I Won’t Drive The Roundabout

A couple of days ago, my Facebook friend Gary Chartier posted this article from USA Today on increasing speed limits on American highways. As it happens, I’m at work on a paper on a traffic-related blog post I wrote here a few months ago, to the effect that police tailgating ought to be regarded as a form of legal entrapment. To that end, I’ve been reading a lot about cars, roads, road safety, traffic, tailgating, police chases, and entrapment. Research aside, I happen to be an unapologetic traffic-ethics bigot inclined to the view that when it comes to driving, it’s my way or the highway. So naturally, I leapt at the chance to pontificate on Gary’s post.

I think higher speed limits have a paradoxical effect. The higher the speed limit, the greater the generalized fear of driving; the greater the generalized fear, the greater the vigilance with which people drive; the greater the vigilance, the fewer fatal (high speed) accidents. (The general pattern has been statistically demonstrated.) Unfortunately, when there are accidents at that speed, they’re more likely than usual to be fatal. In New Jersey, higher speed limits have led to fewer fatalities. That said, I don’t think higher speed limits are a legitimate way of reducing fatalities; I call it “regulation by terror-induced vigilance.” It’s like reducing crime rates through extremely aggressive methods of deterrence.

Incredibly, Nathan Byrd, another of Gary’s FB friends, had the audacity to question my claims right there on Facebook.

I’m unclear what you’re proposing if it seems that the results are better overall.

A related point for surface street driving is that removing lights and putting in more roundabouts and other street formations that give a higher sense of threat seem to keep people alert and thus less likely to get into an accident. Would you similarly disapprove of that kind of change?

Khawaja 2:

I wasn’t so much proposing as remarking on a trade-off and an ambivalence: you can in principle get lower mortality by the judicious use of vigilance-inducing fear, both with higher speed limits and with roundabouts as you say, but it’s not obviously justified to reach a favorable public policy outcome by threatening innocent people, i.e., deliberately, intentionally subjecting them to life-threatening threats. With a day’s reflection (I went through the notorious Somerville Circle last night, in Somerville, NJ) I guess I would say that I disapprove of both changes. It isn’t just that higher limits and roundabouts “give a higher sense of threat.” They deliberately subject people to life- threatening conditions, then regard lower mortality rates as a justification for doing so. I think that procedure violates rights. It basically involves assaulting people in order to keep them vigilant. So it seems a case of doing injustice so that a good outcome may come. But the injustice is a bad outcome of its own.

Though roundabouts generally have lower accident rates than intersections, certain major exceptions are worth noting. Even if the accident rate was low, there is no conceivable way to drive through this circle if you don’t know exactly what to do at each micro-second of your trip through it.

Byrd 2:

It’s a complex issue. I just wondered what you had in mind a bit more, so thanks for explaining.

I’ve also heard that seatbelts and other kinds of safety features that make safety obvious to the driver prompt them to compensate for this by driving in a riskier fashion, and generally that tends to go past the tradeoff point and put them in more danger than they were prior to the safety features being added.

I hear what you’re saying about the ethical implications, but I’m not entirely convinced for a couple reasons.

I don’t know that roadways have an inherent safety/danger level prior to drivers interacting with them. Or at least, it’s not entirely pre-determined, and except in extreme cases, it’s going to have a wide variability across the population of drivers.

Polls have shown that the vast vast majority of drivers believe that they are better than average in driving skill. This is not exactly Dunning-Kruger levels of deception, but it does mean that driver perception plays a big role, even when road designers don’t especially care to influence it one way or the other. (I am one of the few who believes that I am less capable than most drivers, and that actually makes me a safer driver because I make effort to anticipate situations and not put myself in a position to have to rely on those skills.)

It would seem that if we convince someone that a situation is safe when it isn’t, or if it’s predictable that a person will change their behavior to a riskier one if we add extra assurance of safety, that has to be at least as ethically suspect as your concern. We seem to be caught in a no-win situation then where any roadway will necessitate widespread ethical violation. That doesn’t seem quite right somehow.

