A recent study by IIHS tries to frame Advanced Driver Assistance Systems (ADAS) like Tesla’s Autopilot as potentially dangerous, because drivers speed more often while using the systems. The study’s authors found that drivers using no assist features speed 77% of the time, while people using adaptive cruise speed 95% of the time and people using adaptive cruise with lane centering (systems like Tesla Autopilot) speed 96% of the time. Because of this, the study’s authors conclude that fatal crashes are 10% more likely for people using ADAS.
There’s just one problem: Speed limits can’t really be used a as a yardstick to measure safety. As usual, the missionaries for The Church of Automotive Safety are out knocking on virtual doors trying to spread their gospel, and they want us to accept their findings on faith alone.
To be fair, the study’s authors were honest about the limitations of their data (with one exception that I’ll get to below). The bigger problem is that articles like this one take the 10% figure and run with it, falsely concluding that technologies like Autopilot are dangerous. Then, the “safety advocates” on Twitter take that and run even faster. This is how The Church operates — they amplify each other’s mistakes, out of context data, and outright falsehoods, and then feed them to the public.
In the study itself, the authors admit that the difference between the behavior of full-manual drivers and ADAS users isn’t actually very big. In terms of speed itself, people driving manually drove an average of 6.1 MPH over the speed limit. People using adaptive cruise control averaged 7.0 MPH over the limit, while people using lane centering and ACC went 7.1 MPH over the limit. In other words, the technology made a difference of a whole 1 MPH.
There’s more that they readily admit to. The drivers in the study tended to speed more when speed limits were lower, and drivers who had a tendency to speed driving manually didn’t change their behavior much when ADAS technologies were used.
The media and the TSLAQ
chuckleheads “experts” on Twitter didn’t share any of those facts from the study with their audiences, probably because they made for a less interesting story.
I did say above that I felt the study’s authors themselves did one thing dishonest, and that is how they treated speed limits. To be fair to them, this is mostly the result of other studies they were relying on, but looking at the studies they referenced, I think it was misapplied. The “Power Model” used to estimate risks uses the increase in the average speed of traffic, and not really the individual behavior of one driver.
There’s a reason for that: one’s individual accident risk is generally more correlated from one’s departure from the mean speed of traffic (the “flow” of traffic) and not their departure from the speed limits themselves, and there are many circumstances where you’d actually increase your risks of a fatal accident if you were to lower your speed. I covered this in more depth in part 4 of the Church of Automotive Safety series, but I’ll briefly touch on it here.
The study’s authors almost saw what was going on, but didn’t fully appreciate the implications. When they saw that people speed more on roads with lower speed limits, the obvious conclusion should be that the speed limit on that road is improperly set in most cases.
This is where the difference between speed limits and the actual safe speed of travel comes into play. Unless there’s a good reason to do otherwise (a blind curve, a school, or some other elevated risk), it’s common practice to set speed limits around what most drivers are doing, usually by setting the limit at the 85th percentile of measured driver speeds. This makes for a realistic speed limit that most people will naturally follow.
When you set speed limits below the 85th percentile, you end up with a situation where you’re creating two tiers of traffic speeds: one set of people trying to obey the limit and avoid tickets, and another group of people driving at a more natural speed for the road. This creates more chances for accidents, as people’s speeds are differing more from each other. There’s more passing, more people getting stuck behind other people, etc.
This happens far more often than it should because traffic engineers don’t get a full say in how the limits are set. Helen Lovejoy goes to the city council, state legislature, parliament, or Congress and demands that people slow down “for the children.” Then, the law mandates lower limits and strict enforcement, which creates more problems than it solves.
More details in this video:
In the study that the study references (don’t worry about Inception-level confusion — you can find it here), they do point out that there are important reasons to get drivers to slow down below what feels natural for a stretch of road. Increased pollution (not so much a problem with EVs), overall risk of fatalities, and dangers to pedestrians are all important, too.
We can’t really do this with speed limits, though. The real way to get people to slow down is to change the road up and make it feel less comfortable to go fast. Narrowing the road, adding little curb bumpouts, putting in speed tables (not speedbumps, because they tear up vehicles), and even changing the striping of the road all slow people down without risky traffic stops and fines.
I know, I know, revenue will be lost if we don’t send brigands out to extort money from people. I’m perfectly fine with that. If the idea is to save lives, then we should prioritize that first instead of pretending to prioritize safety when the real reason is revenue. Even worse, when you consider the way that traffic enforcers tend to disproportionately pull over minorities, there’s really no excuse for having badly set speed limits.
Don’t expect these kinds of studies and their misuse against companies like Tesla to stop, though. The Church of Automotive Safety needs to keep the money flowing and needs us all to believe in their gospel. If we don’t, they’ll lose out on money and ne’er-do-well upstarts like Elon Musk might even get too powerful for the oligarchy to tolerate. We can’t have that!
Featured image: Screenshot from Tesla’s shareholder meeting in late 2020 showing an Autopilot slide.