The Institute’s results also demonstrate the exceptional strength of Model 3’s all-glass roof, which is supported by a very strong metal body structure and helps protect occupants in roll-over crashes. During testing, the car’s roof was able to successfully resist more than 20,000 pounds of force – that’s more than if we placed five Model 3s on top of the car’s roof at once. And, the roof earned a higher strength-to-weight ratio score than any other fully electric vehicle that IIHS has ever tested.
> The Institute’s results also demonstrate the exceptional strength of Model 3’s all-glass roof, which is supported by a very strong metal body structure and helps protect occupants in roll-over crashes. During testing, the car’s roof was able to successfully resist more than 20,000 pounds of force – that’s more than if we placed five Model 3s on top of the car’s roof at once. And, the roof earned a higher strength-to-weight ratio score than any other fully electric vehicle that IIHS has ever tested.
SEJeff in particular has been a member since 2012 and has 7500 karma, and somehow never saw a post that looked unreadable because of blockquote, including his own.
I don't mean this in jest, I really do hope that you find some peace.
It's not like browsing the internet on a phone is a new concept any more. Little ridiculous how nothing has been done about its frequent use.
Prepending 4 spaces is intended to be used for posting source code, where automatic line wrapping is undesirable.
It's marketing speak and isn't supported by anything in the IIHS's actual public announcement on their most recent vehicle safety ratings. Tesla has received flak in the past for misstating the results of IIHS and NHTSA testing.
The whole Mars thing + no 30k electric car after all these years have me immune.
Tesla and musk are simply not trustworthy.
And as far as I'm aware they never promised a 30k car? They promised a 35k car, which you can buy today (you just have to email them or go in person and ask for it, it's not available online)
From this year. There will be no 35k Tesla.
Anyone who got one was merely lucky.
I'm not always a huge Musk fan, but I don't understand this kind of criticism at all. Going to mars is something the human race hasn't done before, it's no surprise it is taking longer than expected.
But that's not what it says at all. No where does it say the glass won't shatter.
The Model 3 roof has two roll bars, one links the A pillars and the other links the B pillars. That's what's strong.
See pictures of a rolled over Model 3 here: https://electrek.co/2018/07/15/tesla-model-3-rollover-crash/ The glass is broken but the rigid metal frame clearly protected the occupants from harm.
Metal gives and then fails. Glass gives minimally, and depending on tempering, can fail very, very quickly.
If you are not wearing a seatbelt, you run the risk of being ejected from the vehicle, and the higher risk of being 'partially ejected' from the vehicle, which is often worse.
I personally observed only 1 vehicle where the roof caved in on rollover (I believe it was a mid 2000s mustang). If the driver had been a few inches taller, I imagine he would've been hurt pretty badly.
The only other time I saw a roof collapse in wasn't from a roll over, but striking a tree roof first at a high rate of speed.
The only rollover crash I ever observed that the driver couldn't walk away from was a rather fortunate partial ejection. The vehicle only rolled onto its side, pinning the drivers arm between the road and the door as it slid for a few dozen yards.
The durability of the laminated safety glass should not be understated. A fire department's extraction team's approach is likely to be "cut through the glass with a circular saw".
Granted, all of this is anecdotal, with a very small sample size. Either way, wear your seat belt.
When I open TFA, I can't even find the word "glass" when I search for it.
Must have been moved from the Tesla press release to the institutes website press release.
To really answer your question, I've arbitrarily picked a few common vehicles with metal roofs.
The IIHS says the Civic roof withstands 13,195lbs of force: https://www.iihs.org/ratings/vehicle/honda/civic-2-door-coup...
A Subaru Outback: 18,533lbs of roof strength: https://www.iihs.org/ratings/vehicle/subaru/outback-4-door-w...
A BMW 3 series has roughly the same roof strength as a Model 3: https://www.iihs.org/ratings/vehicle/bmw/3-series-4-door-sed...
An Audi A5 is only 16,327lbs of roof strength: https://www.iihs.org/ratings/vehicle/audi/a5-coupe-2-door-co...
You get the idea!
Having very strong pillars is pretty much a prerequisite for surviving a rollover in an EV.
That isn't as big of a difference as I'd have expected.
The Tesla Model 3 is 3552-4072 lbs, while the new Jeep Wrangler (JL) is 3955-4455 lbs.
