12 ms·

We allow companies like Tesla to jeopardize people's lives while the company beta tests "full self driving" technology, but are scared of allowing proven safe t

by prng2021 4y ago

We allow companies like Tesla to jeopardize people's lives while the company beta tests "full self driving" technology, but are scared of allowing proven safe technology like adaptive headlights.

bb88 4y ago
You're not wrong, the NHTSA should be more proactive on this stuff. But to be fair, drunk drivers kill 28 people a day in this country, or about 10k people per year. 3k people die at signalled intersections every year, or around 8ish per day. While Tesla shouldn't be beta testing their self driving cars on public roads, it's clear that human behavior and existing road design is still vastly worse.
- prng2021 4y ago
  Yes drunk driving is a terrible problem but beta testing software on the open road is not the only option available. It's nothing but irresponsible. Here's one of many: use sensors and AI in the car to detect a drunk person behind the wheel. This is HN so I'm sure readers could spend 10 seconds and think of numerous ways this could be accomplished and many otherideas about what could be done when a drunk driver is detected. I also think the NHTSA should have prevented companies like Tesla from behind allowed to put every single control behind a touchscreen. It's extremely dangerous to have to look away for 2 seconds when you're driving. Imagine if (when?) every car in the world adopted this and all drivers are having to look at a screen to turn on windshield wipers.
  - bb88 4y ago
    You may find it interesting the NHTSA is pushing for breathalizers in cars starting in 2026. Personally I'm more in favor of smart road like technologies where a highway has a mechanical or physical properties to assist self-driving vehicles stay in line with the road. Mostly because a 100% software solution will likely not work, particularly in the midwest and great lakes states during the wintertime. It need not be expensive either. Magnetic nails embedded in the road would maybe be the cheapest way, and would be easy for both city and rural populations to add, and be detectable with icy and snowy conditions.
    - andrewflnr 4y ago
      Of course they're doing something dystopian instead of something sensible. But yes, I bet even better standardized road markings would be a huge help for self driving cars. I wonder if there's something stupid simple like a paint additive that would make it easy for machines to track. Magnetic beads?
      - waste_monk 4y ago
        Then any asshole could mix up their own magnetic bead paint and cause havoc. Probably no more than someone painting their own road lines or putting up fake signage does now [1], but if you want something that's better than what we currently have it would need to not only be both machine and human readable, but also authenticated. I always thought it would be quite neat to have those portable roadside displays you see for roadworks and so on also communicate map and driving data (cryptographically signed by your local council or dept of transport). So you might get a sign that says "roadworks ahead, please slow down" for human drivers, and the radio beacon would communicate a change of speed limits inside a certain geo-bounded area that would cause self-driving vehicles to slow down. Or similar stuff for "bridge out ahead, turn around now" causing navigation to reroute. [1] https://techcrunch.com/2017/03/17/laying-a-trap-for-self-driving-cars/ https://techcrunch.com/2017/03/17/laying-a-trap-for-self-dri...
        sgtnoodle 4y ago
        Just have the human readable signs be reasonably standardized, and the self driving car computer can then read them too. It's a lot easier and more future proof to side-channel map updates to cars through a cell network, and allows for some curation. Also, improved cell network coverage seems like a more worthwhile capital expenditure for a local government than a bunch of smart signs that only help 0.001% of drivers. Cynically, I'm sure some politician's spouse's cousin would be happy to win the contract to slap a BLE beacon into a box and bolt it onto a sign for $150k a unit, though... There is an ongoing effort to introduce infrastructure free direct communication for vehicles, using 5.9ghz spectrum. The spectrum was previously allocated to DSRC radio technology the automotive industry pushed for, but it's been 15 years and it hasn't gone anywhere. The FCC is now letting the cell network folk take a crack at solving the problem instead.
  - m463 4y ago
    Have you driven a tesla? I think they help with the driving. That said, touchscreen controls (with tiny targets) instead of dedicated controls and the new square steering wheel with touch controls instead of stalks... gah!
  - wheybags 4y ago
    No need for fancy sensors and ai - I used an electric scooter rental in Germany recently, and it made me do a reaction speed test on my phone before starting. Super easy.
    - JamesBarney 4y ago
      Are they worried you'll kill someone with your electric scooter?
      - exclusiv 4y ago
        It happens as it causes swerves and crashes. Deer do this too. With a deer your best course of action is to try and stop but mow it over if necessary. With a person swerving on a bike or moped, you may not be able to stop in time and most likely swerve.
      - willsmith72 4y ago
        They're worried you'll kill yourself
        JamesBarney 4y ago
        If people can get on an escorted i wonder if they wall or uber. If they walk i could totally see the total number of fatalities being hire than escooting. Similar to how drunken walking has a higher fatality raid than drunken driving/biking.
      - Broken_Hippo 4y ago
