Tesla AP already 10x safer & Tesla already self insuring- game over for FUD

[Pushmi-Pullyu]

2020 M3, showing only an Acceptable rating for the Small Overlap FrontDriver-side crash test. And only an Advanced rating (not Superior) on Pedestrian crash prevention.

I'm not sure what your motive is for cherry-picking just one or two factors out of an extensive series of crash test ratings, but clearly your motive is not objective consideration of all the facts.

The importance of the "small overlap" crash test has been greatly exaggerated in the popular press lately, using some poorly conducted studies. More rigorous analysis shows the relative unimportence of the "small overlap" type of crash. In fact, regarding severe injuries: "The risks of injury (AIS ≥ 2) for the full engagement, offset, and small overlap were 8%, 6%, and 3% respectively." Furthermore, the percentage of reported accidents which qualified as "small overlap", when rigorous methods of counting were used, is quite low.


(source)

I think it's safe to say that the IIHS has a lot of adjustment to do regarding the importance -- or rather, lack of importance -- of the small overlap crash tests.

This lack of rigor in IIHS crash testing is also one example of why the NHTSA's crash testing, and not IIHS's testing, is the gold standard.

 

[bwilson4web]

I'm not sure what your motive is for cherry-picking just one or two factors out of an extensive series of crash test ratings, but clearly your motive is not objective consideration of all the facts.

I agree. Worse, evidence of technical content remains sparse.

When someone brings technical content about their ride, regardless of the technology, I'm curious and want to understand. But when someone tries to tell me about the technology that is sitting on my driveway, inaccurate claims, well it is hard to be sympathetic or attentive.

About 10 years ago, the NHTSA convinced Congress to pass the "Bell the Hybrid Act," an anti-Prius act. They passed a law declaring noise makers would make the Prius and EVs safe to pedestrians and bicyclists. A bad law from a flawed study, I have no use for it but had briefly thought it amusing to use it in defense of an equally flawed claim about Tesla Model 3 pedestrian safety. I am not making that claim today even as a joke.

There was a claim about safety in the 'top 10' vehicles that turns out to be empty. Instead, we see evidence of scraping together what could be poorly harvested from the IIHS web-site. Is the alleged 'son with an M3' as accurate? Does the claimant actually have something as real as this:

Trade-in Prius Prime for Standard Range Plus Model 3, March 26, 2019, Nashville, TN.

Bob Wilson

 

[DaleL]

There are many possible ways in which to rank vehicles on safety. Insurance rates are another indicator of, if not safety, at least the cost of ownership and related to safety. There is an interactive comparison tool at insure.com: https://www.insure.com/car-insurance/car-insurance-comparison/results Unfortunately the interactive tool does not have the Model 3. However, the Ford F-150 trounced the Tesla Model S and Model X. Even the Honda Clarity was cheaper than either Tesla to insure. Of course a large part of that is undoubtedly the high cost of the Teslas.

State Farm has a tool for comparing insurance costs and does include the Model 3. https://www.statefarm.com/simple-insights/auto-and-vehicles/vehicle-insurance-ratings However, it only shows one vehicle at a time. I have captured the image of the Model 3 vs the Clarity PHEV. They are not that much different in initial cost. The Tesla is credited with better vehicle safety, but looses in other non-safety areas.

If a person only wanted safety for themselves, the State Farm tool indicates that the massive GMC Yukon is even safer than either the Clarity or Model 3.

There is also the issue that no matter how safe a vehicle is when purchased, it is likely that next year's model, or some other brand, will be marginally safer. Other vehicle characteristics most also be considered. Let me emphasize, I am not bashing the Model 3; it is a VERY safe car by any measure. I am only questioning those, including Elon Musk, who make the claim that it is the safest car on the road.

 

[R P]

Tesla cars under control of Autopilot have a significantly lower accident rate than those not using Autopilot, and that's a Truth which can't be denied, despite all their EV-bashing efforts.

