First, when a human screws up, it's obvious who takes the blame for the mistake. If an autonomous vehicle does, who pays the fine/goes the prison? Putting the blame on the owner feels like it takes away a lot of the incentives to buy a autonomous vehicle, but I suspect the companies making these things don't want to be the ones paying out for their mistakes, and will probably lobby to prevent it if possible.
Secondly, this isn't just a vehicle thing, it's a technology thing in general. For better or worse, the expectations for safety and security are much higher for new devices than they are for existing ones, or for manual human driven solutions. Some of this is definitely driven or inspired by vested interests (like how pollution and safety regulations for renewable energy sources and nuclear plants are higher than for coal/gas power plants, despite the latter being worse in almost every way), but a lot of it is just humans being deeply skeptical of new tech and major social changes, even if the benefits of the new system clearly outweigh those of the existing one.
We don't usually fine people or send them to prison for mistakes, especially when driving cars. Insurance companies figure this out between each other.
Depends how serious the consequences are. A normal scrape which doesn't damage either car much or cause any real injury? That's something for insurance companies to figure out. A major accident that kills or seriously injures someone (especially due to drunk or distracted driving)? Someone's probably getting a hefty fine and points on their license at minimal, more likely prison.
If a self-driving car gets into an accident that kills a pedestrian or other driver, someone's going to have to take responsibility for that.
> due to drunk or distracted driving)? Someone's probably getting a hefty fine and points on their license at minimal, more likely prison.
In the US, killing someone with your car while being drunk and driving twice the speed limit gets you 10 days in jail: "10 Days for Bobby Cann’s Killer, a Repeat Drunk Driver, Is an Insult" (https://chi.streetsblog.org/2017/01/27/10-day-sentence-for-b...)
Speeding and DUIs aren't mistakes, they're choices. There's a huge difference between the errors self-driving cars make and the equivalent errors humans make: the self-driving cars simply don't make those mistakes. In the category of things that self-driving cars do, insurance overwhelming handles those mistakes in both human and self-driving cases.
Each human is different so we treat them as separate entities. Machines, by design are cookie-cutter copies. If one Waymo did something horribly stupid, then all Waymos would have done that exact same thing in that situation.
Exactly. And that’s why automation tech should be better than humans before widespread adoption.
My experience from scaling products in very different areas (consumer tech, large public infrastructure, internet): That occasional hiccup during small-scale trials looks very different with hundreds or millions of units.
Isn't that a point in favor of the automated cars, that we can then fix/prevent that behavior for every single car from that point forward, whereas with humans, we can't and don't patch every human driver whenever a human does something horribly stupid - and a massive number of human drivers do horribly stupid things every single day, right?
The 1 individual human might learn from their horrible mistake, but not the other x billion humans (nor the many people every minute who drive for the first time).
And even the 1 human who made the mistake might forget it after some years, or make the same mistake again as they become elderly, etc.
The full title is "Why We Demand Perfect Machines Yet Tolerate Human Carnage"
The "Why" makes a difference.
To me, the key is here.
Empathy is a cognitive tool we use to forgive human error. But a machine’s error is different, like it’s the product of a bad corporate decision or poor coding. Often, there isn’t one person to blame or forgive; we are stuck with an opaque version of fate that is hard to accept. Forgiveness needs an agent that perhaps could have done something differently and can be held responsible. When a person makes a serious mistake, we have somewhere to place the blame and, eventually, someone to absolve.
People tend to lose confidence in a machine more quickly than in a human when the two make identical mistakes. We prefer a worse human over a better algorithm if the latter has made an error. Behavioral scientists call it algorithm aversion. With no trusted party to explain what happened, visible mistakes dominate people’s judgments of a better system.
But, by reading the word Carnage, I expected the article to say that the common demand that OTHERS be perfect (projecting one's imperfection onto others, IMO), is a cause of wars and carnage.
"Whataboutism" as an excuse to attack.
I get the feeling that the point is to make machines more like humans in the sense that we empathize with humans, whereas machines are opaque.
And I expect that if that's fone, it will be a sleazy con job.
Having done some programming, I tend to look past the opacity to the designers and coders. And direct my "comments" and feeling of powerlessness [0] to those anonymous people.
