[tomkarlo]

The solution isn't "lossy" enforcement, at least in the case of speed limits. It's to revise the speed limit to a higher, more realistic figure and then enforce that limit as rigorously as practical. This is what the UK and other countries are doing with average speed cameras.

It's been shown that in some cases US speed limits make roads less safe and are lower than the speed limits recommended by the engineers who designed the roadway.

"The design speed for the project was 110 km/h (68 mph). The design speed is like a warranty: nothing in the road design requires a driver to go slower than 68 mph, not even on a wet road at night (the design conditions).

The average speed is not far from the design speed. The 85th percentile speed, which is supposed to be used for setting speed limits, is around 75 mph. A little over by my measurement, which found 1% compliance with the speed limit."

http://marginalrevolution.com/marginalrevolution/2011/09/be-...

Having speed limits where a large percentage of drivers regularly violate them teaches people to break the law casually. (Similarly, COPPA basically teaches kids to lie about their age online.)

[jerf]

"Having speed limits where a large percentage of drivers regularly violate them teaches people to break the law casually. (Similarly, COPPA basically teaches kids to lie about their age online.)"

I'd suggest that there's also a widespread lack of understanding that there's a difference between having a law, enforcing a law, and potentially even having the capability of enforcing the law. It's part of the reason why "law" isn't the answer to every societal problem.

[krichman]

> It's part of the reason why "law" isn't the answer to every societal problem.

I sure wish more people like you had stronger control over the legislative process.

[Vivtek]

You appear to be suffering from the notion that there is a single defined value at which the speed limit will be optimum at all times.

But I entirely agree with your larger point that violation of arbitrary speed limits because of obvious engineering considerations does train people to break the law casually.

[tomkarlo]

The quote I posted explicitly points out that the speed limit is based on worst-case road conditions. Obviously a single-value speed limit can't be optimal in all conditions, for all drivers, in all vehicles. It has to be based on what's safe for most vehicles in most conditions, but there is a single optimal value for those constraints on a particular road.

I'd have no problem with variable speed limits on roads - I think it would make a lot of sense to have speed limit displays that change based on time of day, weather conditions, etc. But until that's practical, speed limits will be set based on worst-case, with a proviso that drivers ALSO have to exercise discretion and further reduce speed in exceptional conditions like fog or snow.

[kvb]

I don't think the data on optimal speeds is that clear cut. See another Marginal Revolution post for a summary of a study indicating that raising the limits generally would have costs outweighing the benefits: http://marginalrevolution.com/marginalrevolution/2011/12/sho....

[AnthonyMouse]

The conclusions are based on the idea that higher speeds create external costs to third parties which aren't required to be internalized by the driver, making the driver more likely to choose a higher speed than would otherwise be optimal.

Which means the conclusion is wrong. The solution is not to have a lower speed limit, it's to require the internalization of externalized costs: Impose a fuel tax that accounts for the full cost of the pollution created to society. Require drivers whose speed has caused a collision to pay higher insurance premiums. Then drivers will have to weigh the cost to others in their decisions and the identified problem goes away.

[tptacek]

Did you read the actual paper, or just Tyler Cowen's blurb? The paper is based on actual data, not just suppositions. It too might be wrong, but the arguments it makes are more compelling than anything in your comment.

[AnthonyMouse]

The "actual data" is just as problematic, because they took it from the speed limit increases in 1987 and 1996. Fast forward two or three decades and we have much safer cars, which will completely change the injury data. Meanwhile modern cars are significantly more efficient and have far better pollution control systems. On page 4 of the paper Benthem writes, "Perhaps surprisingly, the costs from pollution-induced adverse health impacts are about as large as the costs from traffic fatalities." (Which is astounding considering the number of traffic fatalities.) But now we have gasoline powered cars that produce exhaust which is cleaner than the ambient air taken into the engine in certain cities. We have fully electric cars. If your concern is pollution then you would do better with legislative efforts to get older cars off the road and eliminating their pollution entirely rather than only reducing it, which would also save you from unnecessarily slowing down newer vehicles that cause little to no impact on air quality.

[tptacek]

According to NHTSA, the sharpest decline in fatality rate / 100mm VMT preceded 1984, and since 1996 has dropped by something like 6%. What mainstream automotive advances lead you to believe that cars are so much safer that van Benthem's numbers are wildly off?

Similarly: the 1990s dominate van Bentham's numbers and correspond to the introduction of OBD-x. Cars are more efficient now, but by how much? There are hybrids on the roads, but most people don't drive them, and we won't be outlawing Volvos from 2000 any time soon.

[AnthonyMouse]

>According to NHTSA, the sharpest decline in fatality rate / 100mm VMT preceded 1984, and since 1996 has dropped by something like 6%.

http://www-fars.nhtsa.dot.gov/Main/index.aspx

  1995 fatalities per 100M VMT: 1.73
  1996 fatalities per 100M VMT: 1.69
  ...
  2010 fatalities per 100M VMT: 1.11
  2011 fatalities per 100M VMT: 1.10

That's a fair bit more than 6%. And that's the rate notwithstanding the increases in the speed limit that have occurred during that time.

>What mainstream automotive advances lead you to believe that cars are so much safer that van Benthem's numbers are wildly off?

Crumple zones, passenger safety cage, airbags, anti-lock brakes. Better handling to avoid accidents. These things existed in 1996 but weren't present on much of the installed base of cars driven in that year. They were even less common in 1987. And that's just the cars. Light trucks and SUVs were deathtraps on wheels through the 1990s.

>the 1990s dominate van Bentham's numbers and correspond to the introduction of OBD-x.

That's what I'm saying. The new emissions controls had only just been introduced and weren't present on the installed base of cars that were driven on the roads in the years the speed limits were changed.

>There are hybrids on the roads, but most people don't drive them, and we won't be outlawing Volvos from 2000 any time soon.

The problem isn't Volvos from 2000, it's Impalas and Broncos from 1990. Which there were a lot more of on the road in 1996 than there are in 2013. And, once again, if your concern is air pollution, you do better to remove polluting vehicles from the road entirely and replace them with vehicles with modern emissions controls than you do by continuing to operate them at scale and then trying to mitigate the environmental damage of that choice by slowing everybody down.

[UVB-76]

Average speed cameras are actually quite scary. In the UK they have only been used (as far as I'm aware) to enforce temporary speed limit reductions, most typically during roadworks, then removed once the speed limit is returned to normal.

In theory they could be deployed across the motorway network with ease, and overnight force people to abide by the maximum speed limits (which a majority of people do not currently do) or face fines and points on their licence (ultimately leading to a ban)

[timthorn]

No, they're deployed as regular speed cameras too. The M25 cameras are linked to a variable speed limit system to dynamically manage traffic flow; not sure if that fits in your definition of temporary (the cameras are permanent but the threshold speed changes) but there are also many instances of fixed speed cameras both on the trunk road network (eg the A14 near Cambridge) and even in villages (eg in Nottinghamshire)