Tesla's 'Leaked' Email About Using Employees to Beta Test Full Self-Driving Cars Makes No Sense
There's something off about the timing, motivation, and especially the content of Tesla's latest "leak."
Everyone knows I love Teslas. I just drove a Model S across France, and it was marvelous. But I also have to call it like I see it, and you have to be blind or ignorant not to see how Tesla is manipulating the media with this latest "leak" about the company's "full self-driving beta testers."
Remember, it was only last Thursday that the SEC announced they were suing Elon Musk over his tweets. (It was later reported that Musk had turned down what looked to be a very reasonable settlement.) On Friday, the stock collapsed and the shorts cheered. Saturday, Tesla announces a not-as-good-as-the-last-one settlement: Musk, out as board chair; a $40M fine, to be split evenly between Musk, personally, and Tesla.
Also on Saturday, an internal Tesla e-mail is "leaked" to Bloomberg News. The headline:
The strategic leak is a powerful tool among skilled corporate communications people. A tantalizing leak, perfectly timed and delivered to a credible publication, is the best response to a wave of negative press. And what a response this headline happened to be.
I'm not claiming that Tesla leaked this story to Bloomberg News, but if it were a strategic leak—and especially if it was Tesla employee who just couldn't keep this to himself—it's interesting this story didn't originate at the usual Tesla leak bucket, Electrek, but with Dana Hull, one of a few credible journalists covering this sector.
Which brings us to the meat of the leak, or rather, the lack of any meat. Let's deconstruct Bloomberg's article.
"Tesla Enlists Employees to Be ‘Full Self-Driving’ Beta Testers"
That sure sounds impressive. Tesla is only years behind Musk's promises for delivering full self-driving capability. The shorts might see this story as a sign of weakness, but the layman, after seeing Tesla blasted in the media for weeks, likely assumes this story means good things are happening.
Except, the story doesn't make any sense. The first two paragraphs are insane:
Elon Musk has asked for Tesla Inc. employees to test what the company has billed as full self-driving capability and is dangling $13,000 in savings to entice them to help."
"Musk wrote in an email obtained by Bloomberg News that Tesla needed about 100 more employees to join an internal testing program linked to rolling out the full self-driving capability. Any worker who buys a Tesla and agrees to share 300 to 400 hours of driving feedback with the company’s Autopilot team by the end of next year won’t have to pay for full self-driving -- an $8,000 saving -- or for a premium interior, which normally costs $5,000, Musk wrote.
This is so crazy I don't know where to start, so let's just jump in with Full Self-Driving—FSD, to the fanboys—a vague, clickbait-ready term perfect for a bad-news Saturday. Musk's statements about FSD have been all over the map, and Tesla conspicuously avoids the traditional SAE definitions. Tesla's site claims FSD will work "in almost all circumstances." That sounds like SAE Level 4, which I call geotonomy: capable of replacing a human driver, within certain conditions, within a geofence; and the more you limit the fence and conditions—say, a nice, low-density suburb with decent weather during most of the year——the easier the problem becomes to solve. That's why companies like Waymo and Uber test in places like Arizona.
What exactly will Tesla's beta-testers be doing? According to Tesla, every car shipped already comes with all the hardware necessary for FSD. Tesla already gathers camera and radar data from the hundreds of thousands of Teslas on the road, whether or not they've paid for FSD. That's called shadow driving, and every Tesla owner alive currently contributes to Musk's dream through the company's Fleet Learning
So, these 100 employee beta-testers wouldn't add anything to Tesla R&D—unless they're doing something else.
If that "something else" looks anything like the testing going on at Waymo, Uber, Cruise, Aurora, or any of the other companies with billions of dollars at stake, Tesla may be driving itself into a dangerous corner. A professional L4 test driver's job is to sit, hands off, and watch the car do its thing while being ready to take over at any moment.
Consider the expensive lessons learned by Waymo and Uber after millions of miles of testing, numerous crashes, and one death:
- At least two highly trained people should be in the car at all times. Why two? Because the first one will fall asleep from boredom.