        Yes, especially yourself if you aren't wearing a helmet. More commonly, though, the electric scooters cause injuries and accidents. Starting next year, Norway is requiring insurance for them and rental prices will surely increase in response.
      - twobitshifter 4y ago
        It is inexplicable how e scooters are treated like death machines in comparison to 4000 pound SUVs which get no restrictions. Scooters have regulated speed, zones that are banned from use, reaction tests, sidewalk restrictions and on and on. You would think there were thousands of e scooter deaths per year.
  - JamesBarney 4y ago
    If it gets us to autonomy 6 months earlier that could easily save 50x as many lives as deaths it's caused.
- Rebelgecko 4y ago
  Didn't the infrastructure bill have a clause saying that all cars have to start using alcohol detecting interlocks in a few years?
- m463 4y ago
  They say in this article: the National Transportation Safety Board conducted an investigation finding that the government should not be actively restricting headlamp systems that have been shown to have a safety benefit. So I can imagine this sort of thinking coming down on telsa's side of the argument. Personally, although tesla does "market" their technology, I think it is good technology that has probably not only saved a fair number of lives, but also bumpers and fenders. Looking at the storied history of the airbag, I think telsa did a good job assisting the driver (and customer) when the status quo for other car manufacturers was to innovate at a snail's pace and implement safety features mostly when required.
- silisili 4y ago
  Drunk driving is a legal issue. The entire problem is that it's not treated seriously, and super easy to get out of. Yes, even in fatal accidents. If the US really cared, drunk and distracted driving would have appropriate sentencing. For god knows what reason, we're still not there yet.
  - bobthepanda 4y ago
    Some judges have taken the opinion that the infrastructure is so car dependent that you can't have a license revoked lest you lose the entire ability to be a self-sufficient member of society. And in parts of the US, that's true. As a general example, how many nightlife areas have decent public transportation at bar closing to avoid committing a DUI?
    - ReptileMan 4y ago
      taxi/cab exist...
      - bobthepanda 4y ago
        How many people can afford to pay a cab for every trip they make?
        ReptileMan 4y ago
        If you can aford to get wasted in a bar, you afford the car ride to take you home
        silisili 4y ago
        I'm obviously not defending drunk driving, but for many people, remember to double that cost. You either have to take an uber both ways, or get an uber home, then back to your car in the morning. I used to just Uber if I thought I'd be drinking at -all-, but hell that costed me about 60 to 70 bucks which definitely isn't realistic for most young people looking to just have a few beers out.
- rswail 4y ago
  The problem is not drunk drivers but incapacitated drivers, whether due to alcohol, drugs, lack of sleep, distractions (mobile phones), etc. The fix is not arbitrary alcohol blood levels, or extending the "War on Drugs", but with standardized, enforced, in-vehicle continuous testing that ensures that the driver is awake, responsive, and in control of the vehicle. That is something where R&D in AI/ML/software to analyze and detect incapacity could/should be directed. It can be entirely on/in-vehicle, with access to the data only available to law enforcement (and, perhaps, insurance companies), if and when it is needed to determine liability (not as a general "carte blanche" for the data), in the same way that aircraft black boxes are only analyzed for investigation after an incident.
  - londons_explore 4y ago
    I have a suspicion that most accidents involve the worst 10% of drivers. I'd like to see research into ways to identify those drivers and encourage them to improve their driving skill. For example, one way might be to make a register of driving skill public information, and then insurance companies can jack their prices up for those people unless they go on and pass (government funded) extra training courses. Such a register could simply be an open database of insurance claims, or police crash records showing the names of people involved.
    - eesmith 4y ago
      Here's a starting point for you - "Estimating likelihood of future crashes for crash-prone drivers" at https://www.sciencedirect.com/science/article/pii/S2095756415000252 https://www.sciencedirect.com/science/article/pii/S209575641... . > At-fault crash-prone drivers are usually considered as the high risk group for possible future incidents or crashes. In Louisiana, 34% of crashes are repeatedly committed by the at-fault crash-prone drivers who represent only 5% of the total licensed drivers in the state. ... From the findings, it is recommended that crash-prone drivers should be targeted for special safety programs regularly through education and regulations.
    - lotsofpulp 4y ago
      >Such a register could simply be an open database of insurance claims, or police crash records showing the names of people involved. In the US, these are CLUE and A-PLUS https://risk.lexisnexis.com/products/clue-auto https://risk.lexisnexis.com/products/clue-auto https://www.verisk.com/insurance/products/a-plus-commercial/ https://www.verisk.com/insurance/products/a-plus-commercial/
- lozenge 4y ago
  What on earth does drunk driving have to do with Tesla's self driving software? You could argue for Waymo, which is actually capable of going without human assistance, not Tesla which is clearly not going to deliver on that.
  - cypress66 4y ago
    That in theory Tesla fsd will be level 4 or 5 in the future.
- bmitc 4y ago
