Arbitrariness troubles justice and should be avoided where possible. However, Mr Balko is overstating the arbitrariness of the situation. If a person with a blood-alcohol content of 0.08 is genuinely capable of reasonable driving, they're probably not going to be pulled over. (Even if obviously impaired, the odds of getting caught are low, given the ratio of cars to cops in most cities.)...
From the other direction, drinking is not the only activity that impairs drivers, but the appropriate response to that is to target dangerous behaviours as it becomes possible to do so. Texas is, for example, due to take up some bills that would outlaw text messaging while driving, as other states have already done. It's true that the public is hurt by the consequences of an elevated blood-alcohol content, not the BAC itself. However, the same drinking that elevates blood-alcohol levels limits a driver's effective judgment about his own capabilities; the existence of an external signpost, whether it's 0.05 or 0.08, can help clarify their thinking, and the possibility of sanctions can deter risky decisions. Given the potential costs of drunken driving to the people who aren't behind the wheel, I'm not especially bothered by the inconveniences incurred by the drinkers.
E.G. also writes that my complaints about checkpoints are complaints about checkpoints, not DWI laws themselves. But that's kind of the point. The fact that people between .08 and .10 aren't impaired enough to make the sorts of mistakes clearly impaired drivers make necessitates the need for the roadblocks. Which in addition to being constitutionally suspect, are more revenue generators than highway safety endeavors.
E.G.'s argument seems to be that because BAC is some way of measuring alcohol impairment, we should go ahead and use it, even if it isn't a particularly accurate method of measurement, and even if the cutoff point we choose is mostly arbitrary. It's really an argument against consuming any alcohol before driving. Maybe that's what he wants. I'm convinced that's what MADD wants. But let's then at least be up front about that.
To illustrate my point about impairment and recklessness, consider two scenarios: In the first, a driver is coming home at 2 am. Because there are few other cars on the road, he conducts a rolling stop. A cop hiding across the street sees him and pulls him over. The driver blows .081. He's looking at a pretty severe punishment, not for the rolling stop, but for having a level of impairment that's a tick over a mostly arbitrary legal limit. Contrast that to a driver coming home at 2am who hasn't slept in 36 hours and flies through a stoplight at 60 mph. He gets pulled over, but hasn't had a drop of alcohol. He's going to get a much less severe penalty, despite the fact that his infraction was more serious and that he's much more impaired.
E.G.'s endorsement of text messaging bans is another example of someone calling for laws banning allegedly dangerous activity with little regard as to whether those laws will actually work. As I've argued before, how exactly are police going to enforce this law? That is, when you're driving by at 65 mph, how can a police officer sitting in a cruiser tell if you're texting or if, for example, you've pulled up an application like GPS Drive on your phone, and are checking for your next turn? If you're going to argue that drivers shouldn't be using GPS aps on their phone, either, can you really argue that those aps are less safe than fumbling with a map or an atlas? Or reading directions printed out from Mapquest? Are we going to argue that maps, atlases, and GPS devices are all also too distracting and should be banned?
Sure enough, the first study of texting bans has actually shown them to make the highways less safe. Accidents and fatalities went up. The researchers who conducted the study speculate that in jurisdictions with highly-publicized texting bans, motorists lower their phones to avoid detection, which takes their eyes off the road. Ironically, the time you're most likely to get caught texting and driving is the time when it's least harmful—when you're stopped at a light, or driving slowly enough for a cop to see what you're doing.
I realize that "abolish drunk driving laws" sounds a bit like a parody of libertarianism, like arguing for heroin vending machines in elementary schools (E.G. calls it a "stunt argument"). But I'm not arguing in favor of a freedom to drive while obliterated, or that there's some right to drive drunk that outweighs the safety of other motorists and pedestrians. I'm arguing that public safety laws need to be clear, enforceable, and should actually achieve their intended purpose. I'm not sure our current DWI laws meet any of those criteria. (It's just a small sample from one city, but see this recent article from Nashville, where DWI arrests are down by a third due to budgetary woes—to no effect on actual DWI fatalities.)
One other thing. Some commenters here and elsewhere argued that removing the de facto BAC threshold would put more discretion in the hands of police officers. I'm not sure that's true. Reckless driving is already a crime. And police in most jurisdictions can already arrest you for DUI even if you're below .08. The .08 threshold is merely the point at which you're de facto assumed to be intoxicated. Now that most police departments use dash cams, it should be pretty easy to show with video evidence that a motorist was swerving between lanes, driving at excessive speeds, cutting people off, or otherwise driving recklessly. And that is what makes the roads less safe, whether it's caused by too much to drink, sleep deprivation, medication, or disciplining your kids in the back seat.