This article ignores a lot of peer-reviewed literature from planning, public health / injury prevention, and civil engineering in favor of some folksy wisdom of a lieutenant cop. A good starting mental model is that vehicle speeds increase both the frequency and severity of crashes.

The article weakly, and quietly made an argument that I do agree with. Namely, roadway design needs to accompany speed limits to ensure people actually drive the desired speeds. Wide lanes, big setbacks, homogenous environments encourage speeding regardless of what is posted. But let's be clear, we should have our design follow our policy, not our policy follow bad designs that were put in place by roadway engineers of yesteryear.

That message seemed to be lost in the race to tell us something tantalizingly counterintuitive. I love "everything we know is wrong" storytelling as much as the next guy, but only when that's actually backed up with solid evidence. Take for example this gem from the article:

> “We all speed, yet months and months usually pass between us seeing a crash,” Lt. Megge tells us when we call to discuss speed limits. “That tells me that most of us are adequate, safe, reasonable drivers. Speeding and traffic safety have a small correlation.”

That would seem to be at odds with this peer reviewed article (sorry, paywall, but you can read the abstract):

> Respectively, they found evidence for an exponential function and a power function between speed and crash rate. Both types of studies found evidence that crash rate increases faster with an increase in speed on minor roads than on major roads. At a more detailed level, lane width, junction density, and traffic flow were found to interact with the speed–crash rate relation.

Link to article: http://www.sciencedirect.com/science/article/pii/S0001457505...

Priceonomics's banner ad tells me it is selling a new book titled "Everything is Bullshit." Quite right.

I found myself disagreeing with the premise that highway speed limits don't influence people's driving speed. I may simply be part of the 15%, but consider the following anecdata: On I-90 in upstate NY, the "speed limit" is 65. I've gone by several state troopers at 70-75 without issue or confrontation. I don't go faster, not because traffic dictates, but I'm concerned about litigious policemen. To put it in Freakonomic terms, I'm going just slow enough that I won't get caught. If the limit was 10 mph higher, I'd drive 10 faster, as would everyone who was on the road for more than one exit (or governed semi tractors).

There is very little crowding on the stretches of highway I'm most familiar with. If the speed limit was 120, I'd probably cruise around 90 or so, assuming it wasn't rush hour. I propose that there is a diminishing impulse for travel rates as they increase, but also that on well-cleared, divided, limited-access highways, we have diminishing utility for speed limits as a regulating concept. Highway driving is the easiest driving to do. There are more variables in control than anywhere else. Why not re-evaluate the roads, not based on the 85th percentile of drivers living under the threat of litigation based on arbitrary bureaucrats, but on simple road capacity and volume?

I studied Structural Engineering in University, which was under the Civil program, and, at least in Canada, we don't design for what people will go, but for how drivers can handle unexpected conditions, like cars turning out in front of them, a loss of traction due to ice, and visibility at night. I've also seen multiple studies that show that during high visibility / traction days (summer during the daytime) the speed limit can be safely raised to at least 130km/h for most highways that are 100km/h.

One interesting thing I observed while driving to university was that there was this one section of highway that was marked at 60km/h, but everybody drove around 95km/h. Every now and then there would be a cop that would give out a bunch of tickets, but it didn't seem to change the speed people went.

Well it turned out that the person that ordered the sign had a dyslexic moment, and the speed limit was later changed to 90km/h. What was interesting was that the design of highway (the guards, the lane width, the shoulder, the banking of the ramps) were all cues for drivers and they drove the actual intended speed without regard to the maximum speed sign.

I think that's neat.

Disclaimer: I am from Germany.

The speeding system we have here seems to make a lot more sense to me: On most highways, there is no speed limit. The only (theoretical) limit is the motor limitation of 155 mph (250 km/h), which can be taken out, but will almost always lead to losing any kind of motor warranty.

In practice, there are limiting signs to 75 mph (120 km/h) for passages where the road is not good enough or there are narrow curves, unclear conditions etc. In my experience, this system works wonderfully: when the road is good, you can drive as fast as you like, so people are more concious about paying attention to the road and their rear mirrors, especially when hogging up the left lane.

I acknowledge that completely overtaking this system would be impossible for the US, simply because the roads are not up to it and the driving tests are not rigorous enough, but I think Germany serves as a fine example that there would be no harm in raising the speed limits a few notches.

The data from Montana always struck me as interesting. It's not an awful lot of data presented in this report [1], but there was a substantial increase in traffic fatalities when Montana shifted from "Reasonable and Prudent" speed limits to posted-and-enforced speed limits.

