Slashdot Mirror
Should The Media Cover Tesla Accidents? (chicagotribune.com)
Elon Musk tweeted about the accident:
It's super messed up that a Tesla crash resulting in a broken ankle is front page news and the ~40,000 people who died in US auto accidents alone in past year get almost no coverage. What's actually amazing about this accident is that a Model S hit a fire truck at 60mph and the driver only broke an ankle. An impact at that speed usually results in severe injury or death.
The Associated Press defended their news coverage Friday, arguing that the facts show that "not all Tesla crashes end the same way." They also fact-check Elon Musk's claim that "probability of fatality is much lower in a Tesla," reporting that it's impossible to verify since Tesla won't release the number of miles driven by their cars or the number of fatalities. "There have been at least three already this year and a check of 2016 NHTSA fatal crash data -- the most recent year available -- shows five deaths in Tesla vehicles."
Slashdot reader Reygle argues the real issue is with the drivers in the Autopilot cars. "Someone unwilling to pay attention to the road shouldn't be allowed anywhere near that road ever again."
Liberal: "willing to respect or accept behaviour or opinions different from one's own; open to new ideas."
Socialism: "any of various economic and political theories advocating collective or governmental ownership and administration of the means of production and distribution of goods"
Pick what you really mean.
If builders built buildings the way programmers wrote programs, then the first woodpecker would destroy civilization.
Tesla has not been calling AP "self-driving" - you're thinking of Mercedes, describing their awful Autopilot-wannabe. Tesla goes through huge lengths to point out that it's not self driving - to the point that they sell two separate packages, "Enhanced Autopilot" and "Full Self Driving", and the latter tells you that it is not available yet - so that it's physically impossible to think that your car is "full self driving", because either you didn't choose that option, or you didn't receive it.
Contrary to popular myth, AP accidents are rarely from "newbies who mistakenly thought their car was self driving". They're overwhelmingly from experienced users who've had AP for a long time. They get overconfindent in their car's abilities and stop paying attention, just doing things like using their cell phone and only stopping to occasionally grab the wheel so the car won't harass them. Newbies are generally paranoid and hypervigilant.
Musk has not clarified exactly what it was about eye tracking that he thought was not ready for prime time, but I hope it gets remedied and implemented (Model 3 already has the requisite driver-facing camera). If the driver's attention can be ensured, I think it's pretty indisputable that "vehicular sensors and constant attention" plus "human senses and reasoning" is going to be by far the safest option. But you need to ensure that the human is actually paying attention to the road. Requiring torque on the steering wheel is good (better than just a pressure sensor), but not enough.
IMHO.
Give a boy a gun and you arm him for a day. Teach him how to make a gun, and the whole metaphor breaks down.
You want to promote your darling as the next evolutionary step that will replace all existing automobiles. You're especially proud of your " autonomous driving " feature. This accident is news because Tesla supposedly rolled out a safety update that enabled automatic emergency braking but appears to have been limited to vehicles operating under 50mph. ( Whoops, guess we should have upped that a bit )
See, when you promote your vehicle with said safety features and it still ends up crashing just like the " dumb " cars out there, it doesn't shine a positive light on your over-hyped / over-priced* product. ( *Compared to the typical ICE vehicle )
Moral of the story: When in the spotlight, you don't get to pick and choose what people see.
No, the car did not run out of juice (ever), and it was a blown fuse, which only disabled part of the braking power. Indeed, the BBC has admitted that the "running out of charge" event was staged.
The court rulings related to Teslas lawsuit never disagreed with Tesla's claim that Top Gear staged the events. They ruled that Tesla had failed to show material losses, and that a reasonable person would not believe that what happens on Top Gear is not embellished (something I think is false, but that's what they ruled).
Beyond the fakery, then there's the deliberate distortions. Like going on about how the car only went a fraction of its rated range. Ignoring that it still had a 20% charge left when they decided to fake the "dead battery" and push it off the track, what they did would apply to any car. You think you can take your car to the track and drive foot down nonstop and get the same mileage as you get in a steady cruise? But they were trying to give the impression that you only go a short distance in the vehicle, which was simply not in any way, shape or form true.
Give a boy a gun and you arm him for a day. Teach him how to make a gun, and the whole metaphor breaks down.
I think it's pretty indisputable that "vehicular sensors and constant attention" plus "human senses and reasoning" is going to be by far the safest option.
I would dispute it. If humans think that driving control can be left to automation, then most of them are not going to pay much attention. So if the safety does depend partly on the driver it is likely to fall between two stools. Automated driving needs to be all or nothing.
That is Elon's point as well. Disproportionately covering Tesla crashes keeps the general public ignorant of Tesla's actual safety record.
Tesla vehicles currently drive about 2.67 billion miles per year (as of July 2017). That is about 0.083% of all miles driven in the US. This means you would expect about 33-34 deaths per year in Tesla cars, without adjusting for the demographics and behaviors of Tesla drivers compared to the US average. In 2016 there were 5 deaths, making Tesla cars about 6-7 times safer than average (again without adjusting for driver demographics).
Considering the average driver age (by miles driven) is about 43, and Tesla average driver age is 53, you would expect slightly less driver deaths in Teslas. Although considering driver deaths by age are fairly consistent between ages 35-70 (with 16-30 and 75+ being much higher), perhaps there would be another 1-2 Tesla driver deaths if they had drivers of a more average age. So after taking that into account, maybe Teslas are only 4-5 times safer than an average car. Then again your average Tesla driver probably likes to accelerate faster than your average 53 year old driver, so maybe on average Tesla drivers are just as safe or even less safe than average drivers.
Teslas are also much newer than the average car. The average car on the road is about 11 years old, and my guess is the average Tesla is around 3-4 years old. 3 year old cars are about 20% safer than 11 year old cars on average, so again Tesla might kill another 1-2 people per year if they were a bit older. So perhaps Teslas are only 3-4 times safer than your average car.
