Saturday, March 05, 2005

Well, the furnace guy came. In the end, he decided that it was my thermostat. He offered to sell me a new one for $150. I passed, opting to go to Home Depot and buy and install one myself for $49. He was actually cool with that, telling me that he doesn't get commissions on sales and even made sure I knew how to install a thermostat (I did). All done, and the house is nice and toasty again. 3:53:23 PM    ; comment [];  I have been hearing for days about Mark Lucovsky's infamous blog entry saying that "Microsoft can't ship software anymore." It's funny, because having lived many of the same experiences he has, I could think of many completely rational arguments to conclude that. But having just read it, I'm deeply disappointed that what he did say is shallow and not very thought-through at all. First of all, in comparing Windows to Amazon, Mark is making an apples-to-oranges comparison. Windows is client code, and amazon is a server-based service. Now, if we compare Amazon to MSN, we see something very different. MSN fixes bugs and posts updates all of the time -- like Amazon, they do it quietly so very few people notice. But let's even go back to Windows. Windows actually fixes bugs and distributes them in all sorts of ways. There's Windows Update, where particularly for security-related bugs things get turned around very quickly. There are other bug fixes and service packs, less frequently. And then thre are the big new releases with lots of minor bug fixes and lots of new features -- they happen over much longer periods of time. Why? Well, for two reasons. First, because Microsoft is in the business of building software that will work on a wide variety of hardware -- and combinations of harware from diverse vendors. And it's supposed to work with tens of thousands of applications, in over 30 languages. That is a huge test matrix, and it takes a long time to get that even close to right. Second, because our customers tell us that they don't want frequent new releases with lots of new features, because there is a real deployment and training cost for them. They would much rather have the releases spaced out so they can plan, test and deploy in the way that makes most sense for their business. Other Microsoft products, like Money and Encarta, ship on an annual basis and hit those deadlines like clockwork. And then there are business like XBox, where the business model itself has a built-in expectation that a console will be in a market for several years to amortize the up-front costs of R&D and setting up manufacturing (as well as to ensure a healthy and profitable aftermarket). What I expected Mark to say was that Windows is big -- really big, undoubtedly one of the largest software projects in the world, if not the largest. And it's been developed over a long enough period that the original authors of many parts of the code have since moved on, and maintenance of that code has passed on through several generations of owners who struggle with the challenge to maintain the insitutional history of the code, its design and architecture, and key decisions that were made along the way. The Windows team is facing design, construction, and maintenance issues that no one in the world has ever faced before. They do their best to make smart, well-informed decisions, but some days they struggle with issues that have no precedent to inform them. Every year, Microsoft spends more on R&D than it cost to send a man to the moon, and some of the company's projects (such as Windows Longhorn) are just as ambitious in terms of the scale and scope of the endeavor. Yeah, some days they stumble, and when they're working without a net that can get very messy. But I'm still at Microsoft after nearly 17 years because the company has never lost its determination to attempt really hard things. 3:30:02 PM    ; comment [];  Once a year, my team in MSR organizes a big internal event we call Techfest. We take over the conference center on Microsoft's main campus in Redmond for two days, and set up what you might call "geek heaven" -- about 150 demos and 25 lectures representing the best, coolest, and newest work in Microsoft Research's six labs. Techfest is open to all Microsoft full-time employees. Techfest was Wednesday and Thursday this week -- which explains why I haven't been blogging. I haven't been sleeping either. The event is company confidential, because we want to roll out most everything we've got and encourage frank, honest two-way conversations between our researchers and the people who work in Microsoft's product groups. Some of the things we show are well on their way to being incorporated into Microsoft products; the rest run the gamut from extensions of the current state of the art to wild, crazy ideas. There are no canned, hardened demos at Techfest; it's all research prototype code, and our researchers have burned a lot of midnight oil in the past month getting their demos to work. For their part, my team handles "everything else" so that the researchers can focus on their demos. Event planning, logistics, internal marketing, VIPs, giveaways, signage, staff shirts, and all that other stuff that makes for a successful event. This is the fifth year we've done Techfest, and we're starting to get good at some of the wacky stuff that you wouldn't necessarily anticipate you'd need to do. For instance: about 100 researchers who work at our remote labs come here and stay for at least a week -- which means that we need to find temporary office space for 100 more people. Also, an entire ecosystem has grown up around this event, as people try to take advantage of the opportunity of having everyone in town. So there is a whole set of pre-Techfest and post-Techfest meetings, workshops, and offsites that take place, and my team probably finds out about half of them. (I myself was booked for on the day before Techfest, for two separate ones the day after Techfest, and another Monday and Tuesday next week). The event has really taken on a life of its