How do cars do in out-of-sample crash testing? (2020)
danluu.com
danluu.com
Obviously it would be better if they had a deep commitment to safety and made every variation of every model maximally safe. But I'll take it. No such thing as absolute safety, so moves in a better direction are good.
(1) A remarkable example of a private agency that has pushed safety standards beyond what the government would do own its own, and
(2) An organization that has persuaded Americans to buy larger vehicles than they would have otherwise with all the associated costs (e.g. the “affordable car” crisis) and risks (to pedestrians.)
The IIHS is an organization of insurers so they are particularly concerned about quantifiable monetary costs. And when it comes to that much more of the benefit of larger vehicles is in avoided minor injuries such as broken bones which are more common than death and life changing injuries. The public focuses on the latter and the psychology is such that some people will spend another $50k on some German vehicle and spend the rest of their days at the dealer getting it fixed or subject their children to the trauma of riding in a minivan. (To generation X the minivan is like the toxic PFAS GenX)
IIHS claims that compatibility has improved between large and small vehicles but that large vehicles are still a menace to other road users
Maybe now that we live in the age of 1k/corner head and tail lights it'll happen...
The last 12 years mostly replaced attention-preserving tactile controls with attention-demanding screens.
Coexisting with that decreased ability is a race-to-the bottom where new vehicles kill visibility for everyone else (headlights blind ahead; oversizing erases visibility of every car around).
I have never felt less safe on the road.
And for the privilege, everyone's insurance rates climb and climb and climb - unreasonably punishing people who don't drive super-expensive-to-repair vehicles.
But I was thinking more about passive safety improvements from body structure engineering and auto-braking systems.
In the last 12 years we have seen the following:
1. Further proliferation of 360 cameras, blind spot monitors, rear and front cameras, digital rear view mirrors, and even some vehicles with in-dash blind-spot cameras and including them down-market. This directly translates to fewer fatalities, especially of children.
2. Further proliferation of HUDs and including them down-market, including augmented reality HUDs from the germans, allowing drivers to keep their eyes on the road for a larger percentage of the drive. This directly translates to fewer fatalities per mile
3. Better auto-high beam systems, including truly amazing systems from the Germans (to reduce the dazzle issue you brought up above). Also down-market movement of things like auto-turning lights to improve safety of turning at night.
4. Huge improvements in automatic front and rear emergency braking, including for people, cyclists, and children. This is likely the most impactful, and has massively reduced deaths, and minimized the harm of situations that do become harmful.
5. Some cars have automatic evasive turn systems, which occasionally intervene and prevent crashes before they happen similar to 4.
6. Some newer cars have safety stuff like automatic crash based loud noise systems (prepare ears for crash), automatic crash based seat belt pre-tension systems and similar systems to mitigate crash damage. Onstar automatically calls emergency services regardless if you pay and speeds up ambulance and thus life-saving care.
7. Carplay/Android Auto means less time fumbling with deciding where to go. Huge amounts of distraction caused by shit infotainment systems.
So done with people acting like driving is worse because we took 1 or 2 steps back for every 7 we take forward. Ludditism is the human death drive externalized in a lesser form.
As a broader extension of the article's point, testing only crashes is a fundamentally flawed approach to the survivability onion, the safest car is the one that doesn't get into a collision in the first place, i.e. one that's smaller and provides the driver with better visibility. More recent cars may be more heavily armoured but less safe.
I'm also very curious on other impacts the vehicle sizes have on things. Easy to think it contributes to so many cyclist with basically no helmets. Curious if data backs that.
One factor here in the UK is that many of our cities and towns predate cars (sometimes by centuries) and consequently the roads and streets are much smaller. Such that giant cars are a real nuisance for both the driver and everyone else. But there is a definite trend toward larger vehicles, as can be seen by how difficult it is to get them into older parking infrastructure.
Compare to Europe and Japan, US has - MUCH lower energy costs, larger family sizes, much further travel distances, and smaller % of population in pre-modern urban areas with pre-automobile sized roads.
Add to that a less top-down centralized government in the US and the resulting lack of mass transit, particularly inter-city, and you end up with more people, driving bigger cars, further.
Unfortunately due to duration / cost of ownership, families over-purchase in terms of vehicle capacity.
Yes, dad maybe does most of his driving to work alone in the car. But on weekends he needs to be able to lug the kids around in their giant baby seats, strollers, etc.
So most parents I know end up owning 2 vehicles of sufficient size for their overall needs.
I know some families that have an extra car for commuting on top of the 2 family sized cars but this itself is something of a luxury.
Sure, but Dad doesn't need 3,000lbs of payload capacity and a 6ft+ bed and wheel spacers and an extra lift kit to carry a stroller.
I've got two kids and a two-child stroller. There's more than enough space in something the size of a Mach E or a Model 3/Y to have multiple kids in car seats and all their stuff.
In the end pretty much all of that could have fit in my old Accord as well.
But a TON of people have vehicles in the MachE/ModelY crossover footprint. Look at how every brand has 3 cars that size and they all look the same.
All of these, including your old accord are bigger than a Prius.