I could compare it to a miata that is ~2300lbs. But that's not really relevant then, is it? A Jeep Wrangler is basically a truck - so I don't think that's a good comparison.
Compared to other "compact executive cars" (as its class states) it's close to the competition.
The vibe I've got from talking to people is that they think the Model 3 (or Teslas in general) are significantly heavy relative to normal consumer vehicles. It's pretty surprising to them that even a 2 door Jeep Wrangler is actually heavier.
Same with my 1-series. It's obscenely heavy for how small it is.
A more spacious Alto that uses some aluminium and magnesium is around 650kg.
Talking about "not-so-heavy" cars...
That's a Pakistan-only car with no safety features . (From the Wikipedia description, "a car with everything manually operated, and with no safety features. The car lacks airbags, ABS, rear window defogger, side air conditioner vents, seat belt reminder and even rear seat belts.") So yes, it's light. You'll also immediately die in a collision, but you get what you pay for I suppose. (The Alto is sold more broadly, but it's still obsolete)
Not exactly comparing like to like.
Just about every family sedan on the market does >5x body weight.
Sure how many Gs one typically pulls during a car crash. 5 doesn't strike me as a lot, but of course this is specifically about force applied to the roof. I guess you have to not just roll but actually fall on top of it somehow. At any rate, some actual data would be great instead of being amazed at this marketing statement.
Yes, the roof can hold 5 teslas stack one on top of the other, but this doesn't tell you much.
More practical knowledge would be at what speed the car will fall on the roof before the roof collapses? 50km/h 70km/h 100km/h?
I am not sure tests of such an event would be still so favorable.
Or to put it differently - that's an impressive achievement of a situation that will never happen to a Tesla driver. What about performance in crash situation that might actually happen?
There have been some pretty insane examples of this, such as this one: https://www.autoblog.com/2016/05/06/tesla-model-s-crash-germ...
Especially when there are car manufacturers who try to hide flaws which can lead to catastrophic malfunction or had been caught with cheat devices for emissions test.
Also one more thing which is less talked about is Tesla's autopilot being useful for people with disabilities although Tesla hasn't added any specific features for people with disabilities AFAIK.
Tesla's car safety features are not a major selling point compared to its EV features, and moreover are generally features common to EVs (i.e., additional crumple zones, floor rigidity).
Tesla's autopilot still struggles to identify white trucks against a blue sky, and highway dividers, and can't seem to tell when the driver is asleep and has his hands off the wheel. So crash safety features take a back page to known dangers of Tesla cars that make Teslas more likely to crash than other vehicles.
Literally every newspaper and news organization in the US and Europe covered the VW emissions scandal. People went to prison over it. Billions of dollars in fines were paid. The stock price was downgraded by analysts for months. Ford and GM's stock prices are also down significantly despite near-record profits due to missteps in the sedan market. Analysts aren't biased against Tesla, they're biased against any company mis-executing.
IIRC, Toyota was eventually successfully sued because they did not follow industry best practices in developing the sw for the ECU. Any idea if Tesla's software is following, say, ISO26262?
Any idea if Tesla's software is following, say, ISO26262?
With respect to the console? Unknown. With respect to Autopilot? It's clear that they're not, since they're are too many issues and regressions that would have been caught if they were following those standards.
Citation needed. Which other vehicles? Tesla's overall safety record is quite good. Every car has things it does well and things it does badly, and every other manufacturer gets judged on the safety record on balance.
I mean, do you regularly post on HN about, I dunno, Toyota's safety record given the high center of gravity of its SUV offerings that make them "more likely to roll over than other vehicles" and claim that "crash safety features take a back page" to that problem?
Other than that, tesla does publish some data:
That said, people commonly constrain the Tesla comparisons to newer cars or in some other way.
So driving to the airport might well be safer than flying across the country according to the journey metric. Even using the hour metric I often am less than an hour from the airport and take more than a 4 hour flight.
In contrast, Teslas literally drive themselves into trucks, stopped cars, highway dividers, etc., resulting in the deaths of 7 drivers so far in a base of only a few hundred thousand vehicles, versus a rest-of-the-industry statistic of 0 across hundreds of millions of cars.
And finally, Toyota doesn't go around bragging about how its vehicles are the safest cars ever, or misstating (or even outright lying) about IIHS or NHTSA safety tests.
PS: Oddly enough of you replaced those billion miles with average American miles in average cars you would expect 12.5 deaths. Tesla’s autopilot might be less safe than other new cars in their segment especially if driven defensively etc, but that’s harder to quantify.