  > it's clear that human behavior and existing road design is still vastly worse Is that true? Those human statistics take every human car and driver into account, not just Teslas. I have never seen data that shows self-driving cars anywhere close to human driving capability. If we replaced every car on the road with "self-driving" Teslas, it would be mayhem.
  - fallingknife 4y ago
    If we replaced every car, it would be an easy problem. The issue is having to replace one at a time while everyone else on the road is still human.
    - spookie 4y ago
      Or... Just make public transportation work? Is this just that difficult to get on the anglosphere or something? I still cannot grasp what people are trying to do with this, really. Why are they trying to do public transports... With cars? You guys do realize most of accidents come from urban/chaotic areas, right? You don't have to use a car to solve this.
      - fallingknife 4y ago
        The percentage of people in the US that live in an area dense enough where even an ideally designed public transport system would match the convenience of a car is in the low single digits. When I go to European cities with these great public transportation systems everybody loves to talk about, what to I see on the road? Lots of cars. Cars aren't going anywhere.
  - KennyBlanken 4y ago
    Most of the "teslas are safer than other cars!" stats are severely biased because: 1)Teslas are "luxury" vehicles, much higher-end than the average US car. A base Model S, or average model 3, costs twice the average US car cost. They are as a result significantly better equipped with safety features and have better crash engineering. In terms of driver assistance, the auto industry was pretty belligerent about backup cameras being mandated, just to give one example. 2)Teslas are on a whole significantly newer than the average US car. The average age of a car on the road in the US is twenty years old. Of course a Tesla with safety features not available until ~10 years ago (or less) is safer than a 1999 anything. If people said "my electric Mercedes is significantly safer than the average car, so Mercedes should be allowed to beta-test their software, which despite after many years of development is proven to randomly slam on the brakes, or drive into the back of emergency vehicles, and is shown on numerous videos to violate traffic laws, suddenly veer at poles and barriers and pedestrians and cyclists" ...they'd be laughed out of the room on a number of points.
eyelidlessness 4y ago
Let’s be even more clear: the regulations are so overly specific that they don’t even provide any guidance let alone recourse for things they don’t cover. I’m less concerned about the slow movement on proven technology, and more concerned that any amount of novel privately automated technology is just free to let loose on society with no restriction.
RamblingCTO 4y ago
Are you referring to the fact that everyday joe's can use this tech? Otherwise testing self-driving is also legal in the EU (of course with a big process attached and people behind the wheel).
weberer 4y ago
How do you think something becomes "proven safe" without testing?
- bobthepanda 4y ago
  These have been tested and used in Europe and Japan for the better part of the past two decades. We don't have to be pioneers at the bleeding edge of testing, but 15+ years? Come on.
- thallium205 4y ago
  Same way Pfizer, Moderna, and J&J did it!
qwerty456127 4y ago
This is because the only place for progress in America seemingly is within yet-nonexistent unregulated areas. As soon as something gets regulated in the US it mostly is doomed to lag decades behind the rest of the developed world (because American policymakers apparently are not fond of updating existing laws/rules and the bureaucracy is not flexible enough, often prioritizing the letter of the law). Of course there are exceptions (let alone particular regulations which do much more good than harm) but this seems to be the rule.
- snovv_crash 4y ago
  Metamizol (Novalgin) is probably one of the biggest examples of this I know. It's a non-NSAID, non-addictive painkiller that was discovered over 100 years ago and banned in the USA in the 60s. It has a similar danger profile to Diclofenac (Voltaren), which is now prescribed like candy. And it is still widely used in Germany and Switzerland, so there is lots of data on 'modern' populations. But since nobody can patent it, there's no motivation to get it reapproved in the US, so everyone get opioids instead.
  - shellfishgene 4y ago
    While it's widely prescribed in Germany, it is actually pretty restricted, too, but there seems to be a lot of off-label use. It's only supposed to be prescribed for strong pain, or for fever only if other medications don't work. There is a risk of agranulocytosis, which is low, but the reason for the restriction in various countries.
    - snovv_crash 4y ago
      Yes, it's not a paracetamol replacement, more like an option as a combination with paracetamol and ibuprofen to treat pain that would otherwise require opioids.
  - TomSwirly 4y ago
    To be fair, the Netherlands has very tight controls on metamizol as well, and for no clear reason except small-c conservatism, and also a tremendous reluctance for doctors here to give out painkillers. No, I have never asked for painkillers myself, so it isn't personal! However, if God–forbid I came down with pain issues that were treatable by painkillers, you better believe I would be on the dark markets PDQ. (My RSI is no fun, but I deal, and it isn't amenable to painkillers except in mind-numbing doses.)
elif 4y ago
At what point does a "life jeopardize" become a "safety feature" for you? Since 160k FSD beta testers have cost 0 lives, and _human_ drivers in my state kill 16 people per 100k. The fact is, the vast majority of lives are jeopardized by reckless, unaware, distracted, or impaired... Things which don't apply to FSD.