Going off lists like this - https://www.caranddriver.com/news/g43553191/bestselling-cars...
While theres some very high volume truck models, which in fact take up the 2 highest sales by model slots, overall if you add up the far larger number of sedan/crossover/suvs you get to ~2x the overall sales.
I suppose one can argue about how you categorize big SUVs like say the Explorer and Grand Cherokee.. overall I think the numbers still hold.
Completely agree there's still more big trucks on the road than there should be, but '3,000lbs of payload capacity and a 6ft+ bed and wheel spacers and an extra lift kit' is a very specific demographic, and they were never going to be talked into a Prius.
People tell me to just rent a truck when you need one. However that is hard and expensive. Most places won't rent you a truck, they will rent you a truck shaped vehicle with restrictions meaning you can't use it like a truck. When you find an exception the costs it very high and they charge per mile rates - it doesn't take many uses to be worth the extra costs of having a truck that does everything instead even before you account for the hassle of going to get the rental truck.
Of course small cars can do more than most people give them credit for but they are still compromise and there are a lot of things that they can't do.
Dad probably doesn't really need it, ever.
One friend argued he really needed a truck because he bought a lawn mower off Facebook Marketplace and wouldn't have possibly gotten it home without his truck. It's the only truck-like thing he's done in a year or two. Strange, I managed to put a lawnmower in my small crossover/hatchback without any issues.
Another argued he needed a truck because he goes mountain biking. No possible way to take a bicycle without an extended cab 6ft bed.
This evening, I have to take the kids and their stuff to visit the grandparents for an overnight. Their home is 30mi away from me. Even though I've got public transit immediately outside my door, there's no public transit servicing their home. I could take the two-hour bus ride for the bus at the edge of my neighborhood to the edge of the public transit network, it's still >10mi to their home. This trip for this afternoon on Uber is ~$40 each way. For one trip I'll make a few times in a month. Or I can take my car and make it a 25-minute trip that costs me several dollars each way.
We also go visit some family >70mi each way from us on a pretty regular basis. There's definitely no transit available to get to where they live, and an Uber going all the way over there is >$100 each way.
I mostly work from home, the "office" is ~2mi from my home, I love to ride my bike and I take public transit a decent bit, and yet I still manage to put >13,000mi on my car every year. That is how extremely car centric DFW is.
Most people I know who "need" a truck would absolutely be able to do everything they use their truck for with a small hatchback to a medium sized crossover.
What level of sacrifice you make to get what you really need apparently isn't necessarily comparable with anyone else, right? Commuting is a choice. No one is making you work where you do, or live where you do, at least not since 1863.
No judgement here, I'm just pointing it out. Is it possible we're all on the same team? Even the pickup owners?
Once again, let me restate that, and maybe it'll sink in.
> Most people I know who "need" a truck would absolutely be able to do everything they use their truck for with a small hatchback to a medium sized crossover.
These people aren't really doing anything that would seriously impact their lives if they didn't own a truck, but they convinced themselves they needed a truck.
I'm not saying this about all truck owners; I know lots who actually do tow campers and boats, who actually do pickup yards of gravel and what not on a regular basis, who literally do live on a ranch, etc. And I know tons of people who own a truck because how else would they get their mountain bike to the trail across town, or who have such a culture deeply ingrained in their mind they couldn't be seen driving a smaller vehicle for more than a day. That they need it because "it's comfortable", as if a luxury sedan isn't comfortable. For many it is a lifestyle choice they choose to make to drive a vehicle massively oversized for their needs, and is not at all similar to the fact I need a car to to most stores and generally function in society because that's how this city was designed.
If you haven't been to the Southern United States, you probably just don't understand the truck culture I'm talking about.
You may think I'm making some straw man, but I personally know more than a dozen of these people.
What exactly is your point?
Of course everyone has a different life. Who are you to judge the life of others?
I'll be judgy when their excessive life choices make my family less safe, excessively poison our water and air, and ruin our city designs catering to their excess.
There is drastically different cargo space in a midsize (not even 3 row) SUV than a sedan. Even a hatchback is drastically different cargo capacity.
Some of it is you don't know what you need until you need it.
With my non-hatch sedan I had all sorts of problems bringing home even moderately sized normal purchases. I'd have to unpack products to fit them into the car and/or wedge into the backseat with sales rep assistance. Lamps, mini fridges, large home air filters, plants from the garden center, wood/drywall from the hardware store that I'd need pre-cut down just to fit in care.
Some of these purchases were emergency backup mini fridges when my home fridge died & replacement would take a week. Or $10 of wood/drywall to DIY repairs instead of $100s of a contractor. Or paying 5x the price of the materials for home delivery. Likewise for plants vs paying a landscaper.
Just pretty frequent stupid stuff that a tall SUV trunk with a large hatch opening solved. Even a midsize hatchback would do the trick.
Not defending status symbol trucks but he rear seats of compact cars aren't known for getting a lot of ass.
I guess it's rational for me to buy a 25' box truck to daily drive since I used to rent them every couple of years for moving apartments.
