Microsoft’s downfall

A Vanity Fair article tries to pin Microsoft’s downfall on Steve Ballmer:

How could a company that stands among the most cash-rich in the world, the onetime icon of cool that broke IBM’s iron grip on the computer industry, have stumbled so badly in a race it was winning?

First of all, Microsoft was never an icon of cool. Unless you happen to live or work in Redmond I’m betting you’ve never seen anyone wearing a Microsoft T-shirt, let alone someone cool. I bet you’ve never seen a trendy sports car with a Microsoft logