> Abuelsamid said breathalyzers aren't a practical solution because many people would object to being forced to blow into a tube every time they get into the car.
Also those are super easy to defeat. People who have them usually just get someone else to blow in it or just blow into a bottle and then squeeze the bottle into the tube.
people have defeated driver awareness systems with flat-printed pictures of their face.
I don't know if defeat-hardness is a particularly good metric for these kind of systems. Very few systems on a modern automobile haven't been defeated in some way/shape/form.
I think it's more valuable to consider the convenience or life-altering-aspect of the consideration -- few people mind nor even notice their photo being taken, forcing folks to breath into a breathalyzer changes their activities and behavior ; their routine is now disrupted, that's a hard sell.
( I hate additional surveillance options, espescially given the idea of the 'always connected automobile'; it's a system that is prime for abuse by authorities. My consideration of these features in no way implies that i'm happy about either implementation. )
Mobile app using computer vision to roughly determine BAC based on facial and optic features. No one has a breathalyzer but everyone has a phone, so people will be able to get a more objective measure of their inebriation without drunkenly doing math of how many drinks they had. If a cop can estimate it with a flashlight, a phone can too.
The flashlight thing is a Horizontal Nystagmus Test, which I don't think works for passive detection. You have to follow the light with your eyes to trigger the twitching at a particular angle. I suppose you could force the test before starting the car. It's also not a terribly reliable test.
You can have the app indicate instructions for following light. This isn't about hard requirements for starting a car, imo that's quite dystopian. This would be better for heuristics for people so that they have some indicator they should not be driving and to seek out alternative means of transport.
I mean, I'm all for anything within reason than can prevent DUI, but in my travels, cell phone use is a much, much more pervasive problem that is only poised to get worse. It makes me concerned it's flying under the radar.
I'm not attempting to straw man or anything, but curious if others feel the same way.
Won't work on interstates but if they are crossing solid white lines like bicycle lanes on modern roads they they are either drunk, high, or starring at their smartphone.
Some people seem to even purposely straddle the bicycle lane line as some kind of subconscious auto-guidance while using their smartphone while driving which should be a go-immediately-to-jail offense before they murder someone.
Maybe a system that detects erratic driving, regardless of the cause -- sleep/drowsiness, alcohol, phone use, shaving, makeup, book reading (I have seen this!) -- can inform the driver to be more careful. That seems to strike a balance between safety and privacy.
You know, I was sitting here thinking on it and came up with eye tracking, but I like your idea way more. You're right, both would drive erratically and probably easily detected by the PCM/CPU.
Don’t forget talking to passengers, playing with a touchscreen, looking sideways at accidents, clipping toenails, looking back at kids, playing with gps, eating, drinking, smoking.
Do we have existing tech that can disable cellphones? I suppose the driver’s seat could be wrapped in a faraday cage. Perhaps drivers could request an exception, with tougher penalties for exempted drivers who get busted using a handheld device while driving.
On iOS, if your device is connected to the car’s Bluetooth, then you can’t use the device unless you’re using it hands-free (despite research showing that hands-free is just as distracting as holding a phone), or unless you’ve disabled this protective feature on your device.
Personally, I disable it on a per-use basis, and not within Settings.
And actually, my use case for per-use disabling is usually when I'm the driver: I turn on my car and then I pick up my phone to select some music - but I can't because the phone thinks I'm driving even though I'm parked.
But I'm sure most people have it permanently disabled.
Blocking cell phones is legally problematic because it interferes with 911 calls.
I'm sure there are technical strategies to allow true emergency calls without texting, but I feel like there's a lot of ways for that to go haywire with tragic consequences and there's no need to starting playing with that fire just yet.
From the article, it seems the type of systems that are being proposed would also help with this issue.
The general idea is that the vehicle will monitor the driver’s performance to determine if the driver is impaired or not. It seems fairly reasonable. We have lane assist “Autopilot” type systems that are relatively close as it is. Basically the computer would see the driver is not doing similar inputs as what it determines is necessary and would flag the driver as impaired no matter what is causing that impairment.
I would guess lane swerving and late braking from cell phone use would be easy to detect.