While open source can give me a real human to gripe at, I would rather use the open source mechanism to make suggestions, and feel empowered.
Does that put the "humanity" into machines for you? I presume that HAL was closed-source.
[0]Yes, I feel powerless dealing with the home computer when things go off the rails, and while I could write scripts, apps, and flying hot patches, the vendor changes things constantly. And nags me daily, on every damn device when I don't go along.
To a lesser extent this happens with open source, but at least I can see what changed, or as noted above, be part of the whole process.
If my suggestion or code isn't accepted, I can be told why, up to and including "Well, that's because you're an idiot", which has happened in the past, so I can discount the comment, or just try changes on my own.
The chances of any of this actually happening are minuscule, but, getting back on point, it's possible, unlike those black boxes created by anonymous people.
We should stop Tolerating Human Carnage. But the solution is not self driving cars.
The actual way to solve the 'Carnage' is to adopt well proven well studied infrastructure changes. After 40+ years of idiotic car brain policies, some nations have started to push back and actually started to treat road safety more like aviation safety. Constant review and feedback into the profession and infrastructure updates. Vision Zero is approachable even without self driving cars. But even that is looking at the problem in the wrong way.
True improvement comes from modal shift to public transport, along with increasing use in walking and biking. Those mods are saver and healthier by any possible measure, no matter how amazing these self driving machines will be in the future.
This is not only good for less people dying in traffic, its also good for many, many, many other reasons. General population fitness, general health and so on. The list of effects of good urbanism and anti car policy is basically endless.
Cars are the problem, and self driving cars are a solution to part of that problem, and electric cars are a solution to part of that problem. But cars are still overall the actual problem biggest problem. Instead of investing countless billions in these two sub problems, for far, far less money you could do a huge amount to change the actual structure of how transportation works.
Not Just Bikes has a good video on the problem with self driving cars, even if the technology worked as advertised. Its not how we want the future to look like.
Putting each human into an individual multi-ton pod is stupid, unnecessary and dangerous even if you plaster unlimited amounts of technology on each pod.
We know how to create massively better societies in every way, its not dispute in people who research these things. Self driving cars can be a tiny part of that, but mostly in rural and suburban regions. In cities they are not needed.
This hyper focus on human car crashes is to narrow and misses a much bigger discussion, but this is a discussion the car industry and the individualistically driven AI people don't want to have.
It's a good goal to have, and to continue working towards. At the same time, we shouldn't let that goal cause massive numbers of unnecessary deaths because we're avoiding the thing that can save lives along the way.
1.35 million people die per year due to road traffic.
You mentioned that billions are being invested in solving self-driving and electric cars, but that "for far, far less money you could do a huge amount to change the actual structure of how transportation works" - is that true/source?
The cost of a single subway line can cost multiple billions of dollars per mile, a high-speed rail line in a single state can cost hundreds of billions of dollars, etc. - and there are a lot of miles in the US, a lot of states, etc.
And the other big difference is about who pays for it - even if billions of dollars are being invested in solving self-driving and electric cars, it's a lot of private money doing that, whereas building public transportation is mostly taxpayer funds.
Again, they probably are good things to continue investing in, working towards, etc. - but I don't think we should delay solving even a tiny percent of the 1.35 million deaths per year in order to wait for public transportation to be solved. (And per the OP article, it's more like 94% reduction already with the existing self-driving technology, and it keeps getting better.)
> At the same time, we shouldn't let that goal cause massive numbers of unnecessary deaths because we're avoiding the thing that can save lives along the way.
The argument here is that political capital and time should be invest to change people mind about self driving cars.
My argument is that political capital and time should be invested in changing people mind about the fundamental approach to land use and transportation.
> is that true/source?
There is a whole field of study on every aspect of this. Basically the medical community, the city planning community, the ecological community and the transport planners (outside of the US) all generally reach the same conclusions.
Bike lands, road diets, and bus lanes, bus priority are very cheap. Basically free. They are mostly political reallocation's of space. The politics are the hardest part, not the cost.
Rezoning is essentially free, or actually produces huge amounts of value. Its very, very well document what the problem with US land use patterns and transportation systems are.
See Strong Towns or Urban3 analysis for example.
Many of the choice that cause a lot of death, like stroads, are also economically idiotic.