Passenger vehicles with enough back-seat room for adults, a third passenger row, more cargo room, or 4WD capability are expensive and inconvenient enough to rent + cheap and practical enough to daily-drive that the use case would really have to be pretty rare.
The problem is that for all we like to claim to be rational, most people make those decisions emotionally and rationalize as needed to explain it. That’s how you get 90% of SUVs never leaving a paved road because the owner thought it was a lifestyle accessory which made them look richer, but talks like they’re suddenly going to start taking 6 friends backcountry camping year-round.
Yuppies driving Range Rovers in Manhattan is a whole different discussion though.
I saw Russia being slightly higher but figured it was marginal and “we emit slightly less than a petrostate in the middle of bombing the fuck out of their neighbour” was hardly a “win” for the USA…
Tesla Model S is wider than Mercedes G-Class, a literal military transport. That's something.
The danluu article also mentions an iSeeCars report at the bottom that says that Teslas have a high amount of fatalities per mile driven [2]. However, while they claim to be using official US estimates of fatalities data, they normalized it with estimated mileage using unknown proprietary iSeeCars data:
> iSeeCars analyzed fatality data from the U.S. Fatality Analysis Reporting System (FARS). Only cars from model years 2018-2022 in crashes that resulted in occupant fatalities between 2017 and 2022 (the latest year data was available) were included in the analysis. To adjust for exposure, the number of cars involved in a fatal crash were normalized by the total number of vehicle miles driven, which was estimated from iSeeCars’ data of over 8 million vehicles on the road in 2022 from model years 2018-2022. Heavy-duty trucks and vans, models not in production as of the 2024 model year, and low-volume models were removed from further analysis.
Some numbers are also surprising, such as the Model Y having almost twice the fatality rate (10.6 per billion miles) as the Tesla average (5.6) --- suggesting that there's a wide gulf between it and the Model 3 in terms of fatality rates, which seems difficult to explain. Ultimately, iSeeCars is a small VC-funded startup with very few people and so it is unclear if their methodology is actually good or not.
For a better blog post about whether Teslas are safe, here's a post from Brandon Paddock [3] from a year ago, which seems more or less objective and unsurprising. Here is the conclusion that he drew upon analyzing the FARS data:
> Tesla’s fatal accident rate is nearly identical to that of the Audi A4 series, and far lower than a standard Ford mid-size sedan. In this case, the Ford’s accident rate is more than 4 times higher than the Tesla Model 3.
[1] https://www.teslarati.com/tesla-model-3-aces-safety-tests-ch...
[2] https://www.iseecars.com/most-dangerous-cars-study
[3] https://brandonpaddock.substack.com/p/are-teslas-the-most-or...
My guess is that is due to higher average occupancy for model Y vs model 3.
Actually, I wonder... if a car is safe and avoids many accidents, could the unavoidable accidents be worse overall?
Kind of like a drug that cures disease #1 blamed for causing disease #2... but actually the drug worked and the next statistical thing to die of after #1 was #2
Just curious, what kind of critical touch screen controls Tesla has? I thought there's nothing critical.
The only "critical control" that is on the touchscreen is choosing whether to go forward or backward (aka shifting into drive or reverse), which can be done on the touchscreen on the Model S, X, 3, and Cybertruck. Of course, this is typically only done when the vehicle is stopped, so it's fine. There are also backup physical buttons in case you can't use the touchscreen.
No current tesla has stalks on steering wheel, so there are areas to touch on (rotating) wheel for turn signals, wiper, high beams
We can also observe that,
in the IIHS analysis, many
factors that one might want
to control for aren't [...]
One way to see that the numbers
are heavily influenced by
confounding factors is by
looking at AWD or 4WD vs. 2WD
versions of cars. They often
have wildly different fatalty
rates even though the safety
differences are not very large
(and the difference is often
in favor of the 2WD vehicle).
Some plausible causes of that
are random noise, differences
in who buys different versions
of the same vehicle, and
differences in how the vehicle
are used.
A typical AWD will have a transfer case, front driveshaft and other front drivetrain components v.s. a sister car in a 2WD configuration. The overall weight and distribution will be different.
One might e.g. suspect that in a crash the front driveshaft might fold upwards and assist in structurally compromising the passenger cabin.
> To adjust for exposure, the number of cars involved in a fatal crash were normalized by the total number of vehicle miles driven, which was estimated from iSeeCars’ data of over 8 million vehicles on the road in 2022 from model years 2018-2022.
Note that the mileage data for each model comes from iSeeCars, not FARS. We don't know how accurate it is. There could be selection effects because iSeeCars only gets odometer data for cars that are for sale or have been sold, not vehicles that people keep. If for example, someone reserved a Model Y and sold it above the original price as soon as they got it (a common practice for new Tesla models), and the second owner kept the car indefinitely, iSeeCars would only see the odometer reading at the early sale. Tesla mileage would also skew low in sales data for anyone who bought an electric car, then regretted it for one reason or another, and exchanged it for a gas car. And since the rate they are measuring is fatalities per vehicle mile traveled, not per passenger mile traveled, any model that tends to contain more people will skew higher in the stats.
