This is the title of a new book by Jeff Jarvis, (his blog here) one that triggered a really good (I’m not always in love with the magazine’s writings) Business Week story. Focusing on Detroit, the book and the story propose a revolution in automotive design: openness. Don’t design everything yourself, open your engineering processes, let customers and suppliers participate in the design effort. Google is used as the example of open design. Look at how they open themselves to user input by launching a product, say Gmail, as an open beta. Three opens in one sentence… As a result, Google benefits from early customer feedback and uses suggestions and bug reports to improve subsequent revisions.
Moving to eco-friendly cars, practicing what they preach and showing the open (that word again) way to Detroit, Google opened up the iconic Prius to make it a plug-in hybrid. The latter term refers to grafting a supplementary battery, one that, unlike the factory-installed unit, is recharged “from the wall” and offers more purely electric autonomy. You will recall the normal Prius only runs a few miles on its electric motor alone, and slowly. In everyday use, but for the slightest pressure on the “gas” pedal, the ICE (gasoline Internal Combustion Engine) kicks in. The Prius plug-in modification is the grand Silicon Valley Hacking tradition at work and in play: the “mod” requires getting into the Prius’ control software. (If you’re interested, try googling “prius hacking” or similar search terms; you’ll be impressed by the breadth and depth of the “hacktivity”. Very encouraging, a confirmation of the thesis that perversion is one reliable measure of success. See, for example, who invented the floppy, for what and where it ended up. The answer is IBM, to load microcode upgrades to its mainframes eons ago, and you know or knew the third part of the answer.)
Back to the Prius, Google and the auto industry. Let’s cloud the debate with a mixture of facts and opinions. I’ll try to keep the two as clearly identifiable as I can.
First, I always get nervous when the O word is thrown around. This is a sinner speaking as I got myself an OPEN MAC license plate when I landed on Apple’s R&D in 1985, when the Macintosh needed help from third-party manufacturers of hardware plug-ins. But calling something Open doesn’t make it so – or good. In the old days, Microsoft foisted a version of Windows on its flock, got vigorous feedback and kept issuing new versions. Calling something version 1.0 versus calling it a beta doesn’t mean the 1.0 product is more mature or less open to feedback. Leaving Microsoft to its Vista and Windows 7 travails, let’s turn to Apple. The company “enjoys” customer feedback – high volume, varying rabidity. Partly as a result of said feedback, partly as the self-driven development process, Apple ships between 100 and 200 megabytes of software updates every month to every user of Macs and iPhones. This without calling any of the products beta – or open. What’s fundamentally different from Google’s religion, pardon, posture of “perpetual beta” isn’t clear. Yes, in theory, Cloud software makes for a centralized update process in the supplier’s servers instead of in the millions of desktops and laptops. In practice, Google takes more time to “roll out” new Gmail features across its huge server infrastructure than for Apple to update millions of iPhones with a 200 megabytes download last week. In passing, let’s admire the power and reliability of CDNs (Content Delivery Networks) such as Akamai and, we’re told, Limelight used by Apple to deliver the updates.
Moving to cars, in Detroit, Hiroshima or Bavaria, the car design process has been a collaborative one for decades. Makers of subassemblies, ignition systems, brakes, suspension, glass, embedded computers all prod automakers to use their new and improved technology. In turn, car manufacturers prod automatic transmission makers such as ZF (Zahnrad Fabriken, “maker of toothed wheels”, gears) or the Japanese Aisin for more gears, double clutches, the latest fashion, fuel economy and lower weight. Automakers design or manufacture very few components of a car. Actually, several of a noted German manufacturer’s engines were designed by a consulting firm (not called Porsche). Another consultant, Giugiaro Design, works so deeply with its car making clients that, in one case, it went as far as designing the entire car, the production line, building said line and training workers. Also, manufacturers always go to great lengths probing reactions to future and current models through customer clinics.
So, you’ll ask, how come these Detroit open-minded, collaborative designers and customer-listening automakers got in such trouble? The answer isn’t they failed to do What Google Would Do but that They Are Not Google. In other words, they have a radically different culture, the set of below-the-consciousness rules and images that shape thoughts and acts without us being aware of the shaping. Google’s culture will not “take” in Detroit.
