Safety Third: I Was Wrong To Fear The Tesla FSD Beta
An article I wrote a couple months ago got a lot of negative attention. Some people harassed me on Twitter to the point where I blocked a few people. One very prominent person in the industry told my editors that it was “idiotic.” At the time, I thought the Tesla Full Self-Driving (FSD) beta rollout was a potential danger to the public, but I didn’t think regulators should come down on it. What I preferred was that the industry work to provide more driver training and awareness. I even wrote a couple of articles giving people safety tips I had picked up working in other industries and things to watch out for.
It turns out that I was completely wrong, though. Blood isn’t running in the streets. While it seems unlikely that there have been zero crashes, there are no media reports I could find telling us about a crash while using the system. Sure, there was much hand-wringing about the possibility of a crash, but to date none of the people frightened by it have followed up with any “I told you so” stories. If anyone was seriously hurt or killed, we would have all heard about it, and that didn’t happen.
So, I’m going to eat a little crow and admit I was wrong. I’m also going to try to learn from my failure on this.
Safety Doesn’t (and Shouldn’t) Come First
As I tried to figure out where I screwed up in my thinking, I came across Mike Rowe’s videos and writings about “Safety Third.” No, Rowe didn’t say safety isn’t a priority, but we are deluding ourselves if we think it’s our highest priority. If it was, nobody would do anything. We wouldn’t drive, we wouldn’t eat fast food, and we wouldn’t send people to climb on suspension bridges or up antenna towers. We certainly wouldn’t do anything like mountain biking, competition shooting, or hiking out in the wilderness with no cell signal.
All of those things are dangerous, you know. If it saves one life …
The fact is, safety should be a priority but not THE priority. We have to first figure out how to get things done. Then, we find a way to turn a profit or have fun doing things. After that, and only after that, do we analyze the proposed activity to figure out how to make it safer. In other words, we tailor our safety to the things we want to do, and not the other way around.
When we forget that getting things done comes first, we make mistakes that cost people their livelihoods and even their lives due to economic stagnation, depression, and even starvation. If applied in the military, you could end up with a destroyed or occupied country. We applaud heroic rescues, even when the rescuers take unusual risks to save people, so we definitely don’t want police, firefighters, EMTs, and passersby to stop being brave, do we?
The cure of safety can’t be added in such a way that its impacts are worse than the disease or danger we tried to get rid of.
Nobody Else Is Responsible For Your Safety
What Mike Rowe said in a video really hit home on why Tesla’s FSD isn’t causing blood to clog the storm drains in places where the beta has rolled out. It comes down to the error of “Your safety is our priority.”
Mike’s crew went through the first three seasons with no accidents. In season 4, they started having accidents and injuries of all kinds. The job sites they were working in weren’t any more dangerous than the ones earlier in the series, but accidents were going up. He tried to figure out what was happening. What he figured out was that people are actually safer and more cautious when they have to take responsibility for their own safety.
Rowe found studies showing that crosswalks that tell you when it’s safe to walk have more accidents than crosswalks that require the pedestrian to figure out when it’s safe. When people rely on someone or something else for their safety, they forget to look out for themselves. They forget that there’s a difference between compliance with safety rules and safety itself, and one can easily be compliant and still get hurt.
When he got people on his crews to start looking out for themselves again, that extra dose of personal responsibility for safety brought the accidents back down.
Personal Responsibility Works Elsewhere
This concept isn’t just a Mike Rowe anecdote. Plenty of other studies and experiences back it, too. European experiments with removing signage actually reduced accidents. Montana’s accident rates went up when they decided to start having speed limits. Roundabouts, with their one yield sign, are safer than intersections with relatively complicated light systems.
The rules we think are making us safer aren’t doing the job we think they do.
It goes beyond cars and workplaces. A growing number of states have decided to stop requiring people to get a license or permit to carry a gun for personal protection. People against these “constitutional carry” laws always gave dire predictions, sometimes literally saying that blood would run in the streets from all of the extra shootings. In fact, the opposite is true. There have been no increases in violent confrontations, murders, or accidents in the states that have passed this in the past 15 years.
This happened because people actually took safety more seriously now that the state wasn’t requiring them to get a state permit. Instructors I talked to in Arizona said their state-mandated classes went down, but students taking other, more comprehensive, classes went up. Without the state saying “This 3-8 hour class is all you need,” people went above and beyond the old minimums to make sure they were safer while the people who were really a problem (criminals and people with severe mental issues) were never getting a permit to begin with.
Agree or disagree with all of this, it’s well established that making people take personal responsibility for their safety leads to lower rates of undesired outcomes. Telling people that the rules will save them leads to more rule following, but less common sense. It’s all about balance.
“Safety First” and “Your Safety is Our Responsibility” May Have Failed Autonomous Driving Programs, Too
Drivers knew that FSD safety was still legally their responsibility, and they took it seriously. Compare this to Uber, which took responsibility for their test drivers’ safety and had one watch reality TV while the vehicle ran over a cyclist. Venture Beat tells us that Waymo’s trained test drivers were involved in 18 accidents in 20 months. These companies have or had great training programs, but taking responsibility away from drivers with an overly formalized process may have paradoxically created the very problems they were trying to prevent.
The key takeaway here is that you want autonomous vehicle testing to be safe, but you don’t want to make test drivers fall into the complacency trap. Delusions like “Safety First” rob us of perspective. “Your safety is our responsibility” robs people of the personal responsibility and common sense needed to actually keep people safe.
Don’t get me wrong. We need safety rules and we need safety training for unusually dangerous activities, but those rules and that training needs to complement and enhance common sense rather than try to micromanage people and take common sense away from the situation. Given the experience most drivers have, it might be that special training backfires at worst and doesn’t help at best.
Even if I’m totally wrong to listen to Mike Rowe, I was definitely wrong and Elon was right. Nearly three months have gone by with the FSD beta without serious incident. People didn’t need special training to test the FSD beta software.