If you query FARS data to get all Teslas involved in occupant fatalities from 2018-2022, there are 141.[3] If you split out the data by model, there were a total of 22 Model Ys involved in occupant fatalities from 2018-2022 (actually 2020-2022, as the Model Y was released in 2020). If there were 10.6 Model Ys involved per billion miles, then (according to iSeeCars) the total number of miles traveled by Model Ys was 2.07 billion. By the end of 2022, the US had approximately 484,100 Model Ys on the road.[4] If you divide the total vehicle miles traveled by the number of vehicles, then the average Model Y had less than 4,300 miles on its odometer. That seems suspiciously low, as Tesla's 2021 report put the average Model Y at 13,000-14,000 miles per year.[5]
This artifact in mis-measuring mileage would also explain why other recently released models seem to have high fatality rates. The Honda CRV Hybrid was released in 2020. The Buick Encore GX was released in 2020. US production of the Hyundai Venue didn't ramp up until 2020. The Toyota Corolla Hybrid was released in 2019. Most people who buy a new car either sell it quickly (because they dislike it and want something else), or they keep it until it gets old. So anyone collecting milage data from car sales will have data that is skewed low for new models because the higher mileage sales haven't happened yet.
It may be the case that I'm using different reports than iSeeCars did. It's hard to know whether they're looking at vehicles involved in fatal crashes, drivers involved in fatal crashes, drivers killed in fatal crashes, occupants involved in fatal crashes, or occupants killed in fatal crashes. FARS lets you query all of these, and the iSeeCars post doesn't make it clear which data they used.
Another odd thing about the iSeeCars data is that they claim an average of 2.8 vehicles involved in a fatal accident for every billion miles traveled. But in 2022, the US had 13.3 deaths per billion miles traveled by car. Unless the average accident involves 5 fatalities, it seems to me that the iSeeCars data has some significant issues.
1. https://www.iseecars.com/most-dangerous-cars-study#v=2024
2. https://www.nhtsa.gov/research-data/fatality-analysis-report...
3. https://cdan.dot.gov/query I don't know how long they retain generated reports for, but mine was available at https://cdan.dot.gov/files/files/b14dcfb4-1f45-4eeb-abda-744... when I made this comment.
4. https://en.wikipedia.org/wiki/Tesla_Model_Y#Sales states 172,700 sold in 2021 and 231,400 in 2022. Other estimates put around 80,000 sold in 2020, for a total of 484,100.
5. See page 79 of https://www.tesla.com/ns_videos/2021-tesla-impact-report.pdf
hmm, operation "Trick Elon into Wasting His Money" has published its latest trick!
I don't think this is a tree Elon wants to be barking at too loudly. But then again, you never know, every time I read the news I can't tell whether I'm reading The Onion or not.
Look at that list of brands though. You'd have to have to be blinded by motivated reasoning to not think that who buys what might have more than a little bit to do with it.
Edit: Just to be clear, I say this not in defense of any particular brand but in offense of anyone who engages in naive surface level assessment.
I'm not at all surprised if a Kia Rio is a death trap in an accident facing most likely an SUV twice the weight. I am surprised to see Tesla at the top of the fatalities chart though, punching a few weight categories above your average Kia.
And certainly Tesla's aren't cheap right?
People who buy Tesla sedans are basically the "german car weaving with no blinker" demographics of yesteryear so if someone's gonna hit a pole at 100 it'll be them.
Buick sells pretty safe stuff to pretty conservative buyers buy they skew toward age where they'd keel over from an open hand slap so surviving more than a fender bender isn't likely and they're exactly the kind of people who are gonna get t-boned at 60 by something they just didn't see.
Dodge basically sells Challengers and Chargers these days (Ram is separate brand for the timeline listed) and the stereotype seems to generally check out.
Hyundai and Kia are kind of the odd ones out but they sell a lot of low end and small stuff which doesn't exactly attract the least risky buyer demographics and the cars aren't exactly loaded with safety themselves.
The comparison really needs to be done on a higher resolution otherwise you get stuff like Volvo and Lexus looking artificially good because of course nobody dies in high end SUVs and sedans driven mostly by people of non-risky age and decent means and Ford and GM look good because they sell god knows how many fleet vehicles that only get driven responsibly on the clock and it's kind of hard to kill yourself in a 3/4 ton pickup anyway.
Looking at real world results is still important. The BMW F10 530i had zero worldwide fatalities over its entire production. Results like that should speak more than contrived sled tests.
After all, people buying a car already have a strong incentive to purchase something that is safe for themselves, but how many people put any thought into reducing the risk that the force upon others?
For context, pedestrian and cycling deaths have increased for the past decade in North America [0] and it is known that tall blunt hoods increase fatalities [1]. Yet, nothing is being done about it in NA as far as I know.
[0] https://www.npr.org/2023/06/26/1184034017/us-pedestrian-deat...
NCAP has a dedicated test battery.