That tech is nowhere near the reliability needed to lock out vehicle access. My (relatively new) car's lane departure system regularly gets confused by wet roads, construction zones, snow, and other environmental factors. Imagine the hell that would be getting stranded on the side of the road, stuck on hold with your car manufacturer's tech support while in a snowstorm because your car refuses to let you drive.
> In contrast, its important to note that intoxication is an objective measure.
is it? BAC itself is an objective measure, and can be measured quite precisely if one cares to calibrate their equipment. but the actual level of impairment varies a lot from one person to another. all we can really say is that, statistically, most people are not good to drive after some cutoff. it's even more difficult with other drugs. a dose of heroin that would put you in the ER might be the bare minimum to stave off withdrawal for an addict.
> Someone who causes an accident while distracted can just not say they were using their phone and theres no evidence otherwise.
depends how much you care to investigate. a text message sent a minute before the crash is fairly incriminating. even if they are just scrolling instagram, there's probably a timestamped log of that request somewhere.
One of my favorite mythbusters episodes actually tested drunk driving. Tory was the only member who performed very well while intoxicated. They showed the results, but avoided the obvious conclusions:
- This was not the first time that Tory drive drunk, and
- Clearly some people can safely drive with a level of intoxication to that is unsafe for others.
(To be clear, I’m not suggesting that driving drunk is safe or sensible on public roads, just bolstering what the parent said: the BAC level for impairment differs between individuals.)
> BAC itself is an objective measure, and can be measured quite precisely if one cares to calibrate their equipment
If by equipment you mean something that analyzes a blood draw, sure. It is my understanding that breathalyzers vary wildly from person to person because the rate at which the lungs pass alcohol from the bloodstream into the air is not consistent, and they necessarily contain a fudge factor that assumes it is.
Fair points. I have no idea what percentage of people driving are drunk.
Anecdotally, I do know what percentage of people who nearly hit me head on each day are looking down into their lap(if not holding the phone in front of their eyes), and it's nearly 100.
That actually makes me think about how smart we want our cars. I know there's a subset of people that strongly dislike smart TVs for roughly the same reasons.
In the case of the car, it's dependence and having options. A simple car is a car I can count on even when times are difficult. And I've got a lot of agency on how I manage all of that.
Yep, “dumb and reliable” please. I’d hate for my car to decide I’ve become intoxicated based on some brain dead AI and shut down when it’s -20F with no cell service.
Interesting to me that the comments here are so negative, but on anything with self driving cars everyone wants it now to prevent deaths. If DUIs are the #1 source of car deaths (per the article), why wouldn’t we want that removed?
I don’t understand what your point is. First, you can make legitimate claims as to why you should own a gun. That’s not the case for driving drunk. Second, as a society we should be minimizing preventable deaths where reasonably feasible.
>as a society we should be minimizing
preventable deaths where reasonably feasible.
Only up to a point. We don't go round putting people in bubble wrap to keep them safe. We don't expect the government to station a policeman inside every house. But the lack of gun control in the USA is ludicrous as demonstrated by every mass shooting. Back to cars, why not limit every car out there to 40 miles per hour? fatality rates would plummet.
Focus on heart disease and opioid problems. Can be done at the same time.
For example: require an AI system at a drive through which makes a decision if you should be allowed to buy that quarter pounder with nacho cheese sause and half a gallon of Coca Cola based on your silhouette. Or have a scale in the seat of the car connecting to the drive through till…
> First, you can make legitimate claims as to why you should own a gun. That’s not the case for driving drunk.
Don't need to make a case for driving drunk, just need to make the case that preventative measures cause more trouble than they're worth. We had one guy try to make a shoe bomb, and now every commercial passenger has to take shoes off to fly on a plane. It's reasonable to ask where the line is, isn't it?
> Second, as a society we should be minimizing preventable deaths where reasonably feasible.
Clearly you can’t universally say that. Many preventative regulations have solved or drastically improved all kinds of situations. Car in particular have seen them in terms of minimum crash test results, seatbelts, airbags, etc.
Yes reasonable is a subjective term, which is exactly why I used it. Feasible is less subjective. Society always has to make trade offs over time in cost, freedom, etc. Which is to say taking an extreme position will lead to an extreme society. That’s not what we currently live in, nor would it be what most people want.
