When cara is trailing carb at 60ish MPH and carsuddenly brakes, cartraverses 150 feet or so before its driver notices danger and stamps on the brakes, and an additional 150 feet are traversed after the brakes are engaged. (It's more complicated than that and it's more complicated than that, too. If I'm oversimplifying too much at this phase please let me know.) So obviously the stopping power of the vehicle is important. Now, a huge amount of R&D has been done on automobile braking systems. Not only are modern braking systems automatic supersonic hypnotic funky fresh, modern cars can extract modern energy out of them, too. But... well, I'm having a lot of trouble finding credible statistics, but it looks like a large fraction of the victims of fatal car accidents are in vehicles that get rear-ended at high speeds.0 Not only do they cause a lot of immediate deaths, rear-ending accidents also cause a lot of damage to both vehicles and nervous systems (whip-lash), and it's hard to calculate how that compounds over the years, but you know it's a really huge amount of lost QALYs and moneys. What I don't understand is, it naively seems to me that there are many different ways you can get a tailing vehicle to stop faster, ways that don't involve completely hopeless public education drives or expensive 5% improvements on disc brakes. Like, having a special system specifically for applying high friction directly to the road surface in emergencies, either mechanically or via electromagnetism. Or an air brake0.08 or two. Combine those with existing automatic electronic sensor brake activator thingies and you can stop the vehicle almost immediately.1 Wham bam, way less crushed organs and needless suffering. But I never hear anyone talking about this. Is it because modern cars just don't crash into cars in front of them anymore? If so, would it be too expensive to equip older cars with a simple brake-pressure-remotely-activated dedicated stopping mechanism?14 E.g. a government or non-profit program that equipped them free-of-charge on the cars of inexperienced or risk-prone drivers. Or something? I can think of a lot of engineering point and counterpoint that would make it more or less difficult but it still seems feasible, life-saving, and money-saving. What am I missing? What hard steps did I trivialize? I am more interested in automobile engineering steps I naively trivialized, but social engineering steps that I ignored might be more important somehow... what gives?


(This has many connections to both instrumental and epistemic rationality but unfortunately it would be too psychologically difficult for me to point them out. I do not think a meta discussion about this would be profitable, but I may be wrong.)


0 I saw somewhere that most fatal accidents involve only a single car. That agrees with my experience but I remain somewhat skeptical.

0.08 More specifically, some reasonable hybrid of air brake and uber-efficient-mini-parachute. I don't know how negligible the stopping power of air brakes is at freeway speeds.

1 Stopping too fast does indeed hurt the driver but it's rather asymmetric, you can deploy various safety mechanisms in advance unlike the case where you're crashing into an unsuspecting Honda with your Chevy.

14 I don't see why making such add-on systems at least half-automatic should be hard either, it's like 2 cheap cameras and a rudimentary but information-efficient AI... or maybe I just completely underestimate the complexity of those systems.


New Comment
41 comments, sorted by Click to highlight new comments since: Today at 11:10 PM

It's not actually that easy.

Tires can only get so much friction with the road and after that, brakes lock up. Purpose built skids wouldn't be any better. We're easily at that limit. I'm already seeing commercials for cars that hit the brakes for you, but I'm not sure how much time they shave off or what the unintended consequences will be.

If you want to decelerate at more than ~1G you need other ways of doing it. A maximally efficient 10x10 parachute would have to be ~16' diameter to get 1G for a small car at 60mph. At 30mph you're down to a quarter G.

Other methods I can think of would be forward facing rockets (~40lb worth of fuel) with their obvious downsides, an explosively fired anchor and winch system, or a rail on every street everywhere for the brake to grab (like the "driving" ride at disneyland). Nothing easy..

If the obstacle in front of you is a car limited to 1G, then to first order all you have to do is start braking at the same point he started braking. Computers can probably get really good at this. If your problem is that you're driving into trees and poles, then stop driving your car into trees and poles.

[upvoted for being clear, concise, and informative]

To amplify slightly -- Another way to shave off reaction-time is to predict when the car ahead is going to brake. In particular, there's a common case where car A brakes, causing B to brake, causing C to brake. There's a fair bit of ongoing research into so-called open road platooning, the premise of which is that a computer in car A should tell B,C, and succeeding cars "I am about to brake" so that C doesn't need to wait for B to react before beginning to do the same.

