The first suspicion that something was wrong came in the late Eighties. I was a freelance journalist and the sole employee at an underfunded start-up making PC networks. I reviewed a version of Microsoft's MASM Assembler for ubergeek publication .EXE Magazine -- just as well, we couldn't really afford to buy the package -- and uncovered what looked like a few bugs. Some of them were quite serious. Had to be my mistake, of course, a company as big and technically proficient as Microsoft couldn't have let the product get out in that state.
I called technical support, saying that we were using this for commercial software and we needed an answer to our problems. We'll get back to you. Then I called the press office, saying something's up with tech support. We'll get back to you. I gritted my teeth and wrote a critical review. At least that would provoke some response, I thought. Twenty years later, I'm still waiting.
Then we had to make our products work with MS-DOS. They wouldn't, because MS-DOS did some odd things that weren't documented. We knew that products like ours could be made to work, because we saw similar software on the shelves, but Microsoft refused to divulge how. So we had to take apart both MS-DOS and the rival products to work out what was happening, found the magic, applied it. There was no reason for Microsoft to keep it secret, except to freeze out competition. Not that it was in competition with us -- it took a long time for Microsoft to wake up to networking -- but it was happy to pre-emptively lock people out.
These days, of course, we know about Stac, Go, Netscape, Digital Research, Be, Sun -- the list goes on. We know about the company being found guilty of anti-trust law in America, Asia and Europe. We know that without competition, Microsoft doesn't give a fig about product development -- wave your walking stick for the people at home, IE 6 -- and that with competition, it mostly cares about closing it down. This is old news.
We also know new stuff. Microsoft has got its people online, blogging away, and they report that while there are lots of good people doing great things within the company there is little chance of everything hooking up and coming out when required. Layers of management do what layers of management always do, taking the nasty spiky truth at the bottom and gently rebuilding it bit by bit with lighter, fluffier, more palatable fiction as it floats upwards. Bright ideas die on the vine. Things are, as you may have noticed, later than late.
You can't stifle innovation forever, though, and away from Microsoft's control zone the world is moving fast. This column was written on an Ajax-based web service -- not enterprise ready, but then neither was Word 1.0 -- running on top of an entirely open stack, and there's nothing that Microsoft can sell me that works any better for the job I'm doing. Last year, there wasn't a Linux desktop I'd be happy to give to my grandmother. This year, there is. Last year, there was no Microsoft product that made me glad I'd lived long enough to see it. This year, likewise.
Finding a way to prosper in the new world, which will be open, collaborative, distributed and virtual, is far more important. As it stands, the company could no more produce a roadmap describing its part in that world than it could build a motorway to Mars. But that roadmap is far more important than yet another version of Windows. Ray Ozzie knows this; he has the eye of an outsider experienced in big software, and will know to reinvigorate the troops at the same time as setting a new course.
That's the internal reinvention. The external one is just as important. Microsoft needs to build trust as partner and innovator, and that's the sort of reinvention that needs a generational change. Gates and Ballmer made a ferocious team; now Bill is off, the best thing Steve could do for the cause is to follow. He doesn't have to do it now -- to lose one of your head honchos is unfortunate, to lose both is perhaps careless -- but he does have to take his hand off the horsewhip.
Oh, and a fixed version of MASM would be nice.