There is a law in business organization known as Conway's Law. This states that: Organizations which design systems...are constrained to produce designs which are copies of the communication structures of these organizations. The law was stated by Melvin Conway in 1968 so is nearly 50 years old. The corollary runs the other way, namely that organizations produce products that reflect their organizational structure. Conway was a computer scientist, although that term was not used back then, and he was mainly thinking about computer programs. Eric Raymond, another computer scientist who coined the term "open-source software," pointed out a more concrete example: if you develop a compiler with four groups, then you will get a four-pass compiler. Car companies are the same. They are organized the way cars are put together, with an organizational hierarchy that reflects that cars have bodywork, engines, interiors, electronics, suspension, and so on. In fact, little electronics is done inside the car companies. By and large, they spun out the electronics into separate companies: Delphi out of GM, Visteon out of Ford, Denso out of Toyota. Even at the time this seemed a little short-sighted since it was clear that the electronic component of cars was only going to get larger, as it has. In fact, a cynic might say that they did the right thing, they just kept the wrong part. But the growth of electronics in cars is only just beginning. Fully autonomous vehicles may still be some way off, with legal and societal issues as important as technical issues. But the intermediate stages, known as Advanced Driver Assistance Systems, or ADAS, is already here. The most advanced system today is, I think, the Tesla Autopilot. This was turned on just a few months ago, although Tesla has been shipping cars with all the required sensors for a couple of years. In busy traffic, on freeways, and on rural roads, the car can drive itself. The technology relies on cameras and sensors and, compared to a normal in-car network, high-bandwidth networking and high-performance computing executing complex algorithms. It is significant that Autopilot came from Tesla. Since they only build electric cars, their expertise is entirely in electronics and software. Not entirely, they have to build electric motors too, but compared to building a state-of-the-art internal combustion engine, this is straightforward and well-understood (and simple, as there is one moving part). Building the bodywork of a car is clearly not that difficult. Tesla mastered it quickly. Other companies often delegate actually assembling the vehicle to subcontractors. For example, a lot of Porsche Boxsters are assembled not in Stuttgart but by Valmet in Finland, a company you've probably never heard of. But this type of electronics and software expertise is not the expertise of traditional automotive manufacturers. The transition from the importance of the internal combustion engine to the importance of electronics and software represents a major challenge to the incumbent automotive companies (or in the jargon of the industry, the OEMs). First, they already lost a lot of their electronic expertise by spinning it out into separate companies (the Tier 1s). Even the tier 1s do not have good knowledge of video and autonomous vehicle algorithms. They are scrambling to acquire it. Bosch just acquired a team from ST to do vision ADAS products, Delphi just acquired Ottomatika (basically the CMU team from the DARPA challenges), and it seems safe to guess that there will be more such acquisitions since the Tier 1s have to acquire existing teams. They don't really have the capability or the time to build it up in-house. But the biggest challenge to the automotive companies is that they need to reorganize their companies from top to bottom to reflect the new reality. The need for knowledge of the internal combustion engine will not go away any time soon, but it is clearly not the way of the future. Nobody doubts that they can master building electronic motors; indeed, every hybrid car contains at least one. The bigger challenge is that they need to build up capabilities that they currently do not have enough of, and make them the front and center in their organizational structure. It is never easy to restructure a big company and there tends to be an aspect equivalent to generals always fighting the last war. The CEOs of automobile companies used to get there by operational expertise running factories. Today many have come up through the design and marketing side, but none of them have electronics in their veins in the same way as Tesla does. Just this Monday, Daimler Chief Executive Dieter Zetsche told German weekly Welt am Sonntag that a recent trip to Silicon Valley revealed that Apple and Google have made more progress on automotive projects than he had assumed. Mercedes and BMW have some of the most advanced projects of ADAS and it sounds like they are looking for partners here. He told the paper: Our impression was that these companies can do more and know more than we had previously assumed...There were concrete talks. I will not say anything about the content. It was not just about the fact that there is an innovative spirit in the Valley. We know that already. We wanted to see what drives it, and all the things that can be created from it. I expect that some of the automotive companies will survive the transition, perhaps through acquisition. It wouldn't surprise me in the least if GM re-absorbed Delphi, for example. But I suspect not all will survive unscathed through such a major change of their business.![]()
![]()
↧