I would say, though, that I don’t think the purpose of the designs I’ve seen is to literally make things dangerous and then prompt drivers to correct for this error. It’s more a reflection of how human awareness functions and how easy it is for people to drive in a hypnotic state. So, providing a lot more opportunities for a response is helpful to driver awareness.

Think of the various rumble strips set in place on many large highways so that truckers have some prompting to wake up. That sound might be jarring if you’re not ready for it, but in most cases, that’s a very good thing, and I can’t imagine a trucker feeling slighted by it or feeling that their rights have been violated.

Probably a shorter way I could have said all this is that it seems that a certain level of transparency (and thus consent) is the best way to handle such a thing. I certainly don’t like the idea of manipulative road designers treating drivers as rats in a laboratory experiment. But I think that giving drivers extra stimulation while driving is far better ethically than a road that hums you to sleep. In either case, informing drivers of how things work would help reduce the sense of trickery or manipulation.

I don’t agree with that, but given the difficulties of having a conversation like this on Facebook, I invited Nathan to continue the conversation here, and he agreed. Of course, the ball is now in my court to respond. Hope to do so later this week with “Khawaja 3.” But feel free to join in.

Some relevant reading from Tom Vanderbilt’s classic, Traffic: Why We Drive the Way We Do (and What It Says About Us), pp. 178-79.

13 thoughts on “I Won’t Drive The Roundabout

  1. There’s an intersection near a large mall here that was recently changed to a roundabout (or a rotary, as we used to call them in New England; I’m not actually sure what they’re called down here). What used to be a clear rule — a four-way stop — is now a somewhat less clear rule: cars entering the roundabout are supposed to yield to cars already in the roundabout, depending on how close or how fast the arriving car judges the other car to be. I think the change was made in the interest of speed, not safety, though I reckon there’s an appreciable gain of speed only when there are few cars. It certainly feel less safe.

    Also, at night it’s hard to see that it’s a roundabout rather than an intersection; I expect someone’s going to drive straight into the median at some point.


    • P.S. – When I was living in Boston, Kenmore Square (not to be confused with Kendall Square) used to be a traffic nightmare; it was an intersection where streets came in at various angles and it was radically unclear how to get across the rival traffic flow to the other side. I haven’t seen it for decades so they may have improved it now.


    • the change was made in the interest of speed, not safety, though I reckon there’s an appreciable gain of speed only when there are few cars. It certainly feel less safe.

      From many traffic engineers’ perspective, “the change was made in the interest of speed, not safety” is a false dichotomy precisely because things feel less safe at a higher speed. The engineers make changes in the interests of speed partly because speed makes things feel less safe, and when that’s so, people tend to drive more safely. Statistically, that tends to be true, but ethically I find it wrongheaded.

      This paper presents the statistical point, but the relevant causal point is that you get the reductions in crash severity and injury/mortality they mention by deliberately scaring the shit out of everyone going through the intersection. In other words, the gains in safety are purchased through vigilance induced by literal terror.



  2. Khawaja 3
    OK, so here is my response to Nathan’s last comment, which I called “Byrd 2.”

    Here’s something of a reformulation of my view. In the cases I have in mind, engineers are intentionally subjecting drivers to real (not imagined) danger in order to induce fear in them that increases their driverly vigilance. One problem here is that increased speed increases actual danger. Another problem is that drivers are demonstrably bad at judging risk (H.C. Joksch, “Velocity Change and Fatality Risk in a Crash: A Rule of Thumb,” Accident Analysis & Prevention 25:1 [1993]). Even if they have the generalized sense that a roundabout feels dangerous, they don’t have a fine-grained sense of the specific dangers of the particular parts. So even if the numbers work out to produce outcomes that are safer than before, they do so by creating dangers and collateral damage that wasn’t there before. And even if, on average, certain interventions create safer ex post facto outcomes, there are major exceptions to the rule, like Somerville Circle, where the safety outcomes are worse than before.