It's neat stuff. And given that self-parking and self-driving cars are starting to enter the market, I think we may see it deployed within 20 years. At a guess, commercial freight trucks will be early adopters.

Purpose built skids wouldn't be any better.

This 1971 patent says the inventors tested a device that can stop a car from 50 km/h within 2 meters, which indicates deceleration of about 5g.

ETA: Apparently it has "anti-skid projections" on the bottom, which look like sharp protrusions that can dig into the pavement (presumably damaging it during the stopping process). Is there any reason why this might be physically implausible?

It's certainly plausible to decelerate rapidly by grabbing onto the pavement. You can decelerate even faster by running into a concrete wall. I suspect that 5g is more than the car (or its inhabitants?) can sustain without injury. I'm sure that digging into the pavement won't go over well with the local highway department.

Note also that 50km/hr isn't all that fast -- 31ish MPH I think -- and that kinetic energy goes as the square of speed. So at freeway speeds, you'll do proportionately more pavement damage.

I was skeptical of the injury potential at only 5g, since fighter pilots and F1 drivers routinely experience 3g acceleration; but this study%20933-9000 "pdf warning") indicates that 5g is just over the threshhold for whiplash damage. This may also be due to the unusually good physical conditioning of fighter pilots and F1 drivers; physical conditioning certainly has the possibility of reducing neck damage.

More importantly, a pilot's seat is significantly different from a driver's seat, and they're designed to hold the pilot in a way that limits relative movement of different body parts.

I would guess that the design of the device can be adjusted to give a smaller deceleration if needed to prevent injury. To minimize pavement damage, it's probably a good idea to place the device under the automated control of an emergency stopping system that can detect imminent collisions and apply either the brake or this device depending on the deceleration needed to prevent the collisions.

Suhweet, thank you, that was just what I was looking for. :D

It's also dangerous to have deceleration abilities far in excess of other vehicles, which will rear end you if you panic-stop (all out) when it might have been better to slightly crumple the rear of the car in front of you, or at least come to a stop at the last possible moment. If the extra braking only occurred as necessary to stop you from colliding with an object that would have stopped you anyway (e.g. another car), then I guess I can't object. Of course, it would need to sense the distance to that object in order to do so.

Better stopping seems like it would have great benefit if all vehicles possessed the ability, or if the users of it had training or technological aids to avoid the mistake of prematurely using it and getting unnecessarily rear-ended as a result.

One immediate complication that came to mind is that modern cars are just so jam-packed with stuff already that it would take engineers and design time to figure out where to put such a device or mechanism.

This is just speculation, but it seems like if there were a way to immediately deflate tires, you would increase the surface area touching the road, which could increase friction drastically at the cost of probably destroying your tires. It wouldn't work on an icy road very well, of course.

Edit: This subject matters to me a lot more now than it would have 10 hours ago. Now I have a dilemma about whether to let my girlfriend ever drive my car again. It seems Bayes can help me on this problem:

Girlfriend has driven the car, with me as the passenger a total of about 7 times in the past four years. Girlfriend has been in collisions or crashes a total of 3 times in the past four years (100% of those were with me as passenger) Girlfriend has driven the car hundreds of times to get to work or do shopping or hang out with friends while I sleep or stay at home (most of the driving has been in the past 8 months), with no traffic incidents that I know about. Which I would know pretty quickly, since I regularly inspect the car.

So, P(crash|girlfriend is driving me around) ~ 43% P(crash|no me in car) ~ .5%

It seems obvious I should never let her drive me anywhere again. But should I let her drive by herself? I want to say no, but I also don't want to have to drive her everywhere and cut into my sleep time, which isn't a notion I can easily set as a prior. I am tempted to say "not enough information", maybe LW has some clever advice? To be totally honest, even though it was just over 8 hours ago, I still feel dumb and panicky. Curse you, adrenaline!

So, P(crash|girlfriend is driving me around) ~ 43% P(crash|no me in car) ~ .5%

Only if you assume that your girlfriend driving with you in the card and her driving without you in the car are completely different and one conveys no information about the other.

You're right, I should have said P(crash AND girlfriend driving), not given. Thanks.