We allow companies like Tesla to jeopardize people's lives while the company beta tests "full self driving" technology, but are scared of allowing proven safe technology like adaptive headlights.

bb88 4y ago

You're not wrong, the NHTSA should be more proactive on this stuff. But to be fair, drunk drivers kill 28 people a day in this country, or about 10k people per year. 3k people die at signalled intersections every year, or around 8ish per day. While Tesla shouldn't be beta testing their self driving cars on public roads, it's clear that human behavior and existing road design is still vastly worse.

prng2021 4y ago
Yes drunk driving is a terrible problem but beta testing software on the open road is not the only option available. It's nothing but irresponsible. Here's one of many: use sensors and AI in the car to detect a drunk person behind the wheel. This is HN so I'm sure readers could spend 10 seconds and think of numerous ways this could be accomplished and many otherideas about what could be done when a drunk driver is detected. I also think the NHTSA should have prevented companies like Tesla from behind allowed to put every single control behind a touchscreen. It's extremely dangerous to have to look away for 2 seconds when you're driving. Imagine if (when?) every car in the world adopted this and all drivers are having to look at a screen to turn on windshield wipers.
- bb88 4y ago
  You may find it interesting the NHTSA is pushing for breathalizers in cars starting in 2026. Personally I'm more in favor of smart road like technologies where a highway has a mechanical or physical properties to assist self-driving vehicles stay in line with the road. Mostly because a 100% software solution will likely not work, particularly in the midwest and great lakes states during the wintertime. It need not be expensive either. Magnetic nails embedded in the road would maybe be the cheapest way, and would be easy for both city and rural populations to add, and be detectable with icy and snowy conditions.
  - andrewflnr 4y ago
    Of course they're doing something dystopian instead of something sensible. But yes, I bet even better standardized road markings would be a huge help for self driving cars. I wonder if there's something stupid simple like a paint additive that would make it easy for machines to track. Magnetic beads?
    - waste_monk 4y ago
      Then any asshole could mix up their own magnetic bead paint and cause havoc. Probably no more than someone painting their own road lines or putting up fake signage does now [1], but if you want something that's better than what we currently have it would need to not only be both machine and human readable, but also authenticated. I always thought it would be quite neat to have those portable roadside displays you see for roadworks and so on also communicate map and driving data (cryptographically signed by your local council or dept of transport). So you might get a sign that says "roadworks ahead, please slow down" for human drivers, and the radio beacon would communicate a change of speed limits inside a certain geo-bounded area that would cause self-driving vehicles to slow down. Or similar stuff for "bridge out ahead, turn around now" causing navigation to reroute. [1] https://techcrunch.com/2017/03/17/laying-a-trap-for-self-driving-cars/ https://techcrunch.com/2017/03/17/laying-a-trap-for-self-dri...
      - sgtnoodle 4y ago
        Just have the human readable signs be reasonably standardized, and the self driving car computer can then read them too. It's a lot easier and more future proof to side-channel map updates to cars through a cell network, and allows for some curation. Also, improved cell network coverage seems like a more worthwhile capital expenditure for a local government than a bunch of smart signs that only help 0.001% of drivers. Cynically, I'm sure some politician's spouse's cousin would be happy to win the contract to slap a BLE beacon into a box and bolt it onto a sign for $150k a unit, though... There is an ongoing effort to introduce infrastructure free direct communication for vehicles, using 5.9ghz spectrum. The spectrum was previously allocated to DSRC radio technology the automotive industry pushed for, but it's been 15 years and it hasn't gone anywhere. The FCC is now letting the cell network folk take a crack at solving the problem instead.
- m463 4y ago
  Have you driven a tesla? I think they help with the driving. That said, touchscreen controls (with tiny targets) instead of dedicated controls and the new square steering wheel with touch controls instead of stalks... gah!