> Car in particular have seen them in terms of minimum crash test results, seatbelts, airbags
A crash test result and airbags make no difference in how I operate the vehicle. The only issue it causes is a higher price of the vehicle, but most people are fine with that tradeoff.
Seatbelts are a better example for your argument, but it's simple and mature tech that's easy and fast to operate. I have never once been unable to operate my car because something went wrong with the seatbelt.
If breathalyzers are as foolproof as seatbelts, do not reduce the privacy of people who have done nothing wrong, and don't add too much to the cost of the vehicle (both for delivery and maintenance) then I'd welcome them.
i would say even seatbelt warning beeps is taking it too far. Ideally it should deactivate (as some actually do) after a couple of minutes when it's obvious you don't care. The proposed technology is orders of degrees more complicated and therefore more fallible than a humble seatbelt.
Obligatory reminder that toy drones have killed a grand total of zero people but come with geo zones mandated by law that they will refuse to fly into. Meanwhile, you can get into your car and accelerate to 120 mph on city streets. Most of them have about 5x the engine power they could ever need and are actively marketed by their ability to speed.
Not really a valid or interesting comparison. If my drone has a bug that prevents it from going somewhere it should be able to, I say "darn" and play with my toy elsewhere.
That's not an acceptable outcome for a car, though.
Norms - people expressing approval / disapproval, imposing consequences and remedies similar to law, but not in a court, and not with assurances of due process, equal protection.
Law - Regulates behavior using courts, process and judgement, and can impose remedies, costs, and loss of freedom, consequenses.
Neither norms or law actually prevent people from doing things. They are both about accounting for things people have done. Notably the more resources one has, greater means and popularity generally mean lower costs and more moderate consequences.
Physics - The rules of the world do prevent actions! If it's too heavy you can't lift it, etc... The better one understand the world, the more agency one has. Access to technology also contributes to this agency.
Money - If you can't afford it, you are regulated. The more money one has, the more agency one has, generally.
What is being done here is to regulate with real teeth, preventing actions, without actually making laws and going through courts. Policy becomes law in other words, and many people oppose that because it represents a direct attack on their agency, freedom and it's very difficult to hold the people responsible accountable.
All Congress needs to do is weave in a little immunity and suddenly very large numbers of people experience regulation they may find expensive or undesirable and have very little they can do about it.
For me personally, it's another 80/20 problem. We've gotten our 80% result for 20% the effort/cost and now we're into the 80% of the effort/cost for 20% of the gain territory. At least with our current level of technology.
Meanwhile, I think there are other causes that haven't hit that inversion yet. 88,000 people die from diabetes each year. If we've got money that we're willing to splash on preventing deaths, why not take the easy win and have the government pay for everyone's insulin? There's no risk to privacy, no dubious "will this prevent me getting to work when I'm sober" questions, and no questions about efficacy.
If the goal is to prevent death, this seems like one of the most dubious methods of getting there. There are 17 million new cars sold in the US per year. If we're conservative on pricing and say it costs $10/car to add this, that's $170,000,000/year, or roughly $17,000 per prevented death. $17,000 a lot. That's health insurance premiums for like 3 families for a year. That's nearly a 5 year supply of the most expensive insulin on the market. That's enough to send someone to rehab.
The government needs to walk and chew gum at the same time. It’s not like focusing on cars takes away from diabetes, or that they’re ever considered from the same regulatory perspective.
Re diabetes, personally I’d rather see shifts in food supply and diet / offerings in the US. It’s pretty clear that fast food, soda, endless snacks, and the rest that mega food cos “innovate” is driving diabetes and a whole host of other conditions. Not an easy problem to solve under the guise of freedom to eat what you want, but we’ve been able to successfully minimize tobacco usage so why not junk food?
Government providing anything medical comes with a whole pile of political culture war.
this is a terrible idea. I want a car not my own personal nanny. Drink driving is a problem but we have to draw a line somewhere. What next? Will the car monitor music that might make us drive too fast? Will it shut off if we are not concentrating enough on the road? How many more stupid anti driver legislation can lawmakers come up with.
I would like to know if current advanced safety tech such as lane centering and collision avoidance significantly cuts down on highway deaths due to driving impairment? Since these technologies are already starting to become standard, once they get more reliable would that itself make accidents plummet?
I'm not talking about full self driving, but just the currently available assistance technologies that work 95% of the time (and if they improve to work 99.9999% of the time).
