[nshepperd]

You're basically saying that everyone should defect because everyone else does. Which may well be true for a single human driver, but is probably not how autonomous cars as a whole should be programmed.

Nor does it support the claim that "speed itself is not dangerous". If everyone were driving slower, everyone would be safer (which we may actually be able to achieve once a sufficient number of law-abiding robot cars displace human drivers).

[Dylan16807]

>If everyone were driving slower, everyone would be safer

But that's always true. All the way down to 10mph speed limits.

The job of speed limits is not maximum safety at the cost of everything else.

In practice, people can tell the design speed of a road and mostly follow that, no matter what the signs say. Defecting or not doesn't matter very much.

[TeMPOraL]

> In practice, people can tell the design speed of a road and mostly follow that, no matter what the signs say. Defecting or not doesn't matter very much.

It does, because what people can tell is only the design of the stretch of the road they're at. They have no awareness of how it interacts with rest of the roads in the area. By ignoring the rules they make it significantly harder to optimize traffic flow globally. And even if the limits were wrong, it's still better to have bad rules that are actually followed, because then people responsible can actually observe they're bad and change them for the better.

[ubernostrum]

By definition, the design speed of a segment of road comes from taking into account the entire segment. If a segment contains sharp curves, for example, then the entire segment can be designed around that fact.

[Robin_Message]

> In practice, people can tell the design speed of a road

I think there have been quite a few studies showing most people think they are above average at driving. I don't think people do this at all. I think they are overconfident, hurrying apes, with no real conception of stopping distance or kinetic energy, aiming to just miss each other with these multi-ton projectiles. It's sadly not surprising how many people are killed on the roads.

[Pyxl101]

It's not about some particular kind of skill-based estimate of the road's design speed. It's much simpler than that: people drive at the fastest speed at which they feel safe.

If people are driving on a large separated highway with big lanes, plenty of light, and it's straight with no curves, and no traffic for miles, then they might feel safe going 80 or 100 mph if their vehicle can handle it. People on a small, narrow, windy street with pedestrians around will drive another speed. The point is that people make these judgment calls primarily based on instinct, rather than based on signs.

Imagine that you were walking down a path on foot, and one sign said the walking path was supposed to be 1 mph, and another said 2 mph, and another said 3 mph. Would those signs mean much to you, or would you trust your two feet, sense of balance, and situational awareness to find the right speed? People do the same in cars. Dylan16807 is right.

Most people do, anyway. A small subset of people, I believe it's about 5%, follow the sign regardless. These people create traffic problems since they move at a different speed than the rest of traffic. I read a good article about how traffic planners are re-evaluating the idea that slow speed limits are a good thing or beneficial for safety. The conclusion was that slower speed limits do not actually benefit safety; if the posted speed limit is out of alignment with the road, and whether people feel safe, it creates more problems than it helps. The takeaway from the research was that if traffic planners want people to drive more slowly, then they need to create smaller, narrower roads.

[TeMPOraL]

> Would those signs mean much to you, or would you trust your two feet, sense of balance, and situational awareness to find the right speed? People do the same in cars.

Except people don't have speedometers, and cars do. They are there to be used.

> A small subset of people, I believe it's about 5%, follow the sign regardless. These people create traffic problems since they move at a different speed than the rest of traffic. I read a good article about how traffic planners are re-evaluating the idea that slow speed limits are a good thing or beneficial for safety. (...) The takeaway from the research was that if traffic planners want people to drive more slowly, then they need to create smaller, narrower roads.

Society is fixed, biology (or in this case: external conditions) is mutable. It's a sad but I guess unavoidable outcome - since people under given conditions behave predictably like morons, it makes no sense to ask them to be responsible, so let's redesign the roads instead.

[terryf]

> It's sadly not surprising how many people are killed on the roads.

In 2013, US, total number of traffic deaths was 32719. Total number of miles driven: 2,946,000,000,000.

Miles driven per one death: 90,039,426

It's not even one-in-a-million chance of dying. It's one-in-90-million chance. That is really very good odds. Driving is really quite safe and getting safer all the time. Cars manufactured today have safety systems and braking ability, traction control and other safety measures that are basically incomparable to cars made 20-30 years ago. The speed limits are still the same for everyone though. The whole point of a car is to quickly get from one location to another.

[TeMPOraL]

> It's not even one-in-a-million chance of dying. It's one-in-90-million chance. That is really very good odds.

For you individually. Multiply that by the amount of drivers and the average amount of miles driven, and you get multiple deaths per day - almost one hundred a day, actually, in 2013.

The individual odds make drivers feel safe - and thus behave like idiots - and the result is tens of thousands of unnecessary deaths annually.

[sokoloff]

> It's not even one-in-a-million chance of dying. It's one-in-90-million chance.

How many trips are only 1 mile in length?

[ubernostrum]

Being able to sense the design speed of a road is completely unrelated to someone's estimation of their driving skills.