[This comment is no longer endorsed by its author]Reply

I recommend also "retracting" the deletion. It removes the voting ability and as such is slightly closer to what a delete should be. That is, when people are voting you down in the temporal or ideological context they cannot no longer downvote the particular comment half-deleted placeholder.

Interesting ... I was pretty sure previous retractions of mine continued to get votes.

Edit: And it looks like you can do a full delete after a retraction! Beware trivial inconveniences...

Ahh! Fantastic. (You need to refresh the page to show the delete feature.)


You can?

[This comment is no longer endorsed by its author]Reply

My intuition is that car drivers don't care about car safety. Sure, when considering the purchase an automobile the buyer may eliminate those models that have a 1 out of 5 safety rating and pay extra attention to the models that are considered high safety but that's just because you purchase fuzzies that way. You don't want to signal that you don't care about safety and buying a safe car feels like the responsible and adult thing to do.

The same people who buy "safe" cars drive when tired and take insufficient breaks, even though study after study shows this can be as dangerous as driving while intoxicated. And they'll happily play with their iPod or make distracting phone calls even though this too is known to be dangerous.

Studies show that after the introduction of the seat belt people started driving more aggressively and kept less distance. If I recall correctly the discovered trend was that when people feel more safe they drive more aggressively to compensate. Likewise, if the car feels unsafe or unstable people anticipate better and drive more attentively. My own experience backs this up insofar my driving style changes the moment I switch to winter tires (it shouldn't change; I should just be happy my car is now safer). Given that driving is one of the most dangerous activities in life we should be extremely cautious, yet we aren't.

We care only about relative car safety. You don't want to drive in a car that's much less safe than the average car on the road, even though driving risk should be compared to other risks in your daily life and not to the driving risks other people take.


I shouldn't have said that people don't care about car safety. People do care (because if you ask them that's what they'd answer) but their actions are inconsistent with that belief. So it's a form of status quo bias. The current situation is seen as perfectly acceptable and slightly safer cars are seen as slightly better and slightly less safe cars are seen as slightly worse.

In the automobile market where only relative safety matters there is no need to make wild innovations. Marginal changes in the look and feel of a car are much more practical (and profitable).

Note: I didn't look for sources. Claims about results from studies are from memory.

My intuition is that car drivers don't care about car safety.

They definitely do care about safety. Manufacturers keep introducing advanced safety features, and automobile safety is increasing quite rapidly. (Deaths per capita have fallen 34% since 1970. This is in spite of the fact that the percentage of drivers on the road per year has tended to increase, and miles per driver generally increases as well. Some of this change may be due to improved medical care, but that certainly can't explain much of the improvement dramatic improvement.)

In addition, brands like Volvo are focused almost entirely on safety. Why would they invest so much in safety unless people cared about safety?

The same people who buy "safe" cars drive when tired and take insufficient breaks, even though study after study shows this can be as dangerous as driving while intoxicated.

Just.... no. This is wrong for a couple reasons. First of all, people who buy cars with a reputation for safety are also much safer drivers! Volvo is once again a good example. I can't find it now, but I recall reading a study that said that a major reason Volvos are statistically much safer vehicles in terms of driver deaths is just because the type of people who drive Volvos are so much more cautious.

They definitely do care about safety. Manufacturers keep introducing advanced safety features, and automobile safety is increasing quite rapidly.

Cars are getting safer but that doesn't mean much. When I say "Care about safety" I mean that when given the choice between a car that has +10 safety and a car that has +10 pretty people would pick the car that's safer. People don't. People pick cars based on looks first, and other attributes such as power, torque, features and safety as secondary concerns. People are embarrassed to drive in an ugly car but are not embarrassed to drive their friends in a pretty but unsafe car.

In addition, brands like Volvo are focused almost entirely on safety

This too is evidence for safety as a secondary priority! If every manufacturer already had safety as their #1 priority you wouldn't be able to advertise with "We care about safety!". You pick a marketing strategy to get that segment of the population that cares specifically about the marketing message you send. Some car manufacturers try to get those people who care specifically about safety, some try to get those people who care about the environment, some try to get those people who care about fuel efficiency. The market is segmented based on second order preferences. Not on the thing people care about most.

This is wrong for a couple reasons

I didn't mean to imply that people who buy safe cars take fewer breaks or drive when tired more compared to people who drive regular cars. I meant it quite literally that even though they buy a safe car they still take risks that are -- compared to other risk they take on a daily basis -- way out of proportion.