- Test vehicles should include a camera-based Driver Monitoring System. Why? Because being awake ≠ paying attention.
Tesla's leaked e-mail doesn't seem to say anything about special training for their employee volunteers. (If it did, none of the outlets who actually saw the e-mail mentioned it.) Musk said the employee program is first come, first serve—not a great way to vet the suitability of beta testers. Having a Tesla email address does not equal actual training, let alone skill. Will Tesla use sensors in the seats to verify whether there are two people in the car when FSD engages, just to be safe?
Alas, Tesla declined to comment for this story.
That was probably wise, because I don't know how they would answer the problem of their Driver Monitoring System (DMS), which is that they don't have one, or at least an effective one. Safety is a moving target, but I think it's difficult to safely test a hands-off L4 system without a camera-based DMS. The Seeing Machines system works rather well in Cadillac's SuperCruise. Tesla resorts to a cheaper, less reliable torque-based sensor. Why? Cost, probably.
Here's a comparison of torque sensor versus camera DMS safety.
Putting 100 self-driving beta testers on the road with safety protocols inferior to what most companies were using prior to Elaine Herzberg's death seems unwise.
Perhaps Tesla is retrofitting camera-based DMS systems to all 100+ employee cars. It's certainly possible. Euro NCAP, the European vehicle safety rating system, will require DMS starting in 2020. BMW is deploying camera-based DMS in the new X5. It's hard to believe Euro NCAP won't declare camera DMS the standard. If Tesla isn't already looking into it, and Euro NCAP does declares camera DMS the standard, Tesla will lose their 5-star safety rating in two years.
Which brings us back to the employees "volunteering" to beta test FSD. I admire anyone compelled to help Tesla's greater cause. I also admire Musk, warts and all. But unless Tesla has a super special secret safety plan for making sure one of these volunteers doesn't kill someone, becoming a beta tester for the greater good doesn't seem wise.
And that's coming free me, a Tesla fan who wants to believe. But, still, I have to call it like I see it.
Besides, are these people really getting a deal by "saving" even eight thousand bucks on FSD? It depends on when and where FSD will work—if it will ever work—and where the driver lives during the lifetime of the average Tesla. None of the engineers I know believe L4 will work outside of tightly geofenced areas in the next five to seven years. If risking your life and the lives of others to save $5-10,000 sounds cool, be my guest.
Which leaves that sweet premium interior, worth $5,000. I would volunteer in a heartbeat to get that for free. But only if I could get it in white. And I got to choose where and when I had to do my "testing." And there weren't any other volunteers testing near me, or my kids.
Maybe Tesla is doing it right. Maybe Tesla just doesn't want to spend the money on hiring contractors.
Or maybe Tesla has 100 cars with premium interiors they desperately need to move.
Or maybe Tesla's comms team are the hardest working people in comms. So strategic.
Alex Roy is founder of Geotegic Consulting and the Human Driving Association; editor-at-large at The Drive; host of The Autonocast; co-host of /DRIVE on NBC Sports; and author of The Driver. He has set numerous endurance driving records, including the infamous Cannonball Run record. You can follow him on Facebook, Twitter and Instagram.
- RELATEDThe War For Autonomous Driving: 2017 Mercedes-Benz E-Class VS. 2017 Tesla Model SIt’s a DrivePilot vs. AutoPilot cage match fight to the death—potentially yours.READ NOW
- RELATEDThe Battle for Best Semi-Autonomous System: Tesla Autopilot Vs. GM SuperCruise, Head-to-HeadBeing the best isn't just about technological prowess, it's about how well the designers understand human driving—and human nature.READ NOW
- RELATEDTesla Model 3: The First Serious ReviewThe most important car since the Model T is brilliant—with one big caveat.READ NOW
- RELATEDThe Dangerous Difference Between Electrek, Journalism, and TruthEditor Frederic Lambert is to truth what Autopilot is to safety.READ NOW
- RELATEDThe Language of Self-Driving Cars Is Dangerous—Here's How To Fix ItCurrent SAE Levels conflate automation with autonomy. That's a fundamental flaw—so let's ditch the levels entirely for something better.READ NOW