DUI of alcohol or mobile phones is a public heath hazard. A regulation mandating car manufacturers to throttle the speed and acceleration if the vehicle swerves or cross solid white lines can go a long way in saving innocent lives.
On the other hand, the only times I've ever swerved was to avoid something in the road. And if you've ever driven in icy conditions—it's not ideal but it happens—any kind of automatic correction is dangerous. There are also plenty of circumstances where it's okay to cross solid white lines.
So the devil is in the implementation details. How sensitive is it? How does it determine the difference between a driver who's avoiding an emergency versus about to cause one? What does it do when it thinks something is wrong?
Could there be a good version that informs unsafe drivers that they're not okay? Sure. There could also be a lot of bad versions.
Small part of me always distrusts these sleaziest people in Congress, I sincerely hope they are not doing it for making more use of surveillance and tracking and handing over the data to security apparatus of the state.
Would be great if they automatically get a point on their license if this registers repeat drunk driving attempts. Get these negligent, self-centered scum off the road permanently.
I agree for repeat offenders. Drunk driving, though, is a particularly tough thing to stop. It's not typically sober "joe" that decides to drive. The decision is usually made by someone not in a great state to make it.
This is a terrible idea. There's a reason existing interlock systems don't shut off your car. We have a silicon shortage that is affecting availability of cars and congress wants to require more silicon in cars?
It's stuff like this that's been happening with cars over the years that makes me deliberately continue to drive older cars exclusively. I wouldn't buy a brand new car if I were a billionaire.
Mandating something like an interlock would work. But false positives would be absolutely infuriating and possibly life threatening. Imagine having to call the police or similar to breathalyze or do a blood draw to then unlock your car. Just ask people trying to get out from under DUI convictions how unreliable those devices are.
Go after municipalities. Big cities don't have half the drunk driving problems you get in suburbia or even worse rural settings, because they have transit and walkability.
The root problem of drunk driving is that driving is the only option for many people and that's a city planning problem.
More density, more walkability, more transit, less drunk driving. Also greener, healthier living.
Time for the cars to just drive themselves so we can be free to read a book, watch a movie, even have a beer etc. Instead of stressing out trying to drive with all this control. It would give an amazing amount of useful time back to our lives. I used to hang out at work for hours while totally tired just because I was dreading that horrible rush-hour drive home.
I really hate driving now. 5km/h over the limit is an insta-fine in some areas, where the entire road is monitored. When I still drove I used to watch the speedo more than the road sometimes which can't be the intention. And now the car needs to validate my state of awareness? So I have to focus on looking awake and perky as well.
I really can't wait for self driving cars. Luckily right now where I live I don't need or own one. I hope I never will need to own a car again :)
Ps I never drink and drive but this method will probably cause false positives and cause distractions. It's just time to take this full circle IMO. Computers aren't perfect at driving but neither are people.
91 comments
[ 4.9 ms ] story [ 157 ms ] threadAlso those are super easy to defeat. People who have them usually just get someone else to blow in it or just blow into a bottle and then squeeze the bottle into the tube.
people have defeated driver awareness systems with flat-printed pictures of their face.
I don't know if defeat-hardness is a particularly good metric for these kind of systems. Very few systems on a modern automobile haven't been defeated in some way/shape/form.
I think it's more valuable to consider the convenience or life-altering-aspect of the consideration -- few people mind nor even notice their photo being taken, forcing folks to breath into a breathalyzer changes their activities and behavior ; their routine is now disrupted, that's a hard sell.
( I hate additional surveillance options, espescially given the idea of the 'always connected automobile'; it's a system that is prime for abuse by authorities. My consideration of these features in no way implies that i'm happy about either implementation. )
Mobile app using computer vision to roughly determine BAC based on facial and optic features. No one has a breathalyzer but everyone has a phone, so people will be able to get a more objective measure of their inebriation without drunkenly doing math of how many drinks they had. If a cop can estimate it with a flashlight, a phone can too.
I'm not attempting to straw man or anything, but curious if others feel the same way.
Some people seem to even purposely straddle the bicycle lane line as some kind of subconscious auto-guidance while using their smartphone while driving which should be a go-immediately-to-jail offense before they murder someone.
