had to remove the key.

Now, you have to remove the keyfob and press the off button.

That is not consistent.

> the multitude of easily avoidable deaths that lie entirely on the auto manufacturer's shoulders

Ah, but that's not quite how it works, either. There is a certain, small risk involved with being in the world, we can't offload responsibility for any small accident on to the shoulders of manufacturers. Yes, if you can show negligence, there's a case to be made, but you can't.

> On average, that's two deaths a year in a country of more than 260 million passenger vehicles.[1] These accidental deaths are tragic, but it's actually impressive how infrequently they occur.

The deaths and injuries are utterly avoidable with decent design. Also, this is a feature that isn't prominent yet. The number of deaths is only going to get higher until the cars are built to turn the engine off under the right conditions.

> The number of deaths is only going to get higher until the cars are built to turn the engine off under the right conditions.

I assume implicit to this statement, is that some heuristic algorithm needs to be created to turn the car off. This seems like a grossly over-complicated (and failure prone) solution in comparison to just using a key for the ignition instead. If you can do something easily with a strictly mechanical simple solution, why reinvent the wheel?

> I assume implicit to this statement, is that some heuristic algorithm needs to be created to turn the car off.

If you say so. I'm pretty sure just turning the car off if there's nobody sitting in any of the seats and the key is not in the car would be a good start.

Have you never turned your car on before you left to heat up the passenger compartment?

Uhm, apart from the fact that it's actually forbidden, no. Idle takes ages to warm up the engine, it's better to just start driving.

If you live in a cold climate, you often have to let your car idle for a few minutes to heat it up, or your windows will be all foggy and you don’t see a thing.

It's also really hard on the motor if the fluids are cold, as you're more likely to induce wear because of boundary lubrication and similar effects. Additionally, like on my truck, the synchronizers on the transmission are really cranky when they're cold, and it's hard to get it moving without warming up a little bit.

You kinda need the car running for the heaters to work.

Forbidden how?

I'm not sure why remote-start exists if it's forbidden to use it.

> Have you never turned your car on before you left to heat up the passenger compartment?

With the remote start system I have a passing familiarity with, if you remotely start the car and warm it up with no person and no key in the car, the car automatically shuts off fifteen minutes after it has started if someone doesn't get in with a key. Seems to be compatible the simple guidelines I mentioned.

"Key not in the car" should be condition enough. I'm having a hard time thinking of a scenario where you NEED to have the key away from the car while the engine is running.

That would lead to far more accidents where people leave kids or animals alone in a car in direct sunlight, assuming the AC will keep them cool. But as they step away from the vehicle, the engine and AC turn off, leaving anyone in the car to die from heat stroke.

Those people don't need to take the car key with them, but with a keyless car you don't think about the key (that's the entire point of keyless cars). The key is just one of many things in your pockets.

Isn't keeping a car running with children inside incredibly dangerous? What if they release the breaks or shift to drive? I'd think taking the child with you should be safer than leaving them in a running car.

For animals that's different of course. But wouldn't an override be enough?

When I was a kid, I was at my babysitter's house waiting to be picked up by my parents. Another kid's parent showed up, parked her car, and put one of her kids in the front seat. Her other kid ran in the house, and she ran after him. The first kid let the e-brake off, and the car went rolling down the driveway and hit a tree. It didn't roll very fast, so there was no damage and nobody was hurt. But the mom was freaked out as she came back out 10 seconds later and saw the car rolling away...

This was in the late 80s. The car wasn't running. It was a mechanical e-brake. I know a lot of cars have electronic parking brakes now, but most probably still don't.

On another note, I can't think of a single car that will let you take it out of Park without a foot on the brake, or without using an override.

AC could fail for many reasons. Leaving a child in a car where they were dependant on AC for their life would be criminally negligent.

My car key's battery suddenly gives out mid-drive once and I had to listen to a warning tone every few minutes but at least it got me home. It would be incredibly frustrating if the car just decided that it doesn't see the key and shuts it off.

It also breaks the backup mechanism for no-battery situation where you can tap the fob on where the ignition lock would be for key cars to start it up.

Even simpler. If two of three conditions are met, switch engine off:

* Rising Carbonmonoxide levels * No weight on driver seat for 3 minutes * No driving activity for 3 minutes

The first factor alone should already prompt a shutdown though. It wouldn't even matter if the key was in the car, that might happen by accident and isn't a terribly good indicator.

I don't think this is actually simpler. I was deliberately trying to give an example that didn't require any new sensor tech. I don't think the cars have monoxide detectors right now.

Determining whether anybody is in the car doesn't sound trivial beyond the driver seat. The seatbelt warnings are triggered by weight, but people frequently store stuff on their seats, and occasionally that stuff has a weight plausible for a human

Why not just start an alarm 10 seconds after the key is out of range? Or automatically shut off the engine if the key has been out of range for more than 5 minutes. That still prevents accidents while driving (there were cases where children threw keys out of the window) but should be safe otherwise. And programming that timer shouldn't be a lot of work. The detection already exists anyway.