Do you have any stats on this?
That being said, this is simply a Tesla short that used incomplete (but true in the past) statistics to try and make Tesla look bad. They time boxed their statistics to 2016 and despite the post being made in mid 2018.
Tesla for the years 2019, 2018, 2017 has had 7 deaths per year each (6 so far in 2019). Despite the number of Tesla's on the road showing hockey stick like growth, the number of fatalities has remained nearly constant.
If the blog post were to be true, the number of fatalities due to accidents in 2017 should have been 14, then 24 in 2018, and closer to 50 in 2019.
Autopilot isn't solution for disabilities. If you aren't able to drive car without AP, you cannot drive one with AP - you have to be able to intervene and take over at any point of AP operation.
Don't get people killed by spreading misinformation.
Diability can be of various nature, affecting accessbility in various manner. Iam myself disabled, Tesla is not available in my country; I hope to drive one when available and so I keep track of experience of disabled people with Tesla.
That said you can find numerous videos of disabled people testifying Tesla autopilot is easier to use.
I feel you, but don't put your life in hands of a convenience package.
Other cars like Chevrolet Volt and Toyota Prius scored better.
Tesla went for years trying to make the _sticker price_ include "four years of gas savings: $3,500", bringing the 3 to $31.5K.
Of course you couldn't buy one for that price. That was fairly deceptive. Then Tesla fans tried to spin it as "trying to bring honesty to the market by looking at TCO"... except while they were happy to mention savings from gas, they didn't mention the cost of installing 240V, and their electricity costs were somewhat... absent.
I am by no means saying Tesla is less safe, in fact I do believe these claims, I am just not sure if they are truly thorough.
For a while there certain vehicle manufacturers were putting design aesthetic above functional headlights, and even some that didn't had mediocre headlights.
Since IIHS started dinging vehicles for having bad headlights things seem to have genuinely improved in a noticeable way. As a driver that has to drive on the same road as these vehicles (glare/etc) I am happy.
I would guess this is not the real reason. EV's have a much bigger crumple zone and can therefor absorb more energy. Since "nobody" wants to hear that EV's are saver than combustion engine cars, lets put them into another category...
It's worth remembering that modern cars are very highly optimized for the tests they have to pass. I'm not saying the M3 or any specific car is good or bad, just take the results with a grain of salt because these metrics are very much targets.
In terms of the "firewall deformation", from what I understand this is exactly what is supposed to happen in a collision: there is a "crumple zone"  in modern cars as a safety feature to absorb impact.
Edit: Is the firewall is structurally irrelevant in modern cars or is reality not convenient today? The inability for people to disagree without trying to silence each other is why people complain that HN is turning into Reddit.
The survival rates for collisions on highway speeds are in single digits no matter what you do. It's just laws of physics.
Modern luxury sedans have 1m+ crumple zones, and those only make for few percents extra chance at speeds above 60km/h.
Extending crumple zones beyond that is self defeating, as it will only lead to further increase of average car mass, leading to even more violent collisions
Preventing crashes from happening in the first place is far more economically efficient. EU is just few years away from making some forms of ADAS mandatory, and China is realistically talking about centrally controlled "autopilot" being introduced.
As in what, a head-on collision between vehicles traveling at 70mph? Yeah, that's not likely to end well. It's also a highly uncommon event and not what the safety ratings are testing for, nor is it what anyone is realistically expecting their car to protect them from.
The small overlap front test is reasonable. It's a 40mph along the outer edges of the vehicle, like if a passing truck on a local road drifts over the center line.
Most advances past that level just got cars getting unreasonably heavy, and more lethal, as a result, in car to car collisions, which in turn results in even higher expectations being placed.
That's why "safety marketing" tells only one side of the story.
Modern cars feature considerably more features (both in the bodyshell and in terms of active safety systems) which protect both passengers and pedestrians.
It's simply untrue to say we reached peak safety 20 years ago and everything else has just been pointless window dressing.
I buy and sell cars as a side gig, and newer cars (last 10 years or so) are far better equipped to deal with accidents - they have higher waistlines, better active safety features (airbags, stability systems, seatbelt pre-tensioners, lane assist, better-performing ABS systems), and the bodyshells have features such as side impact bars and larger crumple zones.
