Why Microsoft Can’t Make it Work
john posted in memes, technologies on July 1st, 2008
Not Just Another Bash-Microsoft Session
Nice things will be said about Microsoft.
(afterthought: today it was announced that Microsoft has started funding Apache, see end of this article for more)
I ran across this MS research project page recently when I googled “intelligent adaptive interface.” I was looking to see if anyone was working on making computer and small device interfaces that adapt to the user’s behavior. What I had in mind was using one of the newly-ubiquitous tiny Linux devices as a remote control. On my conventional low-budget stereo and tv devices, the remotes have a gazillion buttons, with no logical placement and no hierarchy. The labels are invisible in the dark and require reading glasses in the light, and inscrutable when actually read. You all know the drill.
Well this very scientific and academic page shows a whole series of Microsoft Research projects on the general notion of adaptive interfaces. At a glance you can tell this is Serious Stuff done by Serious but Nice People ( their pictures are there). They are well-organized, use big-time ideas like Bayesian filters and Matrices, 3D interfaces, “a state-of-the-art psychological studies laboratory,” and so forth. They get to re-examine human-machine physical interfaces, and all sorts of interesting problems.
So what have they come up with?
The infamous talking paper clip.
Yep I am not kidding. OK there is a long list of other things of more substance from implementing encryption to new 3D rendering filters.
Now the MS people I have met have been really smart, and I don’t doubt for a minute that working with these folks would be a real intellectual thrill. But I assure you, bet money on it, they will never solve the remote control button problem, or even improve the visual mess that is Windows. How come I know this at a glance, without any special knowledge? You look at that page, what do you think? And chime in if you disagree.
So why can’t they make it work?
They have meetings.
They have too many people and too much money.
Whatever they do will be diluted and absorbed by the layer after layer of other groups just like them at Microsoft.
They start by going back and questioning premises about human psychology.
They have no test track and feedback loop.
They aren’t hungry.
Feedback
For me, the most interesting of these obstacles (because it could, in theory, be overcome at a big company), is the lack of feedback and test track. The rest are the result of working in a company with 30,000 other smart people, where overhead approaches 100% of total effort. What’s feedback in this case? Feedback is not focus groups. Useful feedback comes from actually writing some software and letting it loose on the public to see if anyone likes it, and actually paying attention to what they say and do. The test track is the venue where you can get good feedback with high frequency.
Let’s look at an analogy, motor racing. In racing you have an absolute test: the order of finish. You have a set of rules, intentionally designed to narrow the problem with constraints to the point where it becomes partially soluble, remains quite difficult, and always has room for improvement. You have competitors and a prize which provide motivation. Good engineering practice is reinforced in racing because at every level there are painful time and resource constraints.
Let me use a type of racing I am familiar with to explain why good practice is reinforced:
In motorcycle road racing a team typically races one or two times a month, but they only get to the track on Tuesday sometimes even Thursday before a Sunday race. Because of cost constraints of many kinds, they only get a few hours of practice, and then qualifying time on the track. Every track is different and requires a new chassis setup, and last year’s settings won’t work, because too many other factors have changed. There are multiple variables to set, and they have unpredictable interactions. Teams have no ability to use the track at other times for testing. There is an unshakable deadline, and somebody wins.
So does this incredibly difficult process produce better engineered bikes?
Lap times get faster every year. C’est tout.
Back to Microsoft, just how are they ever to get this sort of feedback, testing, and motivation? The giant company could spin off into small teams and create a competition. But then what is the test? Microsoft can’t afford to release a bunch of half-baked ideas to the public ( cynics disagree, but I promised not to mock ), so the test is inevitably arbitrary and maybe political. This engenders a kind of corporate navel-staring which can’t produce good work. So no, I don’t think they can innovate, because too much is at stake.
What about the famous iPod, iPhone model of innovation at Apple? The market has clearly voted for that in a big way. Apple is a huge company that nevertheless finds it possible to innovate. The Apple Macintosh approach to the personal computer was to control everything, the hardware, the operating system, the user interface, in the name of consistency, reliability and integration. All of which are other ways to say “control.” Yes Apple is more of a control-freak than Microsoft.