> The cost of a single subway line can cost multiple billions of dollars per mile, a high-speed rail line in a single state can cost hundreds of billions of dollars, etc. - and there are a lot of miles in the US, a lot of states, etc.
The US operates on such a low level, there is so much you can do before even getting to those things. Like we are talking minimal competence of running bus lanes in a modern way, or having basic signaling priority.
The cost large infrastructure is a bigger topic, but that's not mostly what I am talking about. Those things are such a insignificant part of US transportation currently it barley worth talking about.
> And the other big difference is about who pays for it - even if billions of dollars are being invested in solving self-driving and electric cars, it's a lot of private money doing that, whereas building public transportation is mostly taxpayer funds.
You are ignoring the larger structure, the road network and the electricity network are deeply political and public. And even beyond that, electric cars have seen political support. But of course the whole reason why so much public investment flows into these things is because the infrastructure is design in a way where there is a gigantic market for cars that you can sell into. The market is downstream of higher level choice in terms of infrastructure and land use. The under taxing and over support of parking alone accounts for many, many billions in subsidies.
But on some level, in the end it comes down to what your society ends up spending totally on transportation and how efficient that is. The investors investing in these things need to make their money back. And the US is generally not very efficient, its just rich enough to care less. If you look at total 'consumption' that looks ok, but then you realize its just a expensive pickup truck that you end up sitting in traffic jams doesn't really improve your live that much.
If you look at how effectively poor people generally get around, the results are not that great. Car dependency cause tons of problems and cost on society and is worst for the poorest.
So I guess I just care less about who provides capital compared to what the outcomes are. The idea that public capital investment is bad and private investment is a universal good that can never be questioned is just not how I think about it.
> in order to wait for public transportation to be solved.
My argument is much wider then just 'public transport'. Its about how all land use and transportation infrastructure is organized.
> And per the OP article, it's more like 94% reduction already with the existing self-driving technology, and it keeps getting better.
Some of that can be questioned but its not my focus. Even if it was 100% reduction, its to narrow of an argument.
I think it ought to be said here in the comments that magazine's front page is anathema to the ideals of the comment section of HN, and whoever is funding these blowhards clearly doesn't share our ideals. Proceed with caution.
I think there are a few things to consider here.
First, when a human screws up, it's obvious who takes the blame for the mistake. If an autonomous vehicle does, who pays the fine/goes the prison? Putting the blame on the owner feels like it takes away a lot of the incentives to buy a autonomous vehicle, but I suspect the companies making these things don't want to be the ones paying out for their mistakes, and will probably lobby to prevent it if possible.
Secondly, this isn't just a vehicle thing, it's a technology thing in general. For better or worse, the expectations for safety and security are much higher for new devices than they are for existing ones, or for manual human driven solutions. Some of this is definitely driven or inspired by vested interests (like how pollution and safety regulations for renewable energy sources and nuclear plants are higher than for coal/gas power plants, despite the latter being worse in almost every way), but a lot of it is just humans being deeply skeptical of new tech and major social changes, even if the benefits of the new system clearly outweigh those of the existing one.
We don't usually fine people or send them to prison for mistakes, especially when driving cars. Insurance companies figure this out between each other.
Depends how serious the consequences are. A normal scrape which doesn't damage either car much or cause any real injury? That's something for insurance companies to figure out. A major accident that kills or seriously injures someone (especially due to drunk or distracted driving)? Someone's probably getting a hefty fine and points on their license at minimal, more likely prison.
If a self-driving car gets into an accident that kills a pedestrian or other driver, someone's going to have to take responsibility for that.
> due to drunk or distracted driving)? Someone's probably getting a hefty fine and points on their license at minimal, more likely prison.
In the US, killing someone with your car while being drunk and driving twice the speed limit gets you 10 days in jail: "10 Days for Bobby Cann’s Killer, a Repeat Drunk Driver, Is an Insult" (https://chi.streetsblog.org/2017/01/27/10-day-sentence-for-b...)
Is this sarcasm?
Bad drivers absolutely get fines and jail time for speeding, DUIs, etc.
Speeding and DUIs aren't mistakes, they're choices. There's a huge difference between the errors self-driving cars make and the equivalent errors humans make: the self-driving cars simply don't make those mistakes. In the category of things that self-driving cars do, insurance overwhelming handles those mistakes in both human and self-driving cases.