- wheybags 4y ago
  No need for fancy sensors and ai - I used an electric scooter rental in Germany recently, and it made me do a reaction speed test on my phone before starting. Super easy.
  - JamesBarney 4y ago
    Are they worried you'll kill someone with your electric scooter?
    - exclusiv 4y ago
      It happens as it causes swerves and crashes. Deer do this too. With a deer your best course of action is to try and stop but mow it over if necessary. With a person swerving on a bike or moped, you may not be able to stop in time and most likely swerve.
    - willsmith72 4y ago
      They're worried you'll kill yourself
      - JamesBarney 4y ago
        If people can get on an escorted i wonder if they wall or uber. If they walk i could totally see the total number of fatalities being hire than escooting. Similar to how drunken walking has a higher fatality raid than drunken driving/biking.
    - Broken_Hippo 4y ago
      Yes, especially yourself if you aren't wearing a helmet. More commonly, though, the electric scooters cause injuries and accidents. Starting next year, Norway is requiring insurance for them and rental prices will surely increase in response.
    - twobitshifter 4y ago
      It is inexplicable how e scooters are treated like death machines in comparison to 4000 pound SUVs which get no restrictions. Scooters have regulated speed, zones that are banned from use, reaction tests, sidewalk restrictions and on and on. You would think there were thousands of e scooter deaths per year.
- JamesBarney 4y ago
  If it gets us to autonomy 6 months earlier that could easily save 50x as many lives as deaths it's caused.
Rebelgecko 4y ago
Didn't the infrastructure bill have a clause saying that all cars have to start using alcohol detecting interlocks in a few years?
m463 4y ago
They say in this article: the National Transportation Safety Board conducted an investigation finding that the government should not be actively restricting headlamp systems that have been shown to have a safety benefit. So I can imagine this sort of thinking coming down on telsa's side of the argument. Personally, although tesla does "market" their technology, I think it is good technology that has probably not only saved a fair number of lives, but also bumpers and fenders. Looking at the storied history of the airbag, I think telsa did a good job assisting the driver (and customer) when the status quo for other car manufacturers was to innovate at a snail's pace and implement safety features mostly when required.
silisili 4y ago
Drunk driving is a legal issue. The entire problem is that it's not treated seriously, and super easy to get out of. Yes, even in fatal accidents. If the US really cared, drunk and distracted driving would have appropriate sentencing. For god knows what reason, we're still not there yet.
- bobthepanda 4y ago
  Some judges have taken the opinion that the infrastructure is so car dependent that you can't have a license revoked lest you lose the entire ability to be a self-sufficient member of society. And in parts of the US, that's true. As a general example, how many nightlife areas have decent public transportation at bar closing to avoid committing a DUI?
  - ReptileMan 4y ago
    taxi/cab exist...
    - bobthepanda 4y ago
      How many people can afford to pay a cab for every trip they make?
      - ReptileMan 4y ago
        If you can aford to get wasted in a bar, you afford the car ride to take you home
        silisili 4y ago
        I'm obviously not defending drunk driving, but for many people, remember to double that cost. You either have to take an uber both ways, or get an uber home, then back to your car in the morning. I used to just Uber if I thought I'd be drinking at -all-, but hell that costed me about 60 to 70 bucks which definitely isn't realistic for most young people looking to just have a few beers out.
rswail 4y ago
The problem is not drunk drivers but incapacitated drivers, whether due to alcohol, drugs, lack of sleep, distractions (mobile phones), etc. The fix is not arbitrary alcohol blood levels, or extending the "War on Drugs", but with standardized, enforced, in-vehicle continuous testing that ensures that the driver is awake, responsive, and in control of the vehicle. That is something where R&D in AI/ML/software to analyze and detect incapacity could/should be directed. It can be entirely on/in-vehicle, with access to the data only available to law enforcement (and, perhaps, insurance companies), if and when it is needed to determine liability (not as a general "carte blanche" for the data), in the same way that aircraft black boxes are only analyzed for investigation after an incident.