A recent Wired magazine article tried to take on the culture at Apple, inevitably making the question into the culture of Steve Jobs:
Jobs’ fabled attitude toward parking reflects his approach to business: For him, the regular rules do not apply. …
… Apple is … more like an old-fashioned industrial titan than a different-thinking business of the future. Apple operates with a level of secrecy that makes Thomas Pynchon look like Paris Hilton. It locks consumers into a proprietary ecosystem. And as for treating employees like gods? Yeah, Apple doesn’t do that either.But by deliberately flouting the Google mantra, Apple has thrived. When Jobs retook the helm in 1997, the company was struggling to survive. Today it has a market cap of $105 billion, placing it ahead of Dell and behind Intel…..
It’s hard to see how any of this would have happened had Jobs hewed to the standard touchy-feely philosophies of Silicon Valley. Apple creates must-have products the old-fashioned way: by locking the doors and sweating and bleeding until something emerges perfectly formed. It’s hard to see the Mac OS and the iPhone coming out of the same design-by-committee process that produced Microsoft Vista or Dell’s Pocket DJ music player.
( I have cut out a rather silly comparison to Google)
This probably isn’t the most profound commentary on Apple, but it does try to get at the culture issue that is clearly different, if a bit distasteful. Jobs is famously a screamer, possessed of a legendary “reality distortion field,” in short, a real prima donna. I worked a long time ago for an outfit with a head man like that, The Santa Fe Opera under John Crosby. He ran the opera like a fiefdom, it was housed on his ranch North of Santa Fe, and he usually decided by fiat that he would conduct one of the operas each season. He used his own stable of buddies to conduct, design, and direct. He was known to summarily dismiss even lowly employees for simple offenses like a noisy muffler. Crosby was at best a mediocre conductor, and his taste could veer into the campy and maudlin, but it’s impossible to conceive of the SFO without him. He created a world-class summer residency for opera out of the juniper, sagebrush, and rattlesnakes of New Mexico. Santa Fe was interesting. It was something to sit outside ready for the incredible New Mexico stars, watching a thunderstorm approach down the Rio Grande valley, and have the orchestra break into the Der Fliegender Hollander, or Lulu, the most depressing show I have ever seen, or Falstaff. At its best, it was thrilling, and people cared.
The various Apple iThings were brand-new products, and Jobs bet a good part of the company on them. Poker players call this kind of gamble, “all-in” and it requires an appetite for risk, and an environment where aggression is rewarded. Other companies don’t seem to be able to follow, even when there is such a clear example to copy. Let’s not just dump on Microsoft, why can’t the Japanese or Koreans or Finns make an mp3 player or cell-phone as good as an iPod or iPhone?
Hell, why can’t anybody even make a good remote control or car radio?
Microsoft goes Open Source?
an afterthought…
Well the cnet article points out previous areas of cooperation between Microsoft and the Open Source world, and favorable differences in the Apache license and the GPL used by many other open applications. So its not quite a shocker. But never mind these inside-the-silicon-axis matters, does this mean Microsoft might have to open up its platform to be able to innovate at all?
Windows is in a pinch. World-wide its tremendous market success has paradoxically left it with few new markets to expand into. Sure the Chinese market may eventually be as big as Europe and America combined, but its not at all clear that they will pay for the software they use. If somehow forced to start observing international IP standards on piracy, they might just use no-cost Linux instead. The latest Windows, Vista, has been a flop technically and in sales. In addition, the hot new market is phones and tiny devices, where Apple owns the flashy-bits and Linux, with help from Google and others, may be the gorilla in the room.
An admittedly far-fetched scenario:
If Microsoft moved very quickly and transformed Windows into a windowing application that could run natively on Linux, and provided software infrastructure ( libraries) for the applications developers to port to the Linux kernel-Windows hybrid, they might be able to drop thousands of developers from Windows, and replace them with some funding for the various Linux groups they’d need. Windows users would get the same look&feel, and Microsoft could probably charge just as much for a product with much smaller development costs. They won’t do it because its risky.
Not doing it might be just as risky, since the Open Source world is catching up quickly in all areas. Another far-fetched idea of three years ago, that one day Linux and friends might be a real competitor for Windows, doesn’t seem as outrageous today. Another and more likely option for Microsoft is to stay the course and milk the Windows franchise for as long as it can, and meanwhile focus on games, small devices, and phones. This might be more profitable in the near term, depending on how much labor costs could be saved or re-used on future-oriented development. It would be much easier for management to do, and therefore its the likely choice for them.
and 
(opt) USB CD/DVD drive 
Cody gives Zonbu four biscuits. 
Bugatti Veyron
2007 Corolla