Pedestrian figures are alarming, but I doubt vehicle mass and shape are the largest factor contributing to their rise. The numbers have increased since 2009. Smart phones and distracted driving are a huge factor here. That combined with NA’s terrible pedestrian infrastructure is a terrible combination.
The fastest way to curtail pedestrian deaths is to build real pedestrian infrastructure and get serious about anti-cell phone technology for the driver’s seat.
Edit: Oh, also several NA cities just stopped enforcing all traffic laws post summer 2020. Can’t ignore that in the post pandemic spike.
Distracted driving might be the biggest factor but it's hard to prevent a human from being distracted. It's easy to make smaller cars, we used to have them.
The shape most definitely has a large impact (pun intended). The visibility from recent SUVs and trucks is abysmal without extensive assistance, like sensors and cameras. [0] Sometimes those cameras and sensors are another source of distraction and increase the cognitive load for what used to be "just look ahead", leading to the distracted driving.
> examined that front visibility with a group of elementary school children, ages 6 to 10, and several adults of different heights in the driver’s seat of four tall, square-hooded vehicles: Ford F-150 and Toyota Tundra pickup trucks and Cadillac Escalade and Jeep Wagoneer SUVs. With the kids seated in a line stretching forward from the vehicle’s front bumper, it took nine to 11 of them before a 5-foot-2 driver could see a child’s head
[0] https://www.nbcnews.com/news/us-news/americas-cars-trucks-ar...
Straight ahead drive overs of unseen pedestrians are vanishingly rare compared to people cornering into pedestrians that are hidden behind pillars. From there the lethality of the vehicle takes over.
The largest of SUVs that you complain about are not increasing in frequency. What is happening is that sedans re being traded in for taller SUVs of comparable footprint and that's where the changes in statistics are coming from.
I'm not sure what's misleading, it's based on data which contradicts your opinion. Every link I can find on this points in the same direction.
As as summary: cars have increase continuously over the past decades and so have their blind spots, between 2012 and 2021 the total number of large cars has increased by 50% and each segment of large cars has seen 25-100% increase, the design can make even smaller cars more deadly, and the number of pedestrian deaths increased by 80% since 2009. [0][1]
> America's cars and trucks are getting bigger, and so are their front blind zones [0] [look for "Getting Bigger"]
> Over the past 30 years, the average U.S. passenger vehicle has gotten about 4 inches wider, 10 inches longer, 8 inches taller and 1,000 pounds heavier [1]
> a blunt profile makes medium-height vehicles deadly too [1]
> Pedestrian crash deaths have risen 80 percent since hitting their low in 2009 [1]
> many safety advocates have also drawn a connection to the growing portion of the U.S. vehicle fleet made up of pickups and SUVs [1]
You call it misleading, I call it supported by data.
[0] https://www.nbcnews.com/news/us-news/americas-cars-trucks-ar...
[1] https://www.iihs.org/news/detail/vehicles-with-higher-more-v...
Deaths are also about what happens when there is contact: High hoods lead to the unfortunate pedestrian hitting their head against the pavement, followed by the car trying to go over it with all its weight. Do the same with the very heavy, but low Tesla Model 3, and the pedestrian goes over the car, which happens to be much safer. The contact with the sedan is at around knee size, many a modern American truck will hit your ribcage, which is a bit more threatening.
A car can be tall and designed to be relatively safe on impact, but it'd look like a minivan. We all know that those kept losing in the marketplace because they didn't look aggressive enough.
So the switch from sedans to small SUVs was a disaster.
To prevent it entirely, perhaps. But there's quite a few things we could do to reduce the incidence of distracted driving:
* More thoroughly educate drivers on just how bad they are at distracted driving. I know that when I ran the red light while talking on the phone, I forswore any future use of the phone while driving, because I clearly couldn't be trusted to drive safely while doing so.
* Rip out the touchscreens of modern cars, and stop providing stuff in ways that requires distracted driving to operate.
* Laws against distracted driving can be more rigorously enforced by the police.
* Penalties could also be harsher. Drunk driving? You can lose your license. Distracted driving? Here's a $500 fine. Very light penalty for the crime that kills more people.
Places like school parking lots are probably where front and rear visibility are of upmost importance, and that's where I think all of the sensors and cameras are critical and should be mandatory. For example, while backing up out of a parking space a rear sensor can detect cross traffic (people or cars) way before you can see it regardless of the vehicle size. My old Mazda Miata would probably fair far worse than my Telluride as rear visibility was poor and the car was so small and low to the ground it was hard to see over other cars.
How do you know? You can't know what you didn't see, and when you run over a squirrel there's no mother squirrel screaming her lungs out to stop your car.
I'm not sure what we're trying to debate here. I agreed front visibility matters but sensors and cameras are also important (the post I was replying to seemed to indicate that we should go back to a time prior to relying on sensors and cameras).
That's the bias. It could very well mean "no problem spotting the ones I had no problems spotting". By definition you can't know how many you never spotted and maybe even ended up under your wheels.
Most drivers only kill a pedestrian once in their life, and relatively few actually are that unfortunate, so it's not something you can easily calibrate for from personal experience (with any luck you never will). They all saw and avoided all the pedestrians that jumped in front of them until they didn't.