Oh, and call the cop immediately while at it.
And actually, my use case for per-use disabling is usually when I'm the driver: I turn on my car and then I pick up my phone to select some music - but I can't because the phone thinks I'm driving even though I'm parked.
But I'm sure most people have it permanently disabled.
I'm sure there are technical strategies to allow true emergency calls without texting, but I feel like there's a lot of ways for that to go haywire with tragic consequences and there's no need to starting playing with that fire just yet.
The general idea is that the vehicle will monitor the driver’s performance to determine if the driver is impaired or not. It seems fairly reasonable. We have lane assist “Autopilot” type systems that are relatively close as it is. Basically the computer would see the driver is not doing similar inputs as what it determines is necessary and would flag the driver as impaired no matter what is causing that impairment.
I would guess lane swerving and late braking from cell phone use would be easy to detect.
Looking forward to things like machine assisted breaking as default as the tech matures to help avoid crashes.
Additionally, “pervasive use” is not a measurement of damage.
Someone who causes an accident while distracted can just not say they were using their phone and theres no evidence otherwise.
is it? BAC itself is an objective measure, and can be measured quite precisely if one cares to calibrate their equipment. but the actual level of impairment varies a lot from one person to another. all we can really say is that, statistically, most people are not good to drive after some cutoff. it's even more difficult with other drugs. a dose of heroin that would put you in the ER might be the bare minimum to stave off withdrawal for an addict.
> Someone who causes an accident while distracted can just not say they were using their phone and theres no evidence otherwise.
depends how much you care to investigate. a text message sent a minute before the crash is fairly incriminating. even if they are just scrolling instagram, there's probably a timestamped log of that request somewhere.
- This was not the first time that Tory drive drunk, and
- Clearly some people can safely drive with a level of intoxication to that is unsafe for others.
(To be clear, I’m not suggesting that driving drunk is safe or sensible on public roads, just bolstering what the parent said: the BAC level for impairment differs between individuals.)
If by equipment you mean something that analyzes a blood draw, sure. It is my understanding that breathalyzers vary wildly from person to person because the rate at which the lungs pass alcohol from the bloodstream into the air is not consistent, and they necessarily contain a fudge factor that assumes it is.
Anecdotally, I do know what percentage of people who nearly hit me head on each day are looking down into their lap(if not holding the phone in front of their eyes), and it's nearly 100.
According to this other current HN story, this mandate aligns well with carmakers' plans: https://news.ycombinator.com/item?id=29182115
In the case of the car, it's dependence and having options. A simple car is a car I can count on even when times are difficult. And I've got a lot of agency on how I manage all of that.
With the TV, it's simple privacy and utility.
The last 18 months have shown that a lot of people disagree with that.
Only up to a point. We don't go round putting people in bubble wrap to keep them safe. We don't expect the government to station a policeman inside every house. But the lack of gun control in the USA is ludicrous as demonstrated by every mass shooting. Back to cars, why not limit every car out there to 40 miles per hour? fatality rates would plummet.
For example: require an AI system at a drive through which makes a decision if you should be allowed to buy that quarter pounder with nacho cheese sause and half a gallon of Coca Cola based on your silhouette. Or have a scale in the seat of the car connecting to the drive through till…
Don't need to make a case for driving drunk, just need to make the case that preventative measures cause more trouble than they're worth. We had one guy try to make a shoe bomb, and now every commercial passenger has to take shoes off to fly on a plane. It's reasonable to ask where the line is, isn't it?
> Second, as a society we should be minimizing preventable deaths where reasonably feasible.
"Reasonably" and "feasible" are subjective terms.
Thank you. You took the words out of my mouth.
Yes reasonable is a subjective term, which is exactly why I used it. Feasible is less subjective. Society always has to make trade offs over time in cost, freedom, etc. Which is to say taking an extreme position will lead to an extreme society. That’s not what we currently live in, nor would it be what most people want.
A crash test result and airbags make no difference in how I operate the vehicle. The only issue it causes is a higher price of the vehicle, but most people are fine with that tradeoff.
Seatbelts are a better example for your argument, but it's simple and mature tech that's easy and fast to operate. I have never once been unable to operate my car because something went wrong with the seatbelt.