The design speed of a road is, in fact, something you're probably only partly consciously aware of as you're driving on it; it's something you get not from a speed-limit sign but from cues like the way curves are built, the length of merging ramps and exit lanes, or how far ahead of something a warning sign is posted. Whether you think you're an above-average driver has nothing whatsoever to do with the fact that you will, consciously or not, learn to pick up on those cues and you will adjust your driving to it unless you're constantly watching your speedometer (and even if you do, if you drive significantly below the design speed, there will be times when you're uncomfortable doing so, though you might not realize why it's making you uncomfortable).

[TeMPOraL]

Exactly. If everyone followed the traffic rules instead of considering themselves smarter than traffic engineers, rules could then be adjusted to accomodate for optimal flow and safety. Alas, drivers know better and you can't optimize anything, because people will ignore the rules anyway.

[ubernostrum]

If everyone followed the traffic rules instead of considering themselves smarter than traffic engineers

Traffic engineers don't set speed limits. Politicians do. If politicians would stop thinking they're smarter than the engineers, the speed limits would match the design and operating speeds of the roads much more often.

[TeMPOraL]

Politicians don't think they're smarter than engineers. They just do what politicians do. Pandering to the public and seeking easy revenue, the latter in this case is possible only because drivers are irresponsible and can be gamed for money by lowering speed limits. Anyway, I wouldn't accuse politicians of being any saner than your average driver when on the road.

[ubernostrum]

You're basically saying that everyone should defect because everyone else does.

So, there are three speeds we need to be concerned with here:

1. Design speed -- this is the speed anticipated and planned for by the engineers who designed the road. It's visible in features like curves, merging areas and so on.

2. Operating speed -- this is the actual speed of traffic on the road once built, and typically is measured as the 85th percentile of observed traffic (i.e., the speed such that 85% of traffic travels at or below that speed).

3. Speed limit -- this is the posted maximum above which vehicles can be stopped and ticketed by an enforcement officer.

In real-world scenarios both the design speed and the operating speed are often higher than the speed limit, because speed limits are often more strongly influenced -- and always downward, when they are influenced in this way -- by factors other than safety (i.e., politics and revenue).

Which is a problem. Good road design anticipates roughly what the operating speed will be, and attempts to match that in the design. Setting a limit which differs significantly from the operating and design speeds is only good for politics ("we're making you safer by slowing down the traffic") and revenue (more tickets issued for speeding); it has no relation to actually-safer roads.

And in general, yes, speed differential is more commonly a danger than simple raw speed; the intuitive explanation is that every event of one vehicle passing another creates an opportunity for a collision, and moving at a speed significantly different from all other traffic (regardless of whether faster or slower) increases the number of passing events which necessarily increases the chance of a collision.

Thus, regardless of posted limits, it is safer to match your speed to that of surrounding traffic. This is not a case of "everyone should defect because everyone else does". It's a case of "the people who posted the speed limit defected, and you shouldn't follow their example".

[TeMPOraL]

You realize that "politics" only works because drivers drive like idiots and generally ignore speed limits? If they were responsible, the operating speed would always be less or equal than speed limit.

[ubernostrum]

As long as ticket revenue goes into the budget of the local government entity issuing the tickets, it doesn't really matter how awful you think drivers are. There are plenty of documented cases of deliberate meddling with safety features to increase revenue. Look at red-light cameras, for example, where local governments have been known to change the light timing in unsafe ways in order to manufacture ticketable violations.

And, again: the safest thing is to keep to the average speed of surrounding traffic. If you deliberately operate your vehicle at a significantly different speed than surrounding traffic, you are deliberately creating a hazard to yourself and others.

[TeMPOraL]

> There are plenty of documented cases of deliberate meddling with safety features to increase revenue. Look at red-light cameras, for example, where local governments have been known to change the light timing in unsafe ways in order to manufacture ticketable violations.

Didn't know that, thanks. If you have any links to some well-documented cases of meddling with safety features like that, I'd be glad to read.

> the safest thing is to keep to the average speed of surrounding traffic.

Safest for drivers, not for pedestrians; but assuming pedestrian-free area, then ok, that is of course true for the reasons you describe (basically introducing any unexpected element on the road is making the situation less safe). But since many (AFAIR most) traffic-related injuries and deaths are caused by speeding, how do you propose we force drivers to slow down? Personally I'm in favour of ALPRs and distance-based speed checks (i.e. distance traveled / travel time > limit == you get ticketed).

That is, if it matters now. Self-driving cars are hopefully around the corner, and they should be able to solve this problem once and for all.

[ubernostrum]

Chattanooga, TN, was ordered to refund 176 fines from a light determined to have an illegally-short yellow phase:

http://www.thenewspaper.com/news/22/2269.asp

This (large) page has a listing of quite a few towns in California where litigation overturned tickets due to illegally-short yellow:

http://www.highwayrobbery.net/redlightcamscameras.htm