    But in some ways, the preceding set of claims bypasses (so to speak) my real complaint: it’s fundamentally immoral (I would say) to treat someone as a mere means to a positive outcome you hope to bring about, and the policy interventions just described amount to doing that. They involve deliberately creating a real mortal danger that didn’t exist before the intervention, then exploiting the fear from the introduction of the danger in order to produce a defeasibly safer aggregate outcome. Doing this would be like creating a police state or regulatory-administrative regime that deliberately operated in a capricious, arbitrary manner in order to produce a generalized fear that, say, lowered the crime rate or increased regulatory compliance (where ex hypothesi regulatory compliance involved safer outcomes). Unless you adopt an old-fashioned sort of consequentialism like J.J.C. Smart’s, none of these things is going to pass ethical muster.

    The cases you have in mind are quite different than that. Rumble strips don’t create a danger where none previously existed; they simply signal the fact that you the driver have steered into a dangerous situation. So there’s no “treating as a means” there. Granted, engineers will introduce curves (“Solomon curves”) into an otherwise straight road (typically an interstate highway) to give drivers something active to do, thereby ensuring that they stay awake. But I don’t think that Solomon curves really introduce danger that wasn’t previously there. (Introducing a curve isn’t the same as introducing danger; Solomon curves are designed to be relatively gentle, easy to negotiate if you’re awake and alert.) It’s more accurate to say that interstate highways without Solomon curves are dangerous; there’s nothing risk-increasing about curves, at least if they’re properly designed. Whereas it’s a demonstrable fact that speed increases mortality in an exponential fashion. It’s not clear whether speed increases the risk of crashing per se, but it demonstrably increases the risk of mortality if you do crash (my inference from the Joksch paper I mentioned above). So the two cases strike me as very different.

    So there is a very clear sense in which a conscientious driver is put in justifiable fear of his life when he drives on a roadway where the speed limit is particularly high: a high speed limit incentivizes excessive speed in the worst drivers, and the faster those drivers drive, the greater the risk of death to anyone proximate to that sort of driver, even if in aggregate the crash and mortality rate goes down. But none of that is true of rumble strips or gentle curves. It would be completely irrational to feel fear at the sight of a rumble strip or a gentle curve. But it makes good sense to be scared at the sight of a pick-up truck or SUV recklessly barreling down a crowded interstate at 80 mph. The problem is, incentivizing that very behavior is part of the engineers’ strategy for safety. They’re purchasing safety through the introduction of danger, then relying on conscientious-but-frightened drivers to push down the rate of crashes and mortality. That strikes me as dangerous (almost trivially so, of course) and deeply unfair.

    Part of our disagreement is purely factual, I think. You say:

    I would say, though, that I don’t think the purpose of the designs I’ve seen is to literally make things dangerous and then prompt drivers to correct for this error.

    In that case, we may just be talking about different things. I’m objecting to engineering designs that literally introduce danger in order to force drivers to correct for it in the hopes of producing some sexy-looking statistic that appears to “demonstrate” a net gain in safety. As I see it, in the most extreme cases, the price of that net gain is injustice and terror.

    One complication for my view is that when it comes to roundabouts, the increase in speed involved is apparent rather than real. In other words, despite what Roderick said in his comment above, it turns out that roundabouts give the appearance of increasing speeds (which is how you get the vigilance effect) while actually decreasing them. I’m guessing that you get apparent increase in speed because you’re moving in a circle: 35 mph looks faster when traversed in a circle than when traversed in linear fashion. So despite the title I gave the post, my strongest objections are to increased speed limits, not roundabouts. But I still think that a lot of roundabouts do more or less what increased speed limits do: they create danger which creates fear which creates vigilance which creates safety. My experience is limited to four circles I regularly drive through in New Jersey: Tonnelle Circle (Jersey City), Somerville Circle (Somerville), Flemington Circle (Flemington), and a smaller, nameless one in Hasbrouck Heights. I just think they’re fucking crazy. Drive through the first three in rush hour in inclement weather, and you wonder each time whether you’re going to make it out alive.

    Though it’s admittedly a complicated issue, I don’t think we end up in the reductio that structures your comment. I’m objecting only to a certain specific kind of engineering design. There’s still plenty of leeway for designs that don’t literally introduce danger into the equation, then demand correction.