own. Despite being a confidential event, every year we offer invitations to a very small number of guests. A handful of them we ask to sign non-disclosure agreements and give them fairly broad access (though not complete access) to the floor. We also invite in a handful of reporters (not under NDA, of course) and show them a pre-selected set of booths for which we've filed all the patent applications and which aren't being kept secret by MSR or a product group for competitive reasons. We do this partly because we want to show off some cool MSR technologies, but also because there's a "story" in the event yourself: just walking onto the floor you realize that Microsoft is a company full of people who are super passionate about technology and the potential it has for transforming people's lives for the better. There is a buzz and an excitement on the floor that you have to experience firsthand to fully appreciate. In preparation for bringing reporters in, we think super hard about what we should show them. Part of that is to do our homework on the individual reporter and understand where their journalistic interests lie. Another part is to make sure that we give each reporter something different -- they don't want to write the same story that every other reporter is writing. It takes us weeks of planning just to get that all right -- since there are real limits to the demos that we can show, we put a lot of time into making sure we have the right stuff to make the trip worthwhile for each reporter. But it seems to have worked -- here's a story on some of the work we're doing on "the basics." Here's another on some very "hands on" (literally) research. Allison Linn wrote a story for Associated Press that got picked up widely -- the only problem here is that the headline makes people think that the teddy bear is being designed as surveillance or babysitting for kids, which is not at all the intent; the demo was just intended to show using vision technology to track faces and have a teddy bear's head movements follow faces and mimic some lifelike reactions to people moving around it. We were also privileged to have a reporter fly all the way from India to cover the event and apparently even the Drudge Report had a link... plus we got articles on the front page of the business section of both the Seattle Times and Seattle P-I. I've been watching the blogs the last few days, and as always when Microsoft is involved the comments are mixed. Generally I expect that and it doesn't bug me. The most disturbing thing though, once again, is the whole "evil spy robot teddy bear" thing which is getting blown further and further out of proportion as it makes the rounds of the blogosphere. It's not "watch" in the sense of "surveillance," it's "watch" in the sense of  "it can track your head as you move around." It's a TOY. It's an experiment in how to make a toy that's more fun and more engaging. Plus, this is EXACTLY why we do research ahead of product development -- so we can try wacky things and find out early whether they are actually good ideas or not. We had several MS execs come through and spend several hours at Techfest too, which is always great fun. My two favorite perennial guests are Bill Gates and Jim Allchin. Jim, in case you are interested, has a Ph.D. in computer science -- he did some important work on a distributed computing system called "clouds" . Both of them look like kids in a candy shop as they walk around the floor. Both Bill and Jim ask our researchers tons of hard questions, which is equally a thrill and a source of stress for them, but in the end a wonderful experience -- imagine being a young researcher from our Beijing lab and having a 10-minute conversation with Bill about your latest creation. It's the thrill of a lifetime for them. Techfest came off nearly flawlessly this week. We had about 6000 MS employees come through, which is a huge success for us. We were thrilled with event, and we're equally thrilled that it's now done. :-) 9:24:31 AM    ; comment [];  My furnace is broken. I woke up at 6am yesterday morning to a cold house. My furnace had reduced itself to a very efficient cold-air circulator; the fan works great, but verily, no heat doth emanate from its heavenly registers. and of course, I'm a software guy, so I'm as useless with a furnace as I am when something goes wrong with my car. I goofed with the thermostat to cycle it a couple of times, with no luck. So with an hour before my kids get up, I tried one last thing: I went downstairs to the furnace room and unplugged the thing. when I plugged it back in, voila! the heat came back on and everything was getting toasty again by the time the kids got up. Last night I got home to the same problem. This time, no luck with the "reboot the furnace" trick. Based upon talking to a guy on the phone last night, the electric igniter is probably shot. Fortunately, the weather is unseasonably warm in Seattle right now, so while I'm bundled up, I'm not risking hypothermia. And the repair guy should be here in 15 minutes... 8:16:26 AM    ; comment [];

© Copyright 2005 Kevin Schofield. Last update: 4/7/2005; 5:27:39 PM. This theme is based on the SoundWaves (blue) Manila theme.

Blogroll

Robert Scoble
Ross Mayfield
Dan Gillmor
Larry Lessig
Joi Ito
misbehaving.net
simplegeek
Ray Ozzie
Doc Searls
Boing Boing Blog
Paul Andrews
Chris Pirillo
Halley's Comment
Instapundit
Scripting News
Beyond the Beyond
Don Box
Mary Jo Foley
WebMink
Dean for America
kuro5hin
John Batelle
PDA and Tablet PC News
AlwaysOn Network
The Old New Thing
PR Opinions
Critical Section
Backup Brain
Seattle P-I Microsoft Blog

Books I Like

The Diamond Age
Cryptonomicon
The Tipping Point
The Art of Human-Computer Interface Design
The Design of Everyday Things
Contact
Earth
The Mythical Man-Month
Peopleware
Wicked French
Linked: The New Science of Networks
As the Future Catches You
Pattern Recognition
The Da Vinci Code
The Man who Stayed Behind
Angels and Demons
The Confusion
Quicksilver
Free Prize Inside


What I'm Reading Now

Fluke