No Fixed Abode: What's the Auto-Point of It All?

Jack Baruth
by Jack Baruth

One of the first things any child learns in the modern technological era is that there are tools for which the true purpose is explicitly stated and tools for which the true purpose is hidden behind some obfuscating official language, legal fiction, or disingenuous disclaimer. Examples of the former: shovels, over-and-under trapshooting shotguns, noise-canceling headphones. Examples of the latter: BitTorrent, “professional” lock-picking kits on Massdrop, the Hitachi Magic Wand.

With the simultaneous democratization of tech and increased frequency of tech-related legislation, more and more things are falling into the category of “used for purposes other than intended, or in a manner other than suggested.” Nobody ever lets the FAA know that they’re going to be flying a Phantom drone over a motocross track, nobody ever deletes their MP3s when they sell their CDs back to Half Price Books, and nobody ever takes the Yoshimura pipe off their GSX-R1000 when they leave Willow Springs and ride back home.

From the moment that the Tesla “Autopilot” feature was introduced, with its copious disclaimers and strident request that the owner keep his hands on the wheel and continue to act just like he was driving the thing himself, the whole world has treated Autopilot like it was Napster. Oh, sure, I’m just going to keep looking ahead with my hands on the wheel, wink-wink, nudge-nudge. The near-universal assumption, one I’ve seen echoed by dozens of Tesla owners, is that Autopilot is, in fact, a functioning autopilot system and all the disclaimers are just there to keep the lawyers happy.

What if that’s not the case at all?

Autopilot isn’t the only system of its type; Car and Driver compared three other semi-autonomous cars to the Tesla Model S. It found that although Tesla had perhaps the least capable and impressive hardware for the task — a complaint echoed by Tesla owners — the Model S easily outperformed the BMW, Mercedes, and Infiniti systems. No software engineer would be surprised by that result; Tesla has clearly been through many more iterative cycles of software development than anyone else, and software matters much more than hardware when it comes to autonomous operation at this current point in time.

In C/D’s testing, the Tesla required 29 interventions in a 50-mile loop, the Infiniti required 93, and the Germans split the difference. Twenty-nine interventions in 50 miles isn’t exactly what you’d a call a self-driving car, yet plenty of Tesla owners have used Autopilot on less demanding or better-marked roads to completely divert their attention away from the vehicle’s operation. Watch the video below and then read the critical parts of the description:

I actually wasn’t watching that direction and Tessy (the name of my car) was on duty with autopilot engaged. I became aware of the danger when Tessy alerted me with the “immediately take over” warning chime and the car swerving to the right to avoid the side collision.

Note 2: In case you’re curious, I’m listening to an audiobook in the background. It’s a Malcolm Gladwell book (excellent book).

Second part bolded to show the truthfulness of stereotypes.

Clearly, this driver is treating “Tessy’s” Autopilot capability in the same manner as the latter kind of technology discussed in the opening paragraphs. You can almost hear him thinking, Of course Autopilot works and can be left alone. They wouldn’t release it if it didn’t work. That keep your hands on the wheel stuff is just for lawyers. Go, Tessy! You can do it without me! This is a great example of what I personally call the antibiotic-resistance effect of legal disclaimers and it also reflects the unspoken idea that Autopilot isn’t really useful unless it allows the driver to completely divert his attention to the Internet or a DVD or the thoughtful perusal of a Malcolm Gladwell audiobook.

Fortunately or unfortunately, Tesla is in the middle of discovering the difference between the attitude our society has to ephemeral, unreal tech products like video games or websites and the attitude our society has to something that you pay real money to physically own or operate. At the risk of sounding trite, the primary characteristic of “beta” software is that it is allowed to crash. The alpha release will crash, and the prod release should not crash. The beta is permitted to crash from time to time.

In reality, Tesla’s so-called “beta testing” feature is nothing of the sort. Autopilot logged well over 100 million miles before anybody was killed using it. Try playing any of the large-scale multi-player video games out there in a “beta release” and you will see crashes and failures on a constant basis. Even in the worst-case scenario of failure, Autopilot just slows the car and demands user intervention. Nobody has ever been randomly and unexpectedly steered into a bridge abutment by Autopilot, nor can anybody claim that they were rammed on the freeway by an out-of-control Autopiloted Tesla.

The “beta testing” label, therefore, is just that — a label intended as an aegis of sorts to discourage lawsuits — and the owners are perfectly aware of the fact. No wonder, then, that they treat Tesla’s caveats about having one’s hands on the wheel and one’s attention forward with similar disregard. The average Tesla owner spends his days using his work laptop for personal purposes despite the explicit warnings on his sign-in screen. Then he listens to music that he “ripped” or borrowed from a friend or a public source. Then he prepares to go home and play a video game for which the consequences of failure, on the part of the player or programmer, amount to nothing more than a “respawn.” Is anybody surprised that people are watching movies and surfing the Web while Autopilot has control of the car?

