[baybal2]

I put to doubt the utility of crash safety ratings.

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.

[volkl48]

> The survival rates for collisions on highway speeds are in single digits no matter what you do.

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.

[baybal2]

For that, sure, but for that even 20 year old cars had good enough structural integrity. By late nineties Japanese cars were already full in into the science of crash safety with concepts like safety volumes, and rigidity belts being fully adopted in mass market cars.

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.

[djaychela]

Sorry, I don't think you're right here.

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).

[ak217]

Great points. I'm not sure how baybal2 arrived at their conclusion. To add to your list, newer cars have NHTSA/IIHS-advised brake assist and obstacle detection, more and better airbags, seat belt load limiters, better use of high-grade structural steel and adhesives in critical joints, and wider tires for better traction.

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.

[baybal2]

I stand my position here.

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.

[djaychela]

Your position seems to have changed - you were initially saying that safety wasn't improving, and cars of 20 years ago were safe enough. Which is patently untrue.

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.

[dsfyu404ed]

20yr ago a 40mph partial overlap would have put whatever you were overlapping with in the foot-well.

[ChickeNES]

Your claim isn't backed up by the data: https://en.wikipedia.org/wiki/Motor_vehicle_fatality_rate_in...

[Tade0]

I drive a 2017 Toyota Auris and (still) own a 00' Corolla.

The difference in the level of safety between these two is patently obvious.

[mywittyname]

Crash safety ratings are based on sound scientific and statistical analysis of real-world crash data and have been proven to save lives.

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).

[Qo2]

I was unable to find a relationship between IIHS safety ratings and IIHS driver death rates for trucks and SUVs based on the most recent set of data released (though this was in 2014, so maybe they have improved their testing methodology since then). Interestingly, the only two vehicles to receive "TSP+" ratings in 2014 were actually among those that driver was most likely to die in - the GMC Terrain and the Chevy Equinox.

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.

[ProfessorLayton]

Most collisions don’t happen at highway speeds and the crash tests are designed accordingly.

[baybal2]

Yes, but collisions below 60km/h got quite survivable back in nineties.

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.

[adamweld]

I think both sides here are right. there have been considerable advances in safety technology in the last 20 years, especially in avoidance systems. these improvements have unfortunately been counteracted by larger and larger vehicle sizes, which make everyone less safe on the road.

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.

[lovehashbrowns]

> 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.

Where are you getting this data from?

[lallysingh]

They encourage spending and R&D in effective safety (vs safety theater).

[693471]

Nearby someone recently died in an accident at an intersection where the speeds don't exceed 30mph. It wasn't head on.

If we can stop these kinds of deaths, we've made incredible progress.