I then found this report [2] which indicates that yearly fatalities are approximately double of what they were in the late 90s (and is now around the third highest fatality rate in the nation [3]). It's a good guess that traffic volume has gone up, but safety technology has improved as well, so there are a number of variables that have to be accounted for. The differences are certainly stark enough that it merits some consideration, though.

Nationwide, traffic fatality rates have been on a steady downward march over the last couple of decades [4], so Montana's tremendous leap in fatalities is absolutely anomalous. If I was debugging it, I'd start looking at what changed around that time. On the face, it certainly looks like posting and enforcing speed limits made the roads less safe, though.

[1] http://www.hwysafety.com/hwy_montana.htm

[2] http://www.mdt.mt.gov/publications/docs/brochures/safety/hsp...

[3] http://missoulian.com/news/local/report-montana-highway-deat...

[4] http://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_in...

IIRC, Sweden has had a "zero pedestrian deaths" goal for the last 10 years or so. Part of that policy was road design, but another pillar was the reduction of urban speed limits to 18 mph, and vigorous enforcement of such.

The number of accidents did not drop substantially, but what did drop was the severity of the accidents. An accident that would have been a fatality in the past is now a broken leg or two.

There's got to be a term for it, but the factors that influence how fast people drive on a road include how well it's been maintained, how congested it is, but also purely visual factors like whether there are trees or signs or buildings crowding the edges of the roads.

There's also the "They aren't serious about it" effect. Warning signs should be reserved for locations where it really would be dangerous to exceed a certain speed, either because there's an intersection just after the crest of a hill, or a reverse-banked curve, etc. So people have been trained to ignore the warnings because they're so familiar and have proven (most of the time!) to be not all that important.

I'd trade a reasonable (15-20mph) speed limit on city streets if it meant people could blast it on the highway.

It's silly to me that so much of the speed limit conversation is around highways, when really the dangerous streets are the fast non-highways (where you see the unfortunate folks walking in the shoulder.) If anything we should slow those down so people don't go faster than 30mph, if we want them to be safe environments for walking.

Sure, speed up the freeway all you want, but don't conflate the freeway's inhumane paved wasteland with every street.

Their reasoning about the dangers of low speed limits make a lot of sense when you're looking at highways, but I don't think they'd necessarily apply to city or residential streets where there are frequent red lights, pedestrians, etc. Of course, the right thing to do there is probably to configure the roads so that people drive more slowly naturally by narrowing lanes, having trees on the side of the road, etc.

I've lived in the US all my life, with my excessive number of years split between Michigan, North Carolina, and northern and southern California, and in all of those places the proper range of speeds was anywhere from the posted speed limit to a hair under 10 MPH over the posted speed limit (I prefer the latter :). Go slower than that and you start blocking traffic, go faster than that and you might get a ticket.

This is when driving on highways and major roads, at least.

I vividly remember the days of 55 MPH speed limits in the US, and it was pretty much the same then (a hair under 10 over was the optimal speed, but 10 over was a lower speed back then).

Anecdotal experience and all, but at least for me and other drivers that I've observed or talked to about it, just because people tend to go over the limit doesn't mean that the limit has no effect.

This is all fine by me. I only mention it because this is exactly the kind of topic overly logical people (and/or inexperienced drivers) tend to get bent out of shape about. Just because the speed limit isn't strictly enforced doesn't mean it's useless, and just because a cop could theoretically take advantage of people's behavior and give a bunch of tickets to people for going 5 over doesn't mean that it's a conspiracy by the police.

I think that the road design is a key factor in how fast people drive, not limits.

Wide lanes, one-way streets, and cleaned-up sides without obstacles encourage driving faster while narrower passes and trees/bushes on the side make you naturally slow down. On a road or street well designed you would drive at the limit even if you didn't know what the limit is because the limit is set to the "natural" speed of the road. This is what I believe is behind the sergeant's thinking.

On residential streets, a very narrow design would be the first thing to do in order to slow down automotive traffic and make the street more pleasant and safe for pedestrians. There are streets where it's nearly impossible to drive faster than 10 mph, so there are really no practical limits for speed governance in road design.

Conversely, on highways even astonishingly high speeds can be reasonably safe such as on German autobahns. Highways are mostly protected against the worst accidents such as head-on-head collisions. Things usually go wrong when someone disrupts the traffic flow by unattentive, blind lane changes. This is a completely different safety assessment than what's in effect on the streets of populated areas.

The problem with speed limits is that people think they are a license to go that speed, or a little faster, regardless of the conditions. Americans are nearly universally useless at slowing down for things like ice, heavy rain and fog. They ignore things like stopping distance and diminished reaction time in poor visibility.

I agree that there is room for debate on speed limits under ideal conditions on well-maintained roads with bridges that have not been left to rot for decades.