And if Volvo drivers indeed drive much safer than other people do this too is evidence for the general disregard for safety by the rest of the population! So again, safety as a secondary concern in practice. Right up there with fuel efficiency, engine noise and the number of cup holders.

I tend to agree with your points. Moreover, caring about whether the car will save your reckless self is not the same thing as caring about safety, when you are the one controlling the car. To borrow from a security industry truism, safety is a process, not a product. Somebody drifting left and right while trying to text a friend is not someone who cares about safety, even if they made extra sure to get a car with side-impact air bags.

/nods.But it was never really consumers who incentivized safety R&D, it was the government. (ETA: As knb pointed out, the previous sentence is obviously completely false.) If there is a relatively easy and beneficially-propagandistic engineering hack that saves many teenage lives per year, why wouldn't governments simply grab it? It seems to me that engineering difficulties are a more parsimonious explanation for the mirage of low-hanging fruit.

ETA: Whoever downvoted this, you're abusing the karma system. Meta: ISTM that karma voting patterns are getting increasingly more ridiculous over time. We might want to think about how to objectively measure this. We might want to think about what it signifies. We might also want to start thinking about how to counteract it if need be.

But it was never really consumers who incentivized safety R&D

This is just wrong. Pretty much every new safety feature which required R&D was introduced on luxury/high end cars on a voluntary basis, as a selling point for the business. Examples off the top of my head, are driver's side airbags, passenger side airbags, airbags that protect the knees, anti-lock brakes, and laminated glass. Usually the governments eventually mandate inclusion on cheap cars, too.

You're right, I used the wrong sentence entirely, I'm not really sure where it came from, though I remember I was thinking about seatbelt laws when I was writing it, and I was thinking about that 'cuz Zed brought up risk homeostasis... that's weird, that's the second time my brain has automatically generated a completely false rationalization out of nowhere in the past month when I was extremely tired. That is effing annoying.

Not seat belts, though.

Actually seat belts do qualify. Ford offered seat belts as options in 1955 and they became standard on new Saabs 3 years later. Seatbelts did not become mandatory anywhere in the world until the 1970s.

On luxury cars first?

I would conclude instead that the seemingly low-hanging fruit is not low-hanging.

Every time you make something that is perceived as a wild innovation you're betting the entire brand of your company on that innovation. The public is completely unforgiving when car manufacturers mess up and the industry compensates accordingly.

Government could enforce it, but which elected official is going to put his weight behind it? There's nothing to be gained and everything to be lost.

I shouldn't have said that people don't care about car safety. People do care (because if you ask them that's what they'd answer) but their actions are inconsistent with that belief. So it's a form of status quo bias. The current situation is seen as perfectly acceptable and slightly safer cars are seen as slightly better and slightly less safe cars are seen as slightly worse.

Perhaps things like playing an ipod in the car, driving tired etc. are due to the bandwagon effect? I think that most people do initially desire vehicle safety, but as you drive more and more, especially when you are driving with other people in the car, you will tend to pick up unsafe driving habits because other people are encouraging this (they do it as well, and nothing bad has happened to them, yet).

Looking at your example of the introduction of seat belts, I'm inclined to think that these people vastly overestimated the added safety of the seat belts because they already considered the car relatively safe after having driven without them for a long enough time. Thus the seat belt meant, to them, that they could take bigger risks and be more aggressive in order to get to their destination faster.

I ceased listening to official numbers on stopping distances when I went out and actually tested them.

Turns out, I could stop with significantly less distance than is physically possible, and so chalked it up as, "this is exaggerated to scare you".

How did you factor in reaction time in a self-experiment? (Or did you just test "distance to stop after brakes engaged"?)

That -- I only compared to the estimated time after you've reacted and engaged the breaks. They typically give both numbers (distance during reaction + distance after breaks are applied). It's still way off. (And the assumed reaction times are extremely conservative too.)

It was probably unfair using a compact car for the tests, but the propaganda gives no such caveats.

Converting these measurements to the metric system would be helpful. Not everyone here is from America.

http://www.google.com/search?q=60+miles+in+km Google does a pretty good job here. The post also doesn't seem to rely on the specific value in any meaningful way to begin with...

New to LessWrong?