Each human is different so we treat them as separate entities. Machines, by design are cookie-cutter copies. If one Waymo did something horribly stupid, then all Waymos would have done that exact same thing in that situation.
Exactly. And that’s why automation tech should be better than humans before widespread adoption.
My experience from scaling products in very different areas (consumer tech, large public infrastructure, internet): That occasional hiccup during small-scale trials looks very different with hundreds or millions of units.
> automation tech should be better than humans before widespread adoption
How much better?
From the article: "Waymo’s vehicles have been involved in 94% fewer crashes that cause serious injury or worse than human drivers on the same roads."
The longer safe autonomous vehicles are obstructed through legislation, the more people die.
Isn't that a point in favor of the automated cars, that we can then fix/prevent that behavior for every single car from that point forward, whereas with humans, we can't and don't patch every human driver whenever a human does something horribly stupid - and a massive number of human drivers do horribly stupid things every single day, right?
Humans learn from mistakes and feel the repercussions. We understand this starting from when we’re infants.
The 1 individual human might learn from their horrible mistake, but not the other x billion humans (nor the many people every minute who drive for the first time).
And even the 1 human who made the mistake might forget it after some years, or make the same mistake again as they become elderly, etc.
The full title is "Why We Demand Perfect Machines Yet Tolerate Human Carnage"
The "Why" makes a difference.
To me, the key is here.
But, by reading the word Carnage, I expected the article to say that the common demand that OTHERS be perfect (projecting one's imperfection onto others, IMO), is a cause of wars and carnage."Whataboutism" as an excuse to attack.
I get the feeling that the point is to make machines more like humans in the sense that we empathize with humans, whereas machines are opaque.
And I expect that if that's fone, it will be a sleazy con job.
Having done some programming, I tend to look past the opacity to the designers and coders. And direct my "comments" and feeling of powerlessness [0] to those anonymous people.
While open source can give me a real human to gripe at, I would rather use the open source mechanism to make suggestions, and feel empowered.
Does that put the "humanity" into machines for you? I presume that HAL was closed-source.
[0]Yes, I feel powerless dealing with the home computer when things go off the rails, and while I could write scripts, apps, and flying hot patches, the vendor changes things constantly. And nags me daily, on every damn device when I don't go along.
To a lesser extent this happens with open source, but at least I can see what changed, or as noted above, be part of the whole process.
If my suggestion or code isn't accepted, I can be told why, up to and including "Well, that's because you're an idiot", which has happened in the past, so I can discount the comment, or just try changes on my own.
The chances of any of this actually happening are minuscule, but, getting back on point, it's possible, unlike those black boxes created by anonymous people.
We should stop Tolerating Human Carnage. But the solution is not self driving cars.
The actual way to solve the 'Carnage' is to adopt well proven well studied infrastructure changes. After 40+ years of idiotic car brain policies, some nations have started to push back and actually started to treat road safety more like aviation safety. Constant review and feedback into the profession and infrastructure updates. Vision Zero is approachable even without self driving cars. But even that is looking at the problem in the wrong way.
True improvement comes from modal shift to public transport, along with increasing use in walking and biking. Those mods are saver and healthier by any possible measure, no matter how amazing these self driving machines will be in the future.
This is not only good for less people dying in traffic, its also good for many, many, many other reasons. General population fitness, general health and so on. The list of effects of good urbanism and anti car policy is basically endless.
Cars are the problem, and self driving cars are a solution to part of that problem, and electric cars are a solution to part of that problem. But cars are still overall the actual problem biggest problem. Instead of investing countless billions in these two sub problems, for far, far less money you could do a huge amount to change the actual structure of how transportation works.
Not Just Bikes has a good video on the problem with self driving cars, even if the technology worked as advertised. Its not how we want the future to look like.
https://www.youtube.com/watch?v=040ejWnFkj0
Putting each human into an individual multi-ton pod is stupid, unnecessary and dangerous even if you plaster unlimited amounts of technology on each pod.
We know how to create massively better societies in every way, its not dispute in people who research these things. Self driving cars can be a tiny part of that, but mostly in rural and suburban regions. In cities they are not needed.
This hyper focus on human car crashes is to narrow and misses a much bigger discussion, but this is a discussion the car industry and the individualistically driven AI people don't want to have.