More interesting than that is whether Autopilot has any credible advantages or benefits when you use it exactly as intended: hands on the wheel, eyes ahead, attention on the road. Most people, if you asked them, would laugh. Of course it’s useless if you use it the way you’re “supposed to.” Isn’t everything nowadays? Are any of us taking our iPhone earbuds out every twenty minutes or stretching our hands every ten during office work?

While I can’t speak for the Autopilot feature — the only way I’ll be driving a Tesla any time soon is if I buy one, something that is unlikely to happen in this decade — I can attest that a semi-autonomous car works very well to reduce fatigue and stress. A few months ago, I drove an Acura TLX to Watkins Glen from Ohio in the dead of night. My co-driver for the trip had immediately turned all of the “features” off, but for the last three hundred miles I got behind the wheel and turned them all back on.

The combination of Lane-Keep Assist and distance-estimating cruise control, though they might seem like insignificant or useless features when considered separately, works to all but eliminate driver fatigue during long hauls. I didn’t surf the web or text people while I was driving. Good thing, too, because I saw and avoided two deer over the course of the stint. I just sat back a little bit, kept a hand on the wheel, and relaxed.

Every so often I’d have to steer the car. Maybe once in three to five minutes. Rarely did I offer any throttle or brake input. And it worked brilliantly. We tend to forget all of the little course corrections and throttle adjustments we make during even the most mundane of freeway drives. Semi-autonomous cars take all of that away. Instead, you’re free to simply pay attention to the road and the surroundings, to look around. Used correctly, autonomous features can make you a much better and more attentive driver over long distances.

I feel like a bit of a traitor to automotive enthusiasm when I write nice things about semi-autonomous cars, but the fact is that long-haul freeway trips are nobody’s idea of a great drive anyway. As long as we have the option to turn the feature off, I don’t have any issue with Autopilot. In fact, I can think of one killer application for it, yet to be implemented: Autopilot for trucks used to tow race cars. I have a ten-hour trip to NJMP after work tonight. If I could set the destination and go to sleep while my truck took me and the Neon to New Jersey, would I do it? Absolutely — and I would pay any price, bear any burden, or sign any disclaimer to do it.

[Image: Honda]

Jack Baruth
Jack Baruth

More by Jack Baruth

Comments
Join the conversation
3 of 121 comments
  • BiturboS4 BiturboS4 on Jul 18, 2016

    Curious what you think, Jack, of autonomous features in a manual transmission car? Does the extra element of having to downshift defeat the purpose of automated cruise control?

    • Jack Baruth Jack Baruth on Jul 18, 2016

      I don't think so. I think most cars can handle most freeway situations in sixth nowadays.

  • Maserchist Maserchist on Jul 18, 2016

    Autopilot was foreseen by the "Fabulous Furry Freak Brothers" at least 40 years ago. The system consisted of a hook on a tow bar conveniently attached to the vehicle in front of you. Hey, it was funny !

  • Dave Holzman My '08 Civic (stick, 159k on the clock) is my favorite car that I've ever owned. If I had to choose between the current Civic and Corolla, I'd test drive 'em (with stick), and see how they felt. But I'd be approaching this choice partial to the Civic. I would not want any sort of automatic transmission, or the turbo engine.
  • Merc190 I would say Civic Si all the way if it still revved to 8300 rpm with no turbo. But nowadays I would pick the Corolla because I think they have a more clear idea on their respective models identity and mission. I also believe Toyota has a higher standard for quality.
  • Dave Holzman I think we're mixing up a few things here. I won't swear to it, but I'd be damned surprised if they were putting fire retardant in the seats of any cars from the '50s, or even the '60s. I can't quite conjure up the new car smell of the '57 Chevy my parents bought on October 17th of that year... but I could do so--vividly--until the last five years or so. I loved that scent, and when I smelled it, I could see the snow on Hollis Street in Cambridge Mass, as one or the other parent got ready to drive me to nursery school, and I could remember staring up at the sky on Christmas Eve, 1957, wondering if I might see Santa Claus flying overhead in his sleigh. No, I don't think the fire retardant on the foam in the seats of 21st (and maybe late 20th) century cars has anything to do with new car smell. (That doesn't mean new car small lacked toxicity--it probably had some.)
  • ToolGuy Is this a website or a podcast with homework? You want me to answer the QOTD before I listen to the podcast? Last time I worked on one of our vehicles (2010 RAV4 2.5L L4) was this past week -- replaced the right front passenger window regulator (only problem turned out to be two loose screws, but went ahead and installed the new part), replaced a bulb in the dash, finally ordered new upper dash finishers (non-OEM) because I cracked one of them ~2 years ago.Looked at the mileage (157K) and scratched my head and proactively ordered plugs, coils, PCV valve, air filter and a spare oil filter, plus a new oil filter housing (for the weirdo cartridge-type filter). Those might go in tomorrow. Is this interesting to you? It ain't that interesting to me. 😉The more intriguing part to me, is I have noticed some 'blowby' (but is it) when the oil filler cap is removed which I don't think was there before. But of course I'm old and forgetful. Is it worth doing a compression test? Leakdown test? Perhaps if a guy were already replacing the plugs...
  • Crown No surprise there. The toxic chemical stew of outgassing.
Next