I have said over and over in past discussions on this forum, that ANY car with the new Driver Assist features (but only comes with Autopilot in Tesla models) will be safer. That IIHS study (done in 2018) I linked to in an earlier post said 41% less accidents in Subaru's that had the EyeSight package (with the Driver Assist features) than those that did not. In those days, Subaru had the longest history and best data for driver assist. Today, most Subaru's now have EyeSight standard (except in their manual drive and sport cars). Eventually, I think most all cars (will probably be mandated) will have all these driver assist features.

The question is why is Autopilot (which includes the Driver Assist features) not standard on all Tesla cars? They could separate Driver Assist from Self Driving and only make the latter optional. Should be just a software thing. Could it be the reason is to make you pay extra, for basic features that should be standard? Or maybe so they can continue to gather statistics that show Autopilot with the Driver Assist features is safer, and market that as a selling point. What about the accidents with drivers that didn't have Autopilot? Are they just guinea pigs for Tesla testing and to support their marketing hype.

This is not Tesla bashing. Just some obvious questions when you look from outside the Tesla bubble.

I want safer cars, period, from ALL manufacturers. No marketing hype, just do it!!

 

[bwilson4web]

The question is why is Autopilot (which includes the Driver Assist features) not standard on all Tesla cars?

Source: https://www.tesla.com/support/autopilot

Do all cars have Autopilot?
All new Tesla cars come standard with Autopilot as of April 2019, which includes Traffic-Aware Cruise Control and Autosteer. . . . ​

This is not Tesla bashing.

Mr. Google knew.

Bob Wilson

 

[interestedinEV]

I want safer cars, period, from ALL manufacturers. No marketing hype, just do it!!

Amen

Tesla has great technology, they have great success but also great hype. There are others with great technology, and I do not know if it is better or worse than Tesla.

If you live in the east valley of Phoenix, you cannot miss Waymo (Google). Just because the technology is not seen everywhere does not mean it is bad or does not exist. They are offering a driverless taxi service in my area. Repeat a driverless taxi (you can get a taxi with a person sitting at the wheel ready to take over or one without anyone at the steering wheel, level 4). You can dismiss it as being in a geo fenced area, good weather, flat roads etc. but they are that confident that are putting their money where their mouth is.

And they have no recently reported accidents, even though hundreds of their cars are driving around my area 24 hours a day. Cruise, Mobileye (Intel) etc are doing the same. Mobileye has a taxi service in Israel where the driving conditions are different. There is lot work being done here and Tesla and everyone else has long way to go. Let us recognize Tesla for the great job they have done, but there are still lot of glitches to work out and that is why insurance companies are cautious. And that Tesla may not have the best technology, we do not know.

And this is not Tesla bashing, it is being factual.

 

[R P]

Source: https://www.tesla.com/support/autopilot

Do all cars have Autopilot?
All new Tesla cars come standard with Autopilot as of April 2019, which includes Traffic-Aware Cruise Control and Autosteer. . . . ​

Mr. Google knew.
Bob Wilson

OK, that's great. And about time... Glad to see they are catching up with the other car manufacturers. My son, with his model 3, had to pay a bunch extra just because he wanted the driver assist features.

 

[interestedinEV]

So when the opening poster says "Tesla AP already 10x safer" what is he comparing with? The previous post talks about others whose technology may be better than Tesla. So I generally dismiss what the OP says.

Is Tesla AP safer than a driver driven car? No one is going to argue that it is not. But there is still much more that is unknown.

Something that insurance companies are concerned about is complacency i.e. over dependence on the autonomous system leading them to ignore avoidable risks. That people can fool the system and become complacent. It is reported that it is easier to fool a Tesla AP then some of the cars. I agree people should not do this, but they do. Insurance companies may take that into consideration when rating Tesla cars. That is reality, some bad apples can spoil it for everyone else.

