So sayeth the Pogue. đ
[Microsoft organization impact after the jump.]
The review skirts around an interesting aspect of Microsoft. At the product/engineering level, Microsoft segments the development team of a product into a number of autonomous programming teams, each composed of a small number of engineers and three managers: technical lead, product manager, program manager. By keeping the programming teams small and making them largely independent, they avoid the Mythical Man-Month problem, which comes about due to reality that the number of possible interactions in a programming team increases as n2. At least until they hit the integration stage (usually during a build and QA partsâbut the idea here is that building and testing can be automated).
Have a larger product? Donât make a team bigger; just add more teams.
This structure is optimized for âreview winsâ in the old-school sense. As long as software reviewers rate software as a set of competitorsâ features that you are missing and a set of bugs and dislikes, as long as these new features and bugs can be distilled into a set of tasks, as long as you can add more programming teams to the product, you can win with the Microsoft method. On a basic level, simply list them out, sprinkle them arbitrary among your teams, set a deadline, integrate and release whatever has passed QA, and iterate until the reviewer gives your product 5 stars/mice/what-have-you.
In the case of Windows Vista, this process is creating an increasingly slower development cycle, produces a outrageously bloated product that is starting to look very schizophrenic. The long development cycle conspires to make the copy-to-win strategy overt. And yet, it has historically worked well for products like Windows, Word, Excel, Outlook, Visual Studio, Internet Exploder, what-have-youâŚ
On the other hand, it failed miserably for PhotoEditor/PhotoDraw vis-a-vis Photoshop/Illustrator; SQL Server vis-a-vis Oracle, DB2, and MySQL; IIS vis-a-vis Apache; ASP and .NET vis-a-vis PHP; Ultimate TV vis-a-vis Tivo; Xbox vis-a-vis Playstation; Windows Media Player/Movie Maker/Photo Gallery/DVD Maker/etc. vis-a-vis Apple iLife, or the Zune vis-a-vis iPod. It also has yet to succeed with any free internet service (MSN, Expedia, MSN Search, MSN SpacesâŚperhaps Hotmail might be an exception, but if so, itâs probably an exception that proves the rule). And this all when leveraging the OS and Office monopolies to the max which generates a near-infinite reservoir of cash and a bundling platform that is the envy of the entire industry.
HmmâŚ
Hotmail was bought by MS when it was the market leader…
Rob. Great point. Also they built their infrastructure on Solaris and are probably organized significantly differently. Thatâs why I said they may be the exception that proves the rule.
If there’s one brutal truth to learn in Silicon Valley it is that it takes more than great engineering and great technology to make great products. Microsoft’s shortcomings and (relative) failures haven’t usually come from poor engineering. Vista is an unusual case where people are pointing their fingers at the engineers for slipping the dates by so long , etc. I don’t know if that is fair or not.
Apple has proven that even bad engineering and technology can be overcome. The first couple of versions of OSX were complete crap. I still their scheduler and threading leave a lot to be desired. But OSX still had enough good ideas in it that you could excuse a lot of the poor execution.
Nice work, OSX.0 was pretty lame and didn’t even look that good compared to 10.4, but hey, even then i knew it was going to be great. It was apple!