The cars need to work when the key battery is dead. My LEAF has an inductive charger near the start button that allows you to "charge" the key just enough to do the security exchange and then start it.

It's quite possible, even likely, that more than 2 people per year would die from environmental exposure if we programmed cars to not work if the key battery dies (as a proxy for "key out of range"). (Drive out to a deserted location, shut down, enjoy the day, get back in to find a dead car key battery. You want to be able to get home rather than stay at the trailhead.)

An alarm would still be fine though? Maybe with an option to deactivate it. I'd want to know when the key gets removed from the car, not having any notification appears wrong.

Mine gives an alarm when you take the key out of the car already. "key not detected" or some such.

My car displays something on the screen, but I don't recall it making a noise.

I've left the car running while stepping into a store because someone was in the passenger seat. If I get too far, my bluetooth music stops, but the air-conditioning keeps going without the key.

All of this still sounds unnecessarily complicated and Rube Goldbergian in comparison to just using a regular physical key.

The ratio to study is not how many deaths per vehicles, but cost-of-implemenation per vehicle vs deaths. The article mentions that the feature can be implemented "for pennies per vehicle." If you could save 1 person's father for $.01 per car, why not?

You could surely prevent this purely in software. The car knows that is hasn't moved, nobody is sitting in the driver's seat and nobody has touched any controls for 10 minutes. Surely it can switch itself off?

So what happens in winter when I go turn my car on thirty minutes before I leave to make sure it's nice and warm?

Could be a special switch on the fob for that. Could beep every five minutes to remind you. Lots of ways to handle it that would save lives at the cost of only minimal inconvenience.

How is beeping every five minutes any different than what cars do currently? Cars with push to start are already required to beep when the key is taken out of the car.

If you add a special switch, that increases costs. Why should car buyers bear extra expenses for less than 100 deaths a year?

Should we require special bathtub sensors that beep when they detects still water because hundreds of people drown in bathtubs each year?

> How is beeping every five minutes any different than what cars do currently?

Would've saved this guy's life. That's different.

> If you add a special switch, that increases costs. Why should car buyers bear extra expenses for less than 100 deaths a year?

How much do you think a switch costs?

If you told me that, for less than a dollar, I could save 50 lives, I'd do it in an instant. It's weird to me that you wouldn't.

30 minutes seems a little extreme. I guess it would sensibly turn itself off.

If you really want that you could have a special "warming" mode.

The pennies per vehicle line confuses me, honestly. Building it in to a car should cost zero cents, since it's just a config change, and apparently a retrofit is in the realm of $5. Where do pennies come in?

A config changes requires someone to design the configuration, right? That's at least x hours of coding, testing, and final validation. Nothing is 'free'.

Config A takes coding, testing, and validation. Config B also takes coding, testing, and validation. Neither one costs more than the alternative.

Not all 260 million vehicles are keyless.

The incident rate per keyless vehicle is likely an order, possibly two, of magnitude higher.

Keyless cars are a tiny fraction of overall vehicles.

The net befit is also minimal, so if they became popular you would expect a ~hundred deaths per year for what? Do save 20$ per car because they lack CO detectors?

CO detectors would make sense (as long as they don’t false-positive on the road, which is also dangerous!)

I suspect keyless entry systems have prevented multiple deaths of children in parking lots, but there’s no way to track that.

It would be easy enough to only use the CO detector when the car is in park or the emergency brake is on. That way it could never activate while driving.

Or just turn the car off after eg 20-30 minutes of idling?

For modern cars, I think that's what's happening actually - engines will power off when stationary and not in gear, but the car is still on and the engine will restart when it's running low on battery power. Thus, people leaving their car in the garage thinking it's off.

If the engine’s off because of stop/start the warning noise should already be much louder and accompanied by dashboard warnings. See also what happens when you leave the lights on.

That’s what doesn’t make a lot of sense here before we dash off to design some sort of countermeasure - there’s either an engine running (and “modern” engines aren’t that quiet, especially in a confined space), or there should be some loud visual/auditory warning. I guess it’s possible the victims didn’t hear either?

It should activate while driving - it just shouldn’t immediately cut the power. It should set off a blaring alarm.

Or how about CO detectors in garages rather than in cars?

Some more math, adding a 1 cent component (not counting one time costs of R&D and tooling) to every new car sold in the u.s. (~17.5 million) would equal out to 175,000; which equates to a spending of $87500 per death prevented, a reasonable amount.

On a counterpoint, 15 times more people die from lightning strikes per year than this, so are kinda going after extreme edge cases here

Yes - especially considering over 15,000 times that died in crashes.

Reply on news.ycombinator.com