 

[R P]

Something that insurance companies are concerned about is complacency i.e. over dependence on the autonomous system leading them to ignore avoidable risks. That people can fool the system and become complacent. It is reported that it is easier to fool a Tesla AP then some of the cars. I agree people should not do this, but they do. Insurance companies may take that into consideration when rating Tesla cars. That is reality, some bad apples can spoil it for everyone else.

That's very true indeed. I rented a car recently that did not have driver assist, not even blind side mirror warning. I found myself wanting to change lanes if the car isn't warning that a car is there. Fortunately, I still look over my shoulder. But it totally illustrates this potential problem.

And if you watch Mayday, there have been a lot of airline accidents due to pilots becoming too complacent and dependent on their autopilots, and not skilled enough at manual controls flying. I am actually a pilot (private, not commercial) so always interested in that type of stuff. That 777 accident at San Francisco a couple years ago was an example of that, and there have been many more.

 

[R P]

Subaru has long been a leader in safety tech, and I see for 2019, they still lead the pack.
https://www.forbes.com/sites/tanyam...t-domestic-brands-nearly-absent/#41d9a07a19f8

And Hyundai is not far behind.
https://www.iihs.org/news/detail/57-models-clinch-2019-iihs-safety-awards

Seems Tesla is far from being the best..., as per this quote from the announcement.
Absent from the winner's circle are Fiat Chrysler's Dodge, Jeep and Ram brands; Ford and its Lincoln brand; General Motors' Buick, Cadillac, Chevrolet and GMC brands; and Tesla. All failed to qualify for either award.

 

[bwilson4web]

This remains a very interesting question:

So when the opening poster says "Tesla AP already 10x safer" what is he comparing with?

A while back, the NHTSA took exception to Tesla claiming they had the safest cars based on NHTSA testing: https://www.southbendtribune.com/ne...cle_d873da01-8267-5b3e-acba-6925cec3324c.html

WASHINGTON — Federal safety regulators accused Elon Musk of issuing “misleading statements” on his company’s Tesla Model 3 last year, sending a cease-and-desist letter after the chief executive claimed it was safer than other vehicles.

The National Highway Traffic Safety Administration admonished Musk for claiming it had “the lowest probability” of injury ever tested by the agency in its safety ratings program, according to the document obtained by government and legal transparency group PlainSite. That’s because the NHTSA said the mass of a vehicle plays a role in passengers surviving crashes, something that’s not directly comparable between cars of differing sizes, despite the Model 3’s 5-star rating.

NHTSA referred the matter to the Federal Trade Commission’s Bureau of Consumer Protection, it said, to determine whether that statements constituted “unfair or deceptive acts or practices.” The documents show NHTSA also subpoenaed information from Tesla on multiple crashes involving its vehicles.
. . . ​

Tesla had taken the raw metrics from NHTSA testing which gave a finer number than the buckets NHTSA reports. For example, grades listed by A, B, C, D, and F actually come from a numerical evaluation of the test with give a 'class ranking.' Without getting into the politics, it was clever to calculate the numbers. But for accuracy, one would want to crash a larger sample set of cars ... one crash test can mislead both higher and lower than a true sampling.

So I'm content ours is very safe car. With the latest updates, it changes lanes better than I can because of the 12 ultrasonic sensors, 8 cameras, dedicated processors, and code base. It is the difference between an automated, race car driver and a 70 year old man.

Bob Wilson

 

[interestedinEV]

Tesla had taken the raw metrics from NHTSA testing which gave a finer number than the buckets NHTSA reports. For example, grades listed by A, B, C, D, and F actually come from a numerical evaluation of the test with give a 'class ranking.' Without getting into the politics, it was clever to calculate the numbers. But for accuracy, one would want to crash a larger sample set of cars ... one crash test can mislead both higher and lower than a true sampling.

Even then 10X better?

Tesla, led by billionaire Elon Musk, has said the Model 3 "was engineered to be the safest car ever built" with the lowest probability of injury of all vehicles ever tested by the National Highway Traffic Safety Administration (NHTSA).