Many of these accidents happen at low speed when pedestrians thinking they have time meet a driver with a giant blindspot right in front of them and doesn't even try to brake.
Search for "frontover" pictures and you'll realize how ridiculously large that blindspot is.
How does this work?
I don’t want my car to be an arm of the nanny state. I don’t want insurance companies spying on my drives or owning my driver monitoring data. But, there could be a simple “place the connected device on this pad and strap it in while you drive” type system, that actually attempts to enforce the policy. Does it get hard once multiple people/devices get in the car? Sure. Could individual drivers just bring a second device along to defeat it? Yes. But for the average solo driver that doesn’t pre mediate their own misbehavior, it could reduce distraction without trampling on privacy or significantly added nuisance.
Why do I have to comply? Is my phone not allowed to be used while in motion or something? Or is this required just to access the infotainment system (hello, car-only phone!)?
If you're putting some kind of radio in my vehicle that tells my (and every passenger's phone) to shut off, I'm going to abuse this signal and broadcast it far more widely than any car would. And then just remove, disable, or block the radio.
> Does it get hard once multiple people/devices get in the car?
It doesn't just become difficult, it becomes logically impossible. Current, common radio technology isn't good enough to isolate the driver's seat.
---
Also: good luck getting any technical chain of trust to work and stay safe/secret between numerous phone and automobile manufacturers. Nobody has been able to achieve this with multi-manufacturer DRM so far - DVD, Bluray, etc. Apple is the exception with regards to key material security: auto manufacturers stand no chance, Xiaomi doesn't care, and my Google Pixel is rooted.
Before phones there were DVD playing head units in cars: parking and GPS bypasses have existed for these for at least 20 years.
Look man, you don’t know what you’re talking about. UWB can easily position a device in a vehicle. BT direction finding is also quite promising.
But I’m not suggesting any advanced locating, let alone a jammer (lol). I’m literally just suggesting an NFC pad with a phone “seat belt” for the primary device connected to the infotainment. Car makes a beeping noise until you rest the phone in the safe spot. That’s it. Nothing any more extreme than existing seat belt and speed limit warnings (which save lives despite their nuisance).
t. Actual automotive engineer for a major OEM that works on infotainment & mobile software, Bluetooth, UWB, & future RF
Why would I pair it?
Do you want a pat on the back for defeating the imagined safety mechanism? You can also buy seat belt plugs to turn the seat belt chime off. Very few people end up doing this in practice (because it’s stupid).
What is your specific aversion to your car expecting you to separate control of your phone while driving alone? Assume you have HFP, A2DP, and/or CarPlay/Android Auto with voice control. Why do you need physical control of your phone while your car is in gear/moving? You don’t.
Mandate driver attention monitoring, and require that prolonged inattention (i.e. N seconds continuous, or N seconds of the past 3 minutes) results in escalating consequences, e.g. a warning chime, followed by activation of the hazard lights, followed by cutting the throttle and engaging lane keeping assist if available, and with several minutes continuous inattention, automatically call emergency services if possible.
Ideally, phone manufacturers and car makers could work together so that the driver inattention chime automatically locks the phone and tells the driver to pay attention (or tells the passenger to tell the driver to pay attention).
Why not ban all old phones and cars too, to really ensure enforcement of this scheme?
(Very confusing the first time, when you don't realize that's what the issue is.)
I'd imagine something similar is at least possible.
If you've ever used an Apple AirTag, they're even better for this; my phone could easily tell me which seat an AirTag is sitting in.
If the US population has doubled since 1980, then it wouldn't be surprising that some population-level statistic (total pedestrian deaths) is at an all-time high! And it could still be at an all-time high while reflecting a substantial increase in safety - if the population doubled (increased 100%) but deaths only rose 25%, for instance.
Similarly, I anecdotally observe that cycling is more popular than it was 20 years ago. So it would not be surprising (to me) to observe that cycling deaths are higher as well.
But without quantifying any of these factors, the significance of these statistics is difficult to evaluate.
Indeed, it is per capita deaths that have increased since 2013 or so [0].
Why speculate when you can Google?
[0] https://www.iihs.org/topics/fatality-statistics/detail/pedes...
Nevertheless, in the same amount of time that it took them to write a shallow dismissal they could have googled the raw data and shared a link. They didn't provide any sources whatsoever for their own claims.
Just because that information was easy to find this time, doesn't mean it is every time. Often when I go looking for statistics on various population level things, I have to do a lot of work to find and massage it into the form I was looking for.
As far as I am concerned you are conceding the argument by doubling down here. And in spite of potentially being right.
I can only speak for myself but when I’m driving, cycling or walking I’m always on the alert for idiots, there are plenty around. In almost all contexts it is people distracted by music and phones and are not paying attention to their surroundings.
And I think cyclists who don’t already have a driving license should have to do a basic provisional theory test. It’s for their own safety to understand the rules of the road.