A slightly different angle on the same topic: Japanese methods and Detroit. Let’s move quickly past the Deming Prize. Edwards Deming, an American engineer, invented quality control methods adopted with glorious success by Japan Inc. Does this kill the notion of incompatible culture as a barrier to adoption of better ways? Not at all. Japanese culture was and is noted for its informed and demanding love of detail. Deming’s ideas were compatible with that culture. In the other direction, Toyota and others invented “just-in-time” manufacturing. They did it in full view, nothing secret. Did Detroit succeed in adopting this Japanese technique? Not really. Because the Detroit culture doesn’t allow for the deep level of collaboration, discipline and, let’s be frank, acceptance of authority, not to say subservience seen in the relationship of suppliers to a Toyota or a Honda. I’ll quickly point at a possible weakness in my argument: I visited Honda’s Marysville (Ohio) plant a few years ago. Red State American workers didn’t look confused nor resentful of the methods imported by a small cadre of Japanese supervisors. But Shoichiro Irimajiri, Honda’s US Chief, was prompt to point out the plant location: Nowhere, USA, far from Detroit…
There is hope for the auto industry, and it doesn’t all come in the form of Japanese factories in the US. The Business Week article mentions the acres of “metal” on dealer lots, meaning the huge inventories of unsold cars. The practice consumes cash and makes it difficult for customers to get the configuration, the mix of options they really want. Yet, “just-in-time” manufacturing and the Net ought to allow one to configure the car and get it delivered a couple of weeks later. Nice theory, dream on says Detroit. Enter the new Mini, BMW’s second incarnation of the British legend. You can order your very own car, one that’s unlikely to be like any other Mini, down to the checkers motif on the sideview mirrors, the color of brake calipers and the stripes on the bonnet, that’s British for hood, to say nothing of interior detailing. Then, still on the Net, you can follow your pride and joy’s progress through manufacturing and transportation across the Atlantic. See a customer’s tale here.
Critics will say this is good for detail-obsessed car buffs willing to pay for snob appeal. Indeed, it allows BMW to get a premium for their Mini while saving on inventories – all the while making their customers happy and proud, a well-debugged algorithm for failure. As usual, bleeding-edge customers are telling us where the mainstream will flow.
Let’s turn to back to Google and to Jeff Jarvis’ excellent book. In 1982, Tom Peters and Robert Waterman wrote In Search of Excellence. They sung the praise and described the methods of successful, excellently managed companies. A couple of years later, Business Week (Nov 15th, 1984, pp 76-88, Oops, Who’s Excellent Now?) surveyed the companies in the book. The result, a good third of these excellent companies were in deep trouble. (See the Criticism section in the Wikipedia article.) What did the authors miss? They probably walked in with a thesis and unconsciously fit facts to their narrative. We all do to some degree. More likely, they fell for the post hoc propter hoc tendency to see cause and effect where there is only a time sequence. “See, we changed the software-debugging process and customer satisfaction doubled as a result”. Perhaps it was just a couple of new versions already in the works and the new debugging process added nothing but a small amount of disruption. We fall for that too. Or the excellent results were the result of a transient, irreproducible combination of circumstances. A major competitor botched its last two products, leaving the market wide open to the lucky company. Or, and we either fall for that, or practice it with as much skill and energy we can muster, the company told Peters and Waterman a really, really good tale, with an optional connection to the company’s complicated circumstances — those were to come out a year or two later. Think Potemkin villages.
What I’m driving at, as you already guessed, is the somewhat random relationship between hagiographic books and their subject’s reality. I love Google, I use their products, I have friends and ex-associates working there and I believe, please note the verb, they’ve made insightful strategic decisions followed by skillful implementation. For example, the founders, Serguei Brin and Larry Page came up with an original idea for search, their page-ranking algorithm. Then, they went for big, powerful VC, even if they could have bootstrapped their business with less capital than Sequoia and Kleiner Perkins could muster. Later, Google worked on their future email service, they stumbled, recognized and energetically embraced the advertising tools that became AdSense and AdWords. All along, unlike less attentive or knowledgeable competitors, they really paid attention to the server infrastructure, built and rebuilt it several times, with the right hires, ending up with the world’s biggest and best server farm, more than one million servers today. Lastly, helped by their experienced, well-connected investors, they acquired adult supervision in the person of Eric Schmidt, the CEO, and a cadre of industry veterans.
Note I pay little attention to the 20% time for personal projects, the consensus building niceties, the free food, massages, dry cleaning, WiFi-equipped shuttle service and other perks. Nice but not guaranteed to last in the current depression. But the scale and quality of the server farm will.
Which ones of these official success factors are really so, I don’t know. What parts of their celebrated culture will one day bite them in the income statement, we don’t know either. But just as managing one million servers is qualitatively different from running hundred machines, the scale of an organization changes its culture, its success factors. The What Would Google Do? question will get very different answers tomorrow, calling in question today’s answers peddled as prescriptions for a better future. Look at Apple: it was once celebrated as a laid-back, undisciplined therefore creative company by one of its founders, Steve Jobs, who once floated a black pirate flag over his building. Today, the same, or not really the same company has become one of the more precisely run Silicon Valley giants, thanks to Tim Cook, the Chief Operation Officer, one of Steve Jobs’ key hires after he came back to build Apple 2.0.
Back to the openness argument, I’m not sure Google is a very open company. Some of what they do is publicized and/or released as Open Source code. But I don’t see the company very open, collaborative about key components of their server farm, we hear they make some of their own servers and might also be making their own switches, much to Cisco’s alleged and huffy annoyance. Nor do they collaborate on ranking bidding algorithms. Understandably.
One last fact.
Google had a fleet of plug-in hybrids, I saw the reserved parking spaces and the charging stations at our San Francisco office. They’re no longer there. The experiment has been concluded. This beta won’t ship for a while. —JLG