Forgetting the clever use of numbers, What about cars that are not tested by NHTSA? Is Tesla safer than Chrysler Pacifica Van operated by Waymo?

 

[R P]

Forgetting the clever use of numbers, What about cars that are not tested by NHTSA? Is Tesla safer than Chrysler Pacifica Van operated by Waymo?

Well, there is the number of accidents, and then there is what happens when you are in an accident. I say the latter is most important, but of course the former is, too. But no matter how good your driver assist is, not going to stop all accidents from other bad drivers. T-boners are among the hardest to protect against.

But yeah, NHTSA certainly doesn't test all cars, usually just the suspect ones.

 

[bwilson4web]

Subaru has long been a leader in safety tech, and I see for 2019, they still lead the pack.
https://www.forbes.com/sites/tanyam...t-domestic-brands-nearly-absent/#41d9a07a19f8

And Hyundai is not far behind.
https://www.iihs.org/news/detail/57-models-clinch-2019-iihs-safety-awards

Is that the December 2018 report?

Bob Wilson

 

[R P]

Is that the December 2018 report?
Bob Wilson

Yes, for 2019 cars. I searched but couldn't find the same report for 2020 cars yet, just announcements that the criteria will be tougher for 2020. Those overlap tests are part of that, as well as pedestrian detection.

 

[bwilson4web]

I'm more interested in: https://www-fars.nhtsa.dot.gov/Main/index.aspx

• Current to 2017 with "Fatalities per 100 Million Vehicle Miles Traveled"
• Individual accident reports can be extracted with make and model of vehicle

In about two years, we should have the 2019, FARS data. Then we can compare relative fatality rates by make and model. Sad to say, we're not seeing enough Tesla deaths for a statistically significant, high confidence sample.

I'm more interested in the National Transportation and Safety accident reports. They do a deep dive with the type of data relevant to the specific make and model. They don't do a population study like the NHTSA but give engineering insights.

Bob Wilson

 

[Pushmi-Pullyu]

The question is why is Autopilot (which includes the Driver Assist features) not standard on all Tesla cars?

Autopilot is now standard equipment on all Model 3's. Not sure about the Model S and Model X.

But you could ask the same question about any safety feature. Why did auto makers resist putting seat belts, and later air bags, into cars? Because it made them cost more, that's why. It's not that auto makers are evil, it's that competition makes lower prices and higher sales more important to a business than trying to make sure every car is as safe as possible.

Take that argument to its logical conclusion, and you could ask why cars are not literally built like tanks, or at least like military armored cars. Just think how much safer we'll all be if they were! But if course if they were, few would be able to afford them.

​

 

[Pushmi-Pullyu]

If you live in the east valley of Phoenix, you cannot miss Waymo (Google). Just because the technology is not seen everywhere does not mean it is bad or does not exist. [snippage of good stuff] You can dismiss it as being in a geo fenced area, good weather, flat roads etc. but they are that confident that are putting their money where their mouth is.

I personally don't at all dismiss it. Going by just what's reported in the popular press, it looks to me like Waymo is ahead of Tesla in semi-autonomous driver tech. As I have said many times, I don't think Tesla will progress much further toward actual Level 4/5 autonomous driving so long as they refuse to use active scanners, likely either phased-array radar and/or lidar.

However, let's not ignore certain facts: Waymo's "robotaxi" fleet operates in (as you said) a geofenced area, one carefully chosen for having streets laid out in an precise grid format, with few if any irregularities. And, as you noted, good (or at least clear) weather most of the year... for those who consider sustained temperatures above 90° for days or even weeks at a time to be "good weather"!

In other words, Waymo carefully chose a suburb of Phoenix for their experimental robotaxi fleet trial because of the simplified driving conditions there. How soon they will be able to expand the use of robotaxis beyond that area, remains to be seen. However, a brief Google survey indicates Waymo is attempting to get regulatory approval for another such robotaxi service area in the Bay Area of California.