Roads in the US are designed almost entirely around the speed and convenience of cars, and don't account for the externalities they impose on everyone else. As cars get larger and more dangerous, and as drivers get more careless, the cost borne by society is only going up. Asking people walking and riding bikes to be even more extra careful not to get killed is not the right solution. Changing the design of our roads and public spaces to make them safer for everyone is.
When driving I’ve had idiots on e-scooters dressed in black with no lights zooming towards me on a one way street at night, I’ve had people walk out in front of me when distracted by their phones. When cycling I’ve often seen other casual cyclists with headphones on, in their own dreamland, no helmet either, and I’ve seen the same when walking about. People are too careless and distracted thinking others are going to look out for them, everyone needs to pay more attention.
I had the following near misses where I would have died or been severely injured if I hadn't been alert: a Tesla coming out of an alley (driver was on his phone, never saw me), old diesel Mercedes running a stop sign (couldn't see the driver), a Ford F150 in a parking lot (guy was fixated on a spot that just opened up).
This does not include the woman in the Lexus who intentionally crawled up on me because I had the temerity to be in the crosswalk when she wanted to be someplace.
Stay safe!
We need to nurture a culture of competency on the roads and excellence in automobile operations. My personal fav idea to help with that is build more race tracks :)
I thought this was a one-off, but I see probably 5-10 of these idiots a week.
If you fall asleep while driving, you’ll just ruin some corn/wheat crops until you get stuck in a rut or get yourself back on the road.
Tried to google for the longest distance anyone drove asleep at the wheel but only getting results about semi-conscious sleepdriving, but I bet it’s quite a distance.
It seems like the general public all agrees that it would be insane to allow arbitrarily powerful e-bikes, and so they are limited to (depending on your jurisdiction) 500 watts and 32km/h. This, despite that there are nearly 0 deaths from e-bikes per year. Cars and motorcycles on the other hand, are orders of magnitude more dangerous and we have no restrictions whatsoever. There are street-legal cars with 1000 horsepower and top speeds of over 400km/h.
Allow anyone to own a sedan or compact SUV with up to 200hp. If you want to own something larger, you need an “oversized vehicle” endorsement. And if you want to own something more powerful, you need a power-level endorsement. And maybe you need to pay extra to have those endorsements, and they can be taken away if you are reckless.
It could be a grandfathering system too, so not to disrupt the entire market. Automatically grant partial endorsements to people who already have those vehicles when the law goes into effect, but require testing within 5 years or when registering any new vehicle that meets those criteria.
But cars and trucks are incredibly dangerous and it’s ridiculous society has basically zero restrictions. (I own a “small” truck and would be happy to have it restricted if it means the roads start to become safer for cycling and walking)
Basic physics and historical collision data collected over decades.
But don't take my word for it! Look up your local official collision data -- how many pedestrians have killed people inside a four-wheel vehicle? What about the other way around? Now repeat for cyclists, mopeds/motorcycles, and other four-wheeled vehicles.
The heavier vehicle kills the lighter vehicle, primarily because of momentum.
It's not the vehicle but the entitled douche bag who thinks they own the road while wildly speeding around, cutting people off and weaving around traffic like a drunkard. No one needs that power save for insecure men.
...or you could look at what is done in countries with a far better safety record and copy it. Hint: street design has a massive effect on the speed at which drivers choose to go. Narrower lanes, curves, intersections with tight corners, raised crossings, speed humps, removing on-street parking next to intersections (daylighting), etc. all reduce fatalities.
Many traffic safety measures cost very little to implement, or even retrofit. You can look up what has been done in Hoboken, NJ -- no pedestrian fatalities since 2017.
In Canada, a standard car license allows driving vehicles up to 11,000 kg (24,250 lbs) [0], including pickup trucks and SUVs that are larger than a WWII Sherman tank [1], such as the Ford F-350 or the Cadillac Escalade [2].
Thus, we are allowing people to drive things that are far more powerful than a car without special licensing requirements -- and without a speed governor.
So it is already pretty lax in terms of licensing requirements for vehicles that kill and maim our neighbors every year. We should be even stricter than we are.
[0] https://drivetest.ca/licences/licences-overview/
[1] https://en.wikipedia.org/wiki/M4_Sherman
[2] https://www.carsized.com/en/cars/compare/cadillac-escalade-2...
ChatGPT-produced summary for the lazy:
+----------------------+----------------------+----------------------+
| Feature | M4 Sherman Tank | Ford F-350 Crew Cab |
+----------------------+----------------------+----------------------+
| Mass | 66,800–84,000 lb | ~7,000–9,000 lb |
| | (30.3–38.1 tonnes) | (3.2–4.1 tonnes) |
+----------------------+----------------------+----------------------+
| Length | 19 ft 2 in–20 ft 7 in| 20 ft 11 in–23 ft 11 in |
| | (5.84–6.27 m) | (6.27–7.28 m) |
+----------------------+----------------------+----------------------+
| Width | 8 ft 7 in–9 ft 10 in | 6 ft 7 in–7 ft 11 in |
| | (2.62–3.00 m) | (2.03–2.43 m) |
+----------------------+----------------------+----------------------+
| Height | 9 ft 0 in–9 ft 9 in | 6 ft 4 in–6 ft 9 in |
| | (2.74–2.97 m) | (1.93–2.06 m) |
+----------------------+----------------------+----------------------+
| Engine | Gasoline/Diesel: | Gasoline/Diesel: |
| | 350–450 hp (260–336 kW)| 385–475 hp (287–354 kW)|
+----------------------+----------------------+----------------------+
| Power/Weight | 10.46–13.49 hp/ton | ~43–53 hp/ton |
| | (8.60–11.09 kW/t) | (36–44 kW/t) |
+----------------------+----------------------+----------------------+These are not really comparable vehicles anymore than a wheelbarrow and a sailboat are comparable. Sure you can compare them but there is no useful purpose to doing so. You are actively being dishonest and misleading by trying to push this comparison.