If breathalyzers are as foolproof as seatbelts, do not reduce the privacy of people who have done nothing wrong, and don't add too much to the cost of the vehicle (both for delivery and maintenance) then I'd welcome them.
There is a ton of low hanging fruit in cars.
That's not an acceptable outcome for a car, though.
Your car already has tons of software that just as easily can have bugs that prevent you from playing with your toy.
My car has just worked for a long time for every use case I've needed it for.
I'd wager that nearly 100% of them involve humans driving.
Note that neither claim actually shows cause.
Regulation of humans happens via these forces:
Norms - people expressing approval / disapproval, imposing consequences and remedies similar to law, but not in a court, and not with assurances of due process, equal protection.
Law - Regulates behavior using courts, process and judgement, and can impose remedies, costs, and loss of freedom, consequenses.
Neither norms or law actually prevent people from doing things. They are both about accounting for things people have done. Notably the more resources one has, greater means and popularity generally mean lower costs and more moderate consequences.
Physics - The rules of the world do prevent actions! If it's too heavy you can't lift it, etc... The better one understand the world, the more agency one has. Access to technology also contributes to this agency.
Money - If you can't afford it, you are regulated. The more money one has, the more agency one has, generally.
What is being done here is to regulate with real teeth, preventing actions, without actually making laws and going through courts. Policy becomes law in other words, and many people oppose that because it represents a direct attack on their agency, freedom and it's very difficult to hold the people responsible accountable.
All Congress needs to do is weave in a little immunity and suddenly very large numbers of people experience regulation they may find expensive or undesirable and have very little they can do about it.
Meanwhile, I think there are other causes that haven't hit that inversion yet. 88,000 people die from diabetes each year. If we've got money that we're willing to splash on preventing deaths, why not take the easy win and have the government pay for everyone's insulin? There's no risk to privacy, no dubious "will this prevent me getting to work when I'm sober" questions, and no questions about efficacy.
If the goal is to prevent death, this seems like one of the most dubious methods of getting there. There are 17 million new cars sold in the US per year. If we're conservative on pricing and say it costs $10/car to add this, that's $170,000,000/year, or roughly $17,000 per prevented death. $17,000 a lot. That's health insurance premiums for like 3 families for a year. That's nearly a 5 year supply of the most expensive insulin on the market. That's enough to send someone to rehab.
Re diabetes, personally I’d rather see shifts in food supply and diet / offerings in the US. It’s pretty clear that fast food, soda, endless snacks, and the rest that mega food cos “innovate” is driving diabetes and a whole host of other conditions. Not an easy problem to solve under the guise of freedom to eat what you want, but we’ve been able to successfully minimize tobacco usage so why not junk food?
Government providing anything medical comes with a whole pile of political culture war.
This madness must be stopped.
I'm not talking about full self driving, but just the currently available assistance technologies that work 95% of the time (and if they improve to work 99.9999% of the time).
On the other hand, the only times I've ever swerved was to avoid something in the road. And if you've ever driven in icy conditions—it's not ideal but it happens—any kind of automatic correction is dangerous. There are also plenty of circumstances where it's okay to cross solid white lines.
So the devil is in the implementation details. How sensitive is it? How does it determine the difference between a driver who's avoiding an emergency versus about to cause one? What does it do when it thinks something is wrong?
Could there be a good version that informs unsafe drivers that they're not okay? Sure. There could also be a lot of bad versions.
It's stuff like this that's been happening with cars over the years that makes me deliberately continue to drive older cars exclusively. I wouldn't buy a brand new car if I were a billionaire.
The root problem of drunk driving is that driving is the only option for many people and that's a city planning problem.
More density, more walkability, more transit, less drunk driving. Also greener, healthier living.
I really hate driving now. 5km/h over the limit is an insta-fine in some areas, where the entire road is monitored. When I still drove I used to watch the speedo more than the road sometimes which can't be the intention. And now the car needs to validate my state of awareness? So I have to focus on looking awake and perky as well.
I really can't wait for self driving cars. Luckily right now where I live I don't need or own one. I hope I never will need to own a car again :)
Ps I never drink and drive but this method will probably cause false positives and cause distractions. It's just time to take this full circle IMO. Computers aren't perfect at driving but neither are people.
Either both mandates make sense or neither does. I think you can probably tell which camp I'm in.