Microsoft’s Qi Lu Talks Future of the Web: Look Out, Facebook, Groupon, Apple, and (Oh Yeah) Google

5/17/11Follow @gthuang

(Page 2 of 4)

understanding of the importance of the rate of heartbeat of an organization. One [aspect of this] is hypothesis validation. Any innovative ideas, whether these are ideas about a new algorithm or particular techniques or a feature that feeds into your search ranking, all at some point require you to verify a particular hypothesis. Even user-facing visible features—-small things, different fonts, different colors—require hypothesis validations. The rate at which an organization can validate hypotheses is one of the hallmarks of a heartbeat.

But one thing I didn’t learn until a few years along that journey is, the talent base you have in an organization—the tools, the processes that enable you to formulate high-quality hypotheses—is equally if not more important. So to summarize one of the learnings I can bring to the table, it’s to help an organization to raise the overall heartbeat.

For me personally, one thing I bring with me is a relentless quest for excellence. It may take time, it may take a long journey, but anything we decide to do, we will settle for nothing short of being the very best. It’s a reflection on the personal level of what I would like to pursue.

X: Critics say Microsoft is too big to move fast. Could you give some examples of what you’ve done to raise its heart rate?

QL: This needs to be put in the specific context. I have this notion of “innate rhythm.” You want to design your cadence, your heartbeat, to tap into the innate rhythm. If you go overly fast, you’re not actually able to be in tune. The key thing is you keep the rhythm that’s in harmony; you don’t want to spin unnaturally faster than the natural rhythm. Different businesses will have different innate rhythms. For all the businesses in the area I’m responsible for at Microsoft, they all have different rhythms. I’m responsible for Microsoft’s data centers, networks, some of the key infrastructure; the innate rhythm for those segments of development is actually much longer cycles. That’s how the whole value chain holds.

Online services is iterative by its nature and its heartbeat is very important. For example, [take] search quality. If you ask the question, “who decides search results (relevance ranking)?”—our view is it’s not us, it’s the collective producers and consumers of the Web that decide what are the right results given particular queries. If you take that perspective, then our job is to be very nimble, in tune with the ecosystem of people producing content and consuming content and services on the Web. So search quality is fundamentally iterative. If you [look at] search logs and you take any particular period of time—a day, a week, or a month—there’s always between 25 percent to one-third of the queries you’ve never seen before. That indicates that we as a society, collectively, are always looking for new things. The high-quality results always need to be evolving. What we need to do is raise the heartbeat so we can always be in tune with the ever-evolving user pursuit of what they’re seeking in the digital universe.

X: But to enable the future, you have to put in new processes and make some changes. What are the most important things you are implementing?

QL: One is infrastructure—storage and compute. One of the important aspects of online services is you have a low-cost way of capturing “digital observances”—every view, every click, every mouse move. You capture those and you run a lot of computations on those. That becomes the baseline for you to be able to formulate and validate hypotheses and do analytics.

We also need to build “enabling capabilities”—usually a combination of talent, skills, and tools. These capabilities become a force multiplier. If you are good at those, it spreads out and … Next Page »

Gregory T. Huang is Xconomy's Deputy Editor, National IT Editor, and the Editor of Xconomy Boston. You can e-mail him at gthuang@xconomy.com. Follow @gthuang

Single Page Currently on Page: 1 2 3 4 previous page

By posting a comment, you agree to our terms and conditions.