- londons_explore 4y ago
  I have a suspicion that most accidents involve the worst 10% of drivers. I'd like to see research into ways to identify those drivers and encourage them to improve their driving skill. For example, one way might be to make a register of driving skill public information, and then insurance companies can jack their prices up for those people unless they go on and pass (government funded) extra training courses. Such a register could simply be an open database of insurance claims, or police crash records showing the names of people involved.
  - eesmith 4y ago
    Here's a starting point for you - "Estimating likelihood of future crashes for crash-prone drivers" at https://www.sciencedirect.com/science/article/pii/S2095756415000252 https://www.sciencedirect.com/science/article/pii/S209575641... . > At-fault crash-prone drivers are usually considered as the high risk group for possible future incidents or crashes. In Louisiana, 34% of crashes are repeatedly committed by the at-fault crash-prone drivers who represent only 5% of the total licensed drivers in the state. ... From the findings, it is recommended that crash-prone drivers should be targeted for special safety programs regularly through education and regulations.
  - lotsofpulp 4y ago
    >Such a register could simply be an open database of insurance claims, or police crash records showing the names of people involved. In the US, these are CLUE and A-PLUS https://risk.lexisnexis.com/products/clue-auto https://risk.lexisnexis.com/products/clue-auto https://www.verisk.com/insurance/products/a-plus-commercial/ https://www.verisk.com/insurance/products/a-plus-commercial/
lozenge 4y ago
What on earth does drunk driving have to do with Tesla's self driving software? You could argue for Waymo, which is actually capable of going without human assistance, not Tesla which is clearly not going to deliver on that.
- cypress66 4y ago
  That in theory Tesla fsd will be level 4 or 5 in the future.
bmitc 4y ago
> it's clear that human behavior and existing road design is still vastly worse Is that true? Those human statistics take every human car and driver into account, not just Teslas. I have never seen data that shows self-driving cars anywhere close to human driving capability. If we replaced every car on the road with "self-driving" Teslas, it would be mayhem.
- fallingknife 4y ago
  If we replaced every car, it would be an easy problem. The issue is having to replace one at a time while everyone else on the road is still human.
  - spookie 4y ago
    Or... Just make public transportation work? Is this just that difficult to get on the anglosphere or something? I still cannot grasp what people are trying to do with this, really. Why are they trying to do public transports... With cars? You guys do realize most of accidents come from urban/chaotic areas, right? You don't have to use a car to solve this.
    - fallingknife 4y ago
      The percentage of people in the US that live in an area dense enough where even an ideally designed public transport system would match the convenience of a car is in the low single digits. When I go to European cities with these great public transportation systems everybody loves to talk about, what to I see on the road? Lots of cars. Cars aren't going anywhere.
- KennyBlanken 4y ago
  Most of the "teslas are safer than other cars!" stats are severely biased because: 1)Teslas are "luxury" vehicles, much higher-end than the average US car. A base Model S, or average model 3, costs twice the average US car cost. They are as a result significantly better equipped with safety features and have better crash engineering. In terms of driver assistance, the auto industry was pretty belligerent about backup cameras being mandated, just to give one example. 2)Teslas are on a whole significantly newer than the average US car. The average age of a car on the road in the US is twenty years old. Of course a Tesla with safety features not available until ~10 years ago (or less) is safer than a 1999 anything. If people said "my electric Mercedes is significantly safer than the average car, so Mercedes should be allowed to beta-test their software, which despite after many years of development is proven to randomly slam on the brakes, or drive into the back of emergency vehicles, and is shown on numerous videos to violate traffic laws, suddenly veer at poles and barriers and pedestrians and cyclists" ...they'd be laughed out of the room on a number of points.