Anyone who has driven internationally will recognize that Americans turn into homicidal/suicidal maniacs when conditions deteriorate--all because they don't understand that the speed limit only applies in ideal conditions.

Driving on U.S. roads in bad weather is a horror show.

I didn't get this from the article, but are there no automated cameras in the USA? In Belgium there are tons (I pass 5 on the 10 minute drive to my girlfriend), but they're a joke. If people know the road they're taking, they drive however fast they like, until they get near a camera. They slow down until they're out of reach and then speed up again. Also only about 1/3 actually have a camera in them, but they get rotated every once in a while.

Actual policemen with a camera are rather rare - considering how much road we have - and they get reported to radio stations. Though they're not allowed to tell listeners where they are exactly, only which road they're on.

Having moved to the US from Germany, it always stroked me as particularly dangerous that it is common to overtake cars on the right and have slow cars and fast cars in any lane. The article mentions that different speeds are dangerous. I think enforcing cars sorting themselves properly into the right lane would make a huge difference in reducing accidents and allowing traffic to o faster. I wish law enforcement could focus on that instead of giving tickets to people who go 5mph/h faster than most other cars.

If there's a state that should win the price for speed limits that make no sense, it must be Hawaii. We have stretches of straight, extremely wide highways with no intersections and a speed limit of 40mph. Then you come to an old, narrow, and curvy road and it's 55mph. Or it could be 35mph. It's like the speed limit is set by a random number generator.

There is very little in this article about the interests of anyone not driving— crash severity is due to speed deltas, and pedestrians and cyclists are almost always moving much slower than automobiles. And a self-interested driver won't slow down to keep pedestrians safe, since pretty all of the consequences of a auto-pedestrian crash fall on the pedestrian.

The lower the speed limit, the easier it is to generate traffic ticket revenue. It's a money game. In California, they tell you that whatever the judge tells you your traffic ticket fine is, say $100, your total is 6 times that when administrative fees are added in.

I believe low speed limits give police officers a legal means to stereotype drivers and pull them over. Since everyone is now speeding, they can pick and choose who to pull over and have a legal reason. Young? sketchy car? minority?

This article continuously states that speed limit signs have little to no bearing on how fast people actually drive, and it says Lt. Megge is actually arguing this. If it has no bearing on how fast people drive, and changing speed limits actually doesn't affect the speed of traffic like they say in the article, why are they changing them? Why are they too low right now? Why does the 85th percentile matter at all? If people pay as little attention to speed limit signs as is mentioned in the article, wouldn't changing them have no effect on safety whatsoever - be it lower or higher, better or worse?

I'm an advocate of getting rid of speed limits completely. Rather than having speed limit signs, post signs that suggest the highest speed for reasonably safe driving. If there is an accident that causes you or someone else damage, the court can use those signs as a way to say that you went against better judgment and assign a greater punishment.

I'd love to know what the best speed to drive at is, rather than having the posted speed limit be the speed at which no one drives except when the police are around.

All traffic laws do one thing: increase the likelihood that all drivers behave predictably. Going a certain speed. Stopping at a stop light. Not changing lanes erratically. It's all about being predictable to other drivers so they have time to react to your presence. Being unpredictable is the quickest way to get into a collision. Speeding but being predictable is a lot safer than going the speed limit and swerving erratically.

An interesting rhetoritcal question is what will be a suitable speed limit when we have roads that are 100% autonomous vehicles?

I used to live in Atlanta where all of the major highways were capped at 55 around the city. The average highway speed was probably 75-80. If you drove the speed limit, you would be the most dangerous car on the road by forcing the flow of traffic to run around you.

Let's just focus on getting self-driving cars out the door as quickly as possible.

Raising the speed limit to the 85%/15% ratio isn't necessarily the safest way to handle the driving. If there's a road where more than 15% of drivers are going faster than the speed limit, one option is to raise the speed limit, but the other option is to change the road so people won't drive so fast on it.

Since this is HN, what about technological solutions? Better speeding cameras could enforce speed limits universally. Self driving cars would obey speed limits automatically, however they could safely drive at much faster speeds than humans (albeit with significantly reduced fuel efficiency, which I think will be an issue.)

undefined

The assumption is that there's no enforcement mechanism, so you might as well give better guidance to the few people who comply voluntarily.

Perhaps the police will just go the other way and have a fleet of drones enforce compliance.

The trick to speed control is to set the speed limit 10-20 mph slower than you expect traffic to actually go, and never or infrequently enforce it until it exceeds +20mph. This is the M.O. for many places already.

Speed does not cause accidents. Bad driving does.

The two are sometimes, but not always, linked.

We should stop focusing on policing speed and start policing bad driving.

Using the article's logic, why have speed limits on country roads at all?