It's a good goal to have, and to continue working towards. At the same time, we shouldn't let that goal cause massive numbers of unnecessary deaths because we're avoiding the thing that can save lives along the way.
1.35 million people die per year due to road traffic.
You mentioned that billions are being invested in solving self-driving and electric cars, but that "for far, far less money you could do a huge amount to change the actual structure of how transportation works" - is that true/source?
The cost of a single subway line can cost multiple billions of dollars per mile, a high-speed rail line in a single state can cost hundreds of billions of dollars, etc. - and there are a lot of miles in the US, a lot of states, etc.
And the other big difference is about who pays for it - even if billions of dollars are being invested in solving self-driving and electric cars, it's a lot of private money doing that, whereas building public transportation is mostly taxpayer funds.
Again, they probably are good things to continue investing in, working towards, etc. - but I don't think we should delay solving even a tiny percent of the 1.35 million deaths per year in order to wait for public transportation to be solved. (And per the OP article, it's more like 94% reduction already with the existing self-driving technology, and it keeps getting better.)
> At the same time, we shouldn't let that goal cause massive numbers of unnecessary deaths because we're avoiding the thing that can save lives along the way.
The argument here is that political capital and time should be invest to change people mind about self driving cars.
My argument is that political capital and time should be invested in changing people mind about the fundamental approach to land use and transportation.
> is that true/source?
There is a whole field of study on every aspect of this. Basically the medical community, the city planning community, the ecological community and the transport planners (outside of the US) all generally reach the same conclusions.
Bike lands, road diets, and bus lanes, bus priority are very cheap. Basically free. They are mostly political reallocation's of space. The politics are the hardest part, not the cost.
Rezoning is essentially free, or actually produces huge amounts of value. Its very, very well document what the problem with US land use patterns and transportation systems are.
See Strong Towns or Urban3 analysis for example.
Many of the choice that cause a lot of death, like stroads, are also economically idiotic.
> The cost of a single subway line can cost multiple billions of dollars per mile, a high-speed rail line in a single state can cost hundreds of billions of dollars, etc. - and there are a lot of miles in the US, a lot of states, etc.
The US operates on such a low level, there is so much you can do before even getting to those things. Like we are talking minimal competence of running bus lanes in a modern way, or having basic signaling priority.
The cost large infrastructure is a bigger topic, but that's not mostly what I am talking about. Those things are such a insignificant part of US transportation currently it barley worth talking about.
> And the other big difference is about who pays for it - even if billions of dollars are being invested in solving self-driving and electric cars, it's a lot of private money doing that, whereas building public transportation is mostly taxpayer funds.
You are ignoring the larger structure, the road network and the electricity network are deeply political and public. And even beyond that, electric cars have seen political support. But of course the whole reason why so much public investment flows into these things is because the infrastructure is design in a way where there is a gigantic market for cars that you can sell into. The market is downstream of higher level choice in terms of infrastructure and land use. The under taxing and over support of parking alone accounts for many, many billions in subsidies.
But on some level, in the end it comes down to what your society ends up spending totally on transportation and how efficient that is. The investors investing in these things need to make their money back. And the US is generally not very efficient, its just rich enough to care less. If you look at total 'consumption' that looks ok, but then you realize its just a expensive pickup truck that you end up sitting in traffic jams doesn't really improve your live that much.
If you look at how effectively poor people generally get around, the results are not that great. Car dependency cause tons of problems and cost on society and is worst for the poorest.
So I guess I just care less about who provides capital compared to what the outcomes are. The idea that public capital investment is bad and private investment is a universal good that can never be questioned is just not how I think about it.
> in order to wait for public transportation to be solved.
My argument is much wider then just 'public transport'. Its about how all land use and transportation infrastructure is organized.
> And per the OP article, it's more like 94% reduction already with the existing self-driving technology, and it keeps getting better.
Some of that can be questioned but its not my focus. Even if it was 100% reduction, its to narrow of an argument.
[dead]
[dead]
I think it ought to be said here in the comments that magazine's front page is anathema to the ideals of the comment section of HN, and whoever is funding these blowhards clearly doesn't share our ideals. Proceed with caution.
Is there a particular thing you'd quote as an issue?