I recently bought a Fiat Seicento as a stop-gap. It was really old-school - light, small, nippy (even with a 55 bhp engine) and great fun to drive. But it was tragically weak and definitely not safe if you had an accident - as backed up by Euro NCAP's assessment of the seicento. Cars like that and the original mini are not something I'd like to have an accident in. (For reference, I did 10 years of gravel rally driving, so I've had a few knocks).
Most of the safety improvements are only very mildly correlated with the weight increases. Of course the safety features add weight, but comparable cars and compact SUVs have grown by at most 10-20% in weight over the past two decades. Most of the weight increase on the road is from pickup trucks (which don't satisfy many of these safety criteria and are bought mostly by people who don't care as much about safety) or people insisting on buying bigger vehicles.
While accident rates fell all across the world significantly with advance of road safety science, and active safety, no question there, but the lethality of car to car collisions has been rising, with most of rise happening in developed countries
The increase in average car weight is the biggest contributor to lethality of vehicle to vehicle collision.
https://www.nber.org/digest/nov11/w17170.html and https://academic.oup.com/epirev/article/34/1/57/493339
> being hit by a vehicle that is 1,000 pounds heavier results in a 47% increase in the baseline fatality probability.
In three decades, the weight of a family car went from under 1 tonne to close to 1.6, with big thanks to SUV overtaking sedan as a default "family car."
And increase in car resistance to mechanical deformation no longer plays a big role when G forces overtake blunt and compression trauma. And there is nothing one can do about them when crumple zones and the "deceleration track" are already maxed out.
The last few car fatalities I saw were all about that: 2 cars seemingly intact besides the front, but both flung violently 10m meters off the road, and people dead without much signs of external trauma. The energy absorption capacity of crumple zones did not save them.
Now you're saying that the issue is that car weights have increased.
That's a different argument - and the links that you cite aren't making the exact point that you are - the 1000lbs/47% figure may well be true, but that's a difference over 30 years, and both papers are actually talking about inequality between colliding vehicles.
The tendency to drive SUVs is an issue, for sure, and if you're in a little car then you don't want to have an accident with a large one where there is a physical incompatibility, but like all things it has become an arms race - I wouldn't want to be in a small car of yesteryear on the roads today because everything modern is bigger - so it's no wonder that people are driving them, as no-one wants to be in the small vehicle that will come off worst in an accident.
Your last few fatalities you witnessed are tragic, but anecdotal. I've seen many situations where people have walked away from accidents that would have killed them 20 years ago - indeed a couple of weeks ago I witnessed a car get T-boned by a good vehicle on an A-road in the UK, and everyone got out and walked away - that certainly wouldn't have happened 20 or more years ago. The statistics for fatalities in the UK (the only country I have checked) don't align with your statement that lethality is rising - it isn't in the UK.
The difference in the level of safety between these two is patently obvious.
The process is essentially this: the IIHS looks at the most common crash scenarios which involve fatalities. They replicate the the crashes in a lab to gather data on the forces exerted on occupants during the crash. Then a test scenario is formalized and future vehicles will be scored based on the forces exerted on occupants during the crash and whether such forces are survivable.
Auto manufactures are supplied information about the tests are are given an opportunity to design their vehicles to perform well in crash testing.
Note, that the IIHS is a independent scientific organization dedicated to reducing fatalities and injuries in motor vehicle accidents and is supported by car insurance companies. These companies wouldn't be funding the IIHS if it didn't have a measurable payoff (humans are expensive to fix).
So far, the only variable that I've found that is predictive of driver mortality is curb weight, and the relationship is not linear. There is a mortality minimum between 4000 - 5000 lbs of curb weight.
Again, this is just for trucks and SUVs, and just those 37 models that offered both 2WD and 4WD versions. Maybe the ratings work better for sedans or for other types of vehicles.
Airbag, seatbelt with pretensioner, rigid cabin, and some crumple zone is all it takes to survive a crash below that speed.
For a healthy, non overweight person, below 40, survival rates already close to 85%-90%.
The reason I'm ringing an alarm here is that survival rates began to slide back since mid naughties exactly because of the trend for more heavier cars being marketed as "safer" resulting in more violent car to car collisions, and more cars going through road barriers.
interestingly the larger vehicles represent a tragedy of the commons, where individuals feel safer in heavier SUVs but overall society is much worse off. If all SUVs were traded for sedans crash survivability would improve about 35% on average.
Where are you getting this data from?
If we can stop these kinds of deaths, we've made incredible progress.