https://www.iihs.org/topics/auto-insurance/insurance-losses-...
Because when I compare between classes, there is a clear trend in the data you provided where large four-door cars have higher rates of bodily injury than midsize cars, and it goes down again for small and mini.
And it makes sense: in a collision, the heavier vehicle crushes the lighter one.
Am I misreading the data?
Edit: though interestingly there's an uptick when you get to the ultra-ostentatious vehicles like Bentley and Lamborghini.
If you bring this up, the amount of visceral reactions, you get from people who demand that you stop talking about the subject is incredible.
They come up with all sorts of crazy hypotheticals (what if I need to rush to the hospital at 105 miles an hour).
Being able to drive what you want, where you want, when you want, and how you want, within reason, is one of the emergent freedoms ever granted by the progress of civilization.
How many people are exceeding posted speed limits in highly populated areas? That is what increases collisions with pedestrians and cyclists, and for that reason I would love to see speed governors installed in all motor vehicles, and not only e-bikes, which ironically are the ones that are the least likely to cause serious injury to others.
Not being killed is also a very popular visceral reason for pedestrians and cyclists to want cars to slow down in our cities, and for motor vehicle regulations to address the yearly carnage they cause every year.
What value should we give to their desire of not being killed, and how should it be weighted against drivers' desire to drive what they want, where they want, when they want, and how they want?
Just a fun experiment: try replacing "driving a car" with "stabbing random people on the street" or any other activity that involves people killing around three people per 100,000 population every year in your country. It's interesting how our perception changes when we get out of the context of motonormativity.
I don't want the independent kinds of safety rating, like what's included in the article to include anything other than what benefits the purchaser of the car.
I am in an adjacent industry where it's getting increasingly common to hear that these ratings are becoming less trustworthy, and part of it is they include points for gadgets and things seen as irrelevant to the end user.
You may not want independent safety ratings to take pedestrians into account, but your insurance company may have other ideas, since pedestrian safety matters to their bottom line. Whose interests do you think matter more to an entity like IIHS?
I want to avoid a case where people return to "just buy the biggest x, x make safe cars".
> As we've seen over the past few years, loudly proclaiming something, regardless of whether or not it's true, even when there's incontrovertible evidence that it's untrue, seems to not only work, that kind of bombastic rhetoric appears to attract superfans who will aggressively defend the brand.
What does by Dann Luu mean by "loudly proclaiming something, regardless of truth, [seems to work]"? Let's try to make the claim more precise and testable.
Since I'm interested in the broader claim, let me "carve out" (put aside) Dan's specific point about attracting superfans, which I find persuasive. With that said, here are some attempts to reframe the broad claim:
Reframing #1: "Some people are persuaded of false things when a company loudly proclaims them." Yes, this seems obvious. But it isn't particularly useful in guiding action. The claim doesn't say anything quantitative.
Reframing #2: "On net, the accuracy of a population's belief goes down when a company loudly proclaims a false belief." Maybe, maybe not. I would expect it largely depends on what happens next; i.e. whether there is some kind of response. Here are two kinds of reactions that favor not lying. First, getting caught in a lie can damage brand reputation. Second, lying calls more attention to an issue which would have otherwise drifted out of public awareness as the news cycle churns. These depend on how much credibility people assign to the organization and their claims.
Reframing #3: "On net, it is in an organization's self-interest to loudly proclaim a falsehood about its quality." Maybe. It depends on many factors. This is the most interesting public policy question! If such a tactic works (see reframing #2 above), it is clearly a negative externality (a downstream effect than an actor does not feel directly).
How does wise public policy address externalities? One classic way is to internalize the cost -- by making the actor feel the pain themself. In #3 above, one way to internalize the externality would be to impose some penalty or cost for lying. How to do this in a targeted, effective, non-delayed, and legal way is non-obvious to me.
By non-delayed I mean temporally-immediate consequences provide better signal to correct a decision. This isn't just human nature; it is fundamental to the credit-assignment problem in reinforcement learning.
The problem of delayed feedback goes deeper than just credit-assignment; it has to do with fairness too. Often, the decision-maker isn't a monolith: the leadership composition changes over time. It may be impractical to punish one cog in the system without imposing a cost on the rest of the system, many of whom had nothing to do with the decision.