On Monday, Apple unveiled iCloud, a new service for remote storage of user data. Some people, including our own Jon Stokes, are skeptical of Apple's chances of getting iCloud to work at scale. And history seems to be on their side. iCloud is at least Apple's fourth attempt to create a viable cloud computing service. The previous incarnations included iTools in 2000, .Mac in 2002, and MobileMe in 2008. As Fortune wrote about MobileMe a few weeks ago, "MobileMe was a dud. Users complained about lost e-mails, and syncing was spotty at best." iTools and .Mac were not exactly resounding successes either.
Apple's perennial difficulty with creating scalable online services is not a coincidence. Apple has a corporate culture that emphasizes centralized, designer-led product development. This process has produced user-friendly devices that are the envy of the tech world. But developing fast, reliable online services requires a more decentralized, engineering-driven corporate culture like that found at Google.
Designers in Charge
Apple has always been primarily a user interface company. A good UI has a minimal set of consistent, well-designed interface elements that economize on the user's scarce attention and take advantage of her muscle memory. The first iPod, for example, wasn't a major breakthrough from a technological perspective (indeed, a lot of commentators sneered at it when it was released), but the experience of using it was vastly superior to other music players on the market at the time. Removing unnecessary features can be more important than adding new ones. Steve Jobs's obsession with removing clutter can be seen in the one-button iPhone, the no-button MacBook trackpad, and the aggressive removal of outdated Mac ports.
Good UI design establishes a consistent visual vocabulary that's used across the product line. Apple uses concepts like column view, a blue circle to mark "new items," and a pentagonal "back" button across iPhone and Mac applications. And it aggressively enforces its UI guidelines on third-party applications.
Apple's UI successes are made possible by its designer-centric corporate culture. Former Apple CEO John Sculley has described designers as the most influential people in Apple's product development process. Centralizing control over product development in the hands of a single designer enables Apple to consistently produce beautiful, user-friendly products.
Google's scalable culture
But this same centralized, designer-driven culture can be a serious weakness when building scalable network services. A good way to illustrate this is to contrast Apple's culture to the dramatically different culture at Google. Google's flagship product barely has a user interface at all. A search engine comes close to the programmer's ideal of a mathematical function: it takes a text string as input and produces a list of URLs as output. When Google entered the market, it didn't even try to distinguish itself from early competitors based on its UI.
Instead, Google distinguished itself by the breadth of its index and the speed and relevance of its results. The success of the search engine, and many of its later products, was largely due to Google's obsession with scalability. And Google's structure and corporate culture are optimized for building scalable systems. Google is organized into small, autonomous teams of engineers who are empowered to make decisions with minimal oversight. Google managers are given dozens of subordinates, which prevents them from micromanaging even if they want to.
Rather than having intimate knowledge of what their subordinates are doing, Google executives rely on quantitative measurements to evaluate the company's performance. The company keeps statistics on everything—page load times, downtime rates, click-through rates, etc—and works obsessively to improve these figures. The obsession with data-driven management extends even to the famous free snacks, which are chosen based on careful analysis of usage patterns and survey results.
This decentralized, data-driven management philosophy often produces cluttered, mediocre user interfaces because there's often no one with the authority to impose a single UI vision on the entire product. But it makes Google adept at building and maintaining the complex infrastructure necessary to offer fast and reliable online services. Getting cloud services to work at scale requires careful attention to thousands of technical details—far too many for any single person to understand or even be aware of.
Improvisation is also crucial. No amount of prelaunch testing will reveal that a network service has scaling problems, because only millions of actual users will push the system to its limits. So engineers frequently have to improvise, making rapid adjustments as usage grows and performance bottlenecks are discovered. Rank-and-file Googlers need to fix problems in their own corner of the Googleplex before anyone else in the company (or the world) even notice them. The company's bottom-up structure gives them the freedom and the resources to do that.
Fourth time's a charm?
In contrast, Apple relies heavily on prerelease testing to make sure its products are perfect when they leave the factory. That works well for iPads and MacBooks, but it's impossible for a networked service. No manager can know enough about a complex online service to fully vet it for flaws. And no amount of internal testing can replicate the experience of having millions of users actually using the product.
Corporate culture matters, and it changes very slowly. Every company is good at some things and not others. Apple is much better at building user-friendly devices than Google, while Google is much better at building scalable network services. The competition between them for dominance of the mobile OS market may ultimately turn on which of these skills proves more valuable to consumers.
Steve Jobs is apparently hoping that iCloud will break Apple's losing streak on building cloud-based devices, and he's hoping that the tens of millions of dollars the company has poured into its gigantic new data center in North Carolina will make the difference this time. But he may be stuck with the strengths and weaknesses of the company he has built over the last 35 years. We wish him well, but we won't be surprised to see stories in the coming months about iCloud having performance and reliability issues.
reader comments
152