eyelidlessness 4y ago

Let’s be even more clear: the regulations are so overly specific that they don’t even provide any guidance let alone recourse for things they don’t cover. I’m less concerned about the slow movement on proven technology, and more concerned that any amount of novel privately automated technology is just free to let loose on society with no restriction.

RamblingCTO 4y ago

Are you referring to the fact that everyday joe's can use this tech? Otherwise testing self-driving is also legal in the EU (of course with a big process attached and people behind the wheel).

weberer 4y ago

How do you think something becomes "proven safe" without testing?

bobthepanda 4y ago
These have been tested and used in Europe and Japan for the better part of the past two decades. We don't have to be pioneers at the bleeding edge of testing, but 15+ years? Come on.
thallium205 4y ago
Same way Pfizer, Moderna, and J&J did it!

qwerty456127 4y ago

This is because the only place for progress in America seemingly is within yet-nonexistent unregulated areas. As soon as something gets regulated in the US it mostly is doomed to lag decades behind the rest of the developed world (because American policymakers apparently are not fond of updating existing laws/rules and the bureaucracy is not flexible enough, often prioritizing the letter of the law). Of course there are exceptions (let alone particular regulations which do much more good than harm) but this seems to be the rule.

snovv_crash 4y ago
Metamizol (Novalgin) is probably one of the biggest examples of this I know. It's a non-NSAID, non-addictive painkiller that was discovered over 100 years ago and banned in the USA in the 60s. It has a similar danger profile to Diclofenac (Voltaren), which is now prescribed like candy. And it is still widely used in Germany and Switzerland, so there is lots of data on 'modern' populations. But since nobody can patent it, there's no motivation to get it reapproved in the US, so everyone get opioids instead.
- shellfishgene 4y ago
  While it's widely prescribed in Germany, it is actually pretty restricted, too, but there seems to be a lot of off-label use. It's only supposed to be prescribed for strong pain, or for fever only if other medications don't work. There is a risk of agranulocytosis, which is low, but the reason for the restriction in various countries.
  - snovv_crash 4y ago
    Yes, it's not a paracetamol replacement, more like an option as a combination with paracetamol and ibuprofen to treat pain that would otherwise require opioids.
- TomSwirly 4y ago
  To be fair, the Netherlands has very tight controls on metamizol as well, and for no clear reason except small-c conservatism, and also a tremendous reluctance for doctors here to give out painkillers. No, I have never asked for painkillers myself, so it isn't personal! However, if God–forbid I came down with pain issues that were treatable by painkillers, you better believe I would be on the dark markets PDQ. (My RSI is no fun, but I deal, and it isn't amenable to painkillers except in mind-numbing doses.)

elif 4y ago

At what point does a "life jeopardize" become a "safety feature" for you? Since 160k FSD beta testers have cost 0 lives, and _human_ drivers in my state kill 16 people per 100k. The fact is, the vast majority of lives are jeopardized by reckless, unaware, distracted, or impaired... Things which don't apply to FSD.