Is Tesla a car company?
I recently had the opportunity to try out a Tesla Model X, the company’s top-end model. The opportunity did not come from Tesla, with which I maintain a good relationship, but through Móviles.com, a Spanish internet company where some people read my page on a regular basis. The vehicle was equipped with Autopilot 2, the second version of Tesla’s Autopilot, which is still in its development phase and with multiple functions in beta.
As I have said on other occasions, I am not specially a car person, and my interest in the automotive sector is based on the incredible pace of disruptive innovation it is experiencing. Very few industries have found themselves in a situation in which the main component of their product (in this case, the internal combustion engine) changes, (to a battery-powered engine), as well as the way it is used (human driving, assisted and autonomous) as well as ownership models (from product to service. The opportunity to study this process close up is a real privilege. But for me, driving a Tesla is about technology, not about performance.
That said, the vehicle is incredible. I won’t go into details: there is plenty of stuff on the internet. My job was to test certain functions and to understand what differentiates this brand from others. It was not the first Tesla vehicle I had driven: the former was a Model S without Autopilot. In both cases, the engineering, the attention to detail or the interface of use was excellent: this is clearly a high-end vehicle.
The Autopilot 2, for those who do not know the history in detail, is not only the second major revision of Tesla’s assisted driving function, but the first after the only fatal accident of a driver using its predecessor. Joshua Brown’s accident changed many things for Tesla: not only was he an unconditional fan and a person very dear to the company, but his death could be related to problems derived from certain sensors of the Israeli company Mobileye, recently acquired by Intel. Tesla worked with Mobileye systems until July 2016 when, following the Brown accident, the Israeli company said its automatic braking systems would not recognize oncoming vehicles that turned side on until 2018.
In reply, Tesla, which was, according to that statement, potentially responsible for the accident for using an inadequate sensor, commented that its systems used not only those sensors, but dozens of components to determine the appropriate response to any given scenario, and announced a break with the Israeli company and its decision to develop sensors in house using components supplied by Comma.ai, the company created by George Hotz. Although it was never clear who had broken with who, shares in the Israeli company fell by 7%, despite Tesla not being a major customer from a quantitative perspective. The subsequent investigation by the National Highway Traffic Safety Administration (NHTSA) into the accident blamed the driver, who was watching a Harry Potter movie rather than the road while cruising at high speed. It also concluded that Tesla’s Autopilot was responsible for a 40% reduction in accident rates.
The Autopilot 2 is Tesla’s attempt to rebuild a power steering system from new components, with all that entails in terms of testing and adaptation: in many ways, the functionality of Autopilot 2 when I tried it was still lower than that obtained with Autopilot 1, but since then it has been updated. In any event, my experience was consistent with what I would expect from a beta product in beta: in Germany it cannot be called Autopilot. It’s a very powerful driving assistance system that allows the vehicle to circulate almost autonomously on many occasions, but it’s not an autopilot. In fact, in my enthusiasm to test it, I experienced a few scares, such as its inability to “see” certain obstacles. Make no mistake: we are talking about the safest vehicle in the world, the best place to be sitting in the unfortunate event of having a car accident.
But when I tested Autopilot 2, it reflected the mentality of a technology company, not an automotive one: a product is launched and then updated and improved when it is already in the hands of customers. No car manufacturer I know of would do that. But given the low incidence of system-related crashes, Tesla does so for one reason: drivers are aware that they are part of a system being constantly improved thanks to their input.
My impression of Autopilot 2 was relatively disappointing. Calling something Autopilot when it isn’t, no matter how much it hopes to be in the not too distant future, is a little unsettling. When, driving under normal circumstances, the system cannot be activated more than a third of the time, or worse, when I have the feeling — and not entirely unfounded — that my car cannot see the cyclist or motorcyclist in front, the only conclusion is that this is a promising product, but also potentially dangerous. After a few days, I got used to the idea that this is not an autopilot or even close to being one yet, and that whatever its superiority over other vehicles, one needs to drive as one would in any other car. In the traffic of Madrid, where it has little experience, things can get chaotic: lanes tend to be significantly narrower than in the average US city, and are sometimes misaligned at intersection, so the distance between other vehicles in Autopilot can be a little unsettling.
Driving the Tesla Model X has been a wonderful experience in every way: without a doubt, it is one of the best cars I’ve ever been in. But above all, I was struck by the difference in approach between the vehicles I have been driving and testing my whole life, built by traditional automotive brands, and this. The difference starts with the smallest details of the interface, which is as simple and intuitive as any well-designed app, and continues with the idea of ”I am participating in a beta-test” that you can obviously avoid if you prefer. Tesla’s are updated regularly: it is common to get into your vehicle in the morning and find a notice that it has been updated in OTA mode overnight, accompanied by information on the changes and a menu that allows you to see new functions in beta. The philosophy of testing functions in beta is core to the brand.
Moving a big, heavy vehicle like this from a smartphone app is impressive and certainly a completely different approach to other brands, and is just one example. While Tesla assumes that its users will be smart enough to test functionality prudently and use their judgment, the traditional car industry has for many years assumed that their customers are fools and that if any functionality can be related to a design flaw, they will find themselves in court. Tesla’s impressively low accident rate seems to validate the system, but on the other hand, it is not without risk and susceptible to problems in the event that a customer decides to take legal action over a fault. But then rather than customers, Tesla seems to have fans.
What else to say about the Tesla Model X? The design of the interface is fascinating, the overall usability is fantastic, it has self-opening doors with sensors that prevent them from hitting something or somebody nearby, along with many other details in the build that clearly put this vehicle in the luxury market. But above all, it is obvious that this is not the product of a traditional carmaker, but a technology company, or even a battery manufacturer, or maybe something else entirely. In short, Tesla is not for everyone: neither in terms of price or its philosophy. We will see how much of that philosophy is retained after the launch of the Model 3.
(En español, aquí)