The only thing that isn't disputable is that Teslas are far safer than your average car. If every car was as safe as a Tesla, its possible that 75% or more of all traffic deaths would be prevented. It's hard to tell just how many 10's of thousands of lives would be saved each year if all cars were as safe as Teslas because there are so many factors, but it would certainly be a lot.
-- All that is necessary for the triumph of evil is that good men do nothing. -- Edmund Burke
Useless statistic. Yes, that's ostensibly lower than the average (one per 13,000 cars) for all registered vehicles in the U.S., but without knowing how many miles the Teslas were driven, we can't know if that's actually low or high.
Typically, the number of accidents (and, thus, fatal accidents) is proportional to the number of miles driven, not the number of cars. Some cars sit in somebody's front yard rusting, and never even see the road except when another car isn't working. And people who are wealthy enough to afford Teslas are more likely to live close to work, and thus have shorter commutes, so they are exposed to fewer opportunities for wrecks. They're also less likely to be driving home for an hour or more after a long day of work, and thus less likely to suffer from fatigue-related crashes.
And even if you assume all of those confounding factors don't exist, there's still the elephant in the room, which is that most folks use AP only on streets where it has worked well for them in the past. Thus, the potential for AP-caused accidents is artificially reduced. If some other driver used it differently, that other driver could have very different results, making a general "this makes driving safer" conclusion impossible to reach without much more fine-grained data in which you compare the crash rates for various types of driving (city streets versus highways, urban versus rural, straight versus windy) independently with AutoPilot off versus on.
And realistically, you also need to separately compare AP unavailable versus AP off, because drivers may behave differently when they have deliberately disabled AP versus drivers who do not have AP. (This can determine to what extent regular use of AP makes drivers less situationally aware over time.)
In short, comparing the number of crashes to the number of vehicles is so prone to being skewed by other variables that it is almost useless as a metric for the safety of the vehicles. You might as well throw darts at a dartboard.
Personally, I think that AutoPilot reduces driver fatigue, which likely improves safety on the whole. But I'm not willing to state that definitively without actual data, which Tesla has thus far refused to provide. That's unfortunate, and it makes me wonder if they have something to hide. After all, if the data really were in their favor, you would expect them to be quick to release it. Unless, of course, they just haven't bothered to do any analysis, in which case I wonder about their competence.
In other words, I would say to Tesla, "Data or GTFO."
Check out my sci-fi/humor trilogy at PatriotsBooks.
The problem is that car accidents are in general vastly under-reported by the media. Until the last couple years, the single most dangerous thing you did was to get into a car (surpassed only recently by drug overdoses). On average, about 1 in 102 people you know are fated to die in a car accident. Compare to the odds of some of the other things the media devotes a disproportionately high (or low) amount of coverage time:
Suicide: 1 in 91
Police killed on duty: 1 in 104 (1.1 million officers / (135 per year * 78 year lifespan normalization)
Homicide by gun: 1 in 285
Drowning: 1 in 1,086
Fire: 1 in 1,506
Choking: 1 in 3,138
Killed by police: 1 in 4,336 (325.7 million / (963 * 78 year lifespan)
Complications from pregnancy: 1 in 5,965 (325.7 million / (700 * 78 year normalization)
Terrorism in U.S.: 1 in 28,033 (325.7 million / (3277 * 78 year lifespan / 22 years sample))
Killed by deer: 1 in 34,797 (325.7 million / (120 * 78 year lifespan)
Gun accident: 1 in 8305
Lightning: 1 in 114,195
School shootings: 1 in 121,033 (325.7 million / (138 * 78 year lifespan normalization / 4 years sample))
Dog attack: 1 in 132,614
Plane crash: 1 in 205,552
Terrorism in U.S. excluding 9/11: 1 in 248,954
Shark attack: 1 in 3,690,101 (325.7 million / (43 * 78 year lifespan / 38 year sample)
If news reports were truly unbiased, you'd expect to see:
Roughly 3x as many reports about fatal car accidents than gun homicides.
5x as many reports of women dying from pregnancy than reports of terrorism fatalities (including 9/11, 77x without).
39x as many stories about people dying of choking on food, versus school shootings.
43x as many stories about fatal car accidents than police shootings.
91x as many reports about suicides than gun accidents.
Over 100x as many stories about people being killed by deer, than killed by sharks.
The truth is the media picks and chooses which stories they want to publicize, whether it be because of their unusual and provocative nature (e.g. Tesla crashes, plane crashes, school shootings, shark attacks), or to serve a political agenda.
Meanwhile, in the real world, you'd find (if you actually hung out with Tesla owners) that when people get their Tesla, they tend to drive it more than the vehicle that it replaced, not less. For multiple reasons. One, it's a new car; they want to drive it. Two, it's a fun car, which reinforces #1. People often go on road trips in them. And three, it's cheaper to operate. The consequence of this is that multiple car households try to put as much distance on the Tesla as possible and minimize the distance spent driving their other vehicle.
And if you want to talk about demographics, EVs are most popular among people who own homes, not people who rent apartments / condos. Aka, the suburbs, not downtown.
As for whether people tend to use AP more or less in certain situations, that's irrelevant. AP isn't self driving; it's a combination of a human and the vehicle, and the result that matters is how the two interact. If the human - in driving 1/3 to 1/2 of the average vehicle's distance - does so in cases that AP handles best... well, good.
But trying to break apart AP and non-AP is beside the point. Associated Pres is trying to portray Teslas as dangerous. The numbers they gave to "prove their case" do precisely the opposite.
Give a boy a gun and you arm him for a day. Teach him how to make a gun, and the whole metaphor breaks down.