    Incidentally, Dunning-Kruger-type effects (even weak ones) are precisely why we should not rely on drivers to correct for dangers that engineers introduce: even if it doesn’t change the aggregate figures, the introduction of new danger creates cases where over-confident drivers under-estimate risks that wouldn’t otherwise have been there. And frankly, there is a real puzzle here: if Dunning-Kruger effects obtain in the case of driving, how could driver vigilance improve driver safety at all? How could systematically over-confident drivers manage to improve safety outcomes when subjected to increased risk? Does Dunning-Kruger not extend to cases of mortal fear? Or are the safety outcomes themselves exaggerated? There are a lot of unasked and unanswered questions here.

    Finally, though safety features like seatbelts do create a moral hazard, I don’t think they really “prompt” drivers to “compensate.” Drivers do that, to be sure, but that is an unjustified response to the presence of the feature. Whereas fear of danger is a completely warranted response to the presence of real danger. That said, the existence or magnitude of the compensation effect is itself hotly debated. It’s called the Peltzman effect, after the economist Sam Peltzman, who first argued for its existence (“The Effects of Automobile Safety Regulation,” Journal of Political Economy 83:4 [August 1976], 677-726). I haven’t worked through the technical literature disputing the existence of the Peltzman effect, but it’s worth noting that there is one, at least as sophisticated as Peltzman’s own research. As a non-expert, it’s just unclear what inference to draw about the whole thing.

    Anyway, I’ve gone on and on about this–to the point of overkill, so to speak–but thanks for your comments, which have forced me to clarify some things that were fuzzy in mind. Though they’re obviously not identical, there is some important overlap between engineers’ exploitation of fear to create aggregate safety, and cops’ use of tailgating to instigate moving violations. I’m mostly thinking about the former to get some clarity on the latter, my real quarry.


    • “I’m guessing that you get apparent increase in speed because you’re moving in a circle”

      I was thinking that when traffic is low you get an increase in speed because you don’t have to stop at a stop sign, and going nonstop around a (small) circle is faster than approaching an intersection, coming to a full stop, and then proceeding onward.

      When traffic is high, and you have to yield to perhaps several cars that are in the roundabout, you don’t get an increase in speed, because you’re stopping for several cars rather than one.

      “Doing this would be like creating a police state or regulatory-administrative regime that deliberately operated in a capricious, arbitrary manner in order to produce a generalized fear that, say, lowered the crime rate or increased regulatory compliance.”

      Not a big fan of Han Feizi and the Qin Dynasty, I take it.


      • Re roundabouts and speed: that may often be true, but at the roundabouts I’m familiar with in New Jersey, there’s no way to avoid slowing down almost to a full stop, at least momentarily, because traffic is often coming from so many directions that caution simply demands that you do so. And the sight lines are so bad that you end up having to do so. Plus, traffic coming into the circle from elsewhere may be coming so fast that there is effectively no such thing as “when traffic is low.” There might be fewer cars than usual in the circle and in the general vicinity, and the latter may seem as though they’re far away, but if they’re headed your way off of an interstate highway, they will reach the circle with alarming speed. Given all this, for all practical purposes, the “when-traffic-is-high” condition is the constant. But despite the fact that you have to slow down, it feels as though you’re going fast.

        But as I said, what I’m describing may well be specific to the roundabouts we have here in New Jersey, or at least to the ones I happen to drive on. In general, what you say sounds plausible enough.

        I had never heard of Han Feizi and the Qin Dynasty before you mentioned it. So I looked them both up, naturally enough, on Wikipedia.

        Devoting the entirety of Chapter 14, “How to Love the Ministers”, to “persuading the ruler to be ruthless to his ministers”, Han Fei’s enlightened ruler strikes terror into his ministers by doing nothing (Wu wei). The qualities of a ruler, his “mental power, moral excellence and physical prowess” are irrelevant. He discards his private reason and morality, and shows no personal feelings. What is important is his method of government. Fa (administrative standards) require no perfection on the part of the ruler.[8]

        This sounds like the best of Machiavelli, Adolph Eichmann, Henry Kissinger, and Donald Trump all wrapped up in one. What’s not to like?


  3. Pingback: Nightcap | Notes On Liberty

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s