​

 

[Pushmi-Pullyu]

Even then 10X better?

Well of course not. There have been claims pop up in comments to InsideEVs articles, lately, claiming that a Tesla car under Autopilot is "2x as safe" as an average car. Personally I'm somewhat skeptical of that claim. I think it needs a lot more study and data before we can treat it as established fact. However, the claim in the OP of Autopilot making Tesla cars 10x as safe, or 10x safer is... ummm... somewhat tangential to reality. Elon Musk has in the past said that the goal of autonomous cars should be to reduce the accident rate no more than 10% of the current rate.

But that's the ultimate goal, not any claim of what Autopilot can currently do. Furthermore, it may be an aspirational goal rather than a realistic one. Elon wanted SpaceX to reduce orbital launch costs by a factor of 10, but last I saw reported, SpaceX had only managed a 5-fold improvement. Reducing launch costs by 80% is an astounding feat, so let's not belittle that! But it's somewhat short of the aspirational goal.

Forgetting the clever use of numbers, What about cars that are not tested by NHTSA?

Are there any cars sold as new, at least in first-world countries, that are not subjected to crash testing before being sold? My understanding is that crash testing is mandatory for a car model to be sold as "new" in the USA, and presumably in EU countries and most if not all other first-world countries. (See the list of regulatory agencies in different countries here.) Not every model is tested every year. Significant changes in models, particularly model year updates involving changes to the car body, may require new testing. Possibly there is a gray area there regarding just what is or isn't a significant change requiring new testing?

There may be an exception to all new car models requiring crash testing before sale: I see a claim on Quora that, in the U.S., a car with a production of less than 1000 units (is that per year, or a total production run?) can get an exemption from crash testing.

The NHTSA may itself not conduct that many crash tests in a year, but it does license companies to do such testing.

 

[interestedinEV]

Well of course not. There have been claims pop up in comments to InsideEVs articles, lately, claiming that a Tesla car under Autopilot is "2x as safe" as an average car. Personally I'm somewhat skeptical of that claim. I think it needs a lot more study and data before we can treat it as established fact. However, the claim in the OP of Autopilot making Tesla cars 10x as safe, or 10x safer is.

Here is the problem I am having with this discussion thread. Crash testing rankings and safety are being considered synonymous and being interchangeably used. Crash resistance is a very important part of overall car safety. But there are other factors which play a part, for example headlight performance. (@R P says a lot of what I am saying)

Even highly rated vehicles can have issues For example, IIHS lately has been ranking headlight performance, and it's found that a model scoring high on crash tests might have only acceptable ratings for lights. Or some versions of that vehicle might have poor lights, while others have good lights.
.

That article continues

Ratings aren't the whole picture. Even if you carefully sift the IIHS and NHTSA scores, you still want to consider additional safety gear, NHTSA advises. It recommends these: forward collision warning and automatic braking (both of which IIHS requires for its Top Safety Pick+ rating); lane departure warning to alert you if the car drifts into an adjoining road lane; and rear-view camera for safe backing and parking.

So when someone says "Tesla is X times safer then every other car", can you define how you measure overall safety? If crash resistance or IHS ranking or CR ranking etc. is the only criteria, then you cannot make an absolute statement.

We have some very safe cars today, and each car provides safety on several dimensions. Suburu is very safe car and Tesla is a very safe car. If you tell me Tesla is safer than Suburu because it has good crash resistance and (note the and) Autopilot, that is your analysis and others can agree or disagree.

So let me leave you with a hypothetical question. I buy a Chyrsler Pacificia identical to one Wyamo bought. But then Waymo adds sensors and the AV stuff. However they do not change the crash resistance or headlights or seat belts etc. Which is a safer car, the only I use or the one Waymo uses?

(Corollary question: Can @101101 or someone rank these cars in order of safety)
a) Chrysler Pacifica off the dealer lot
b) Chrysler Pacificia with Wyamo controls
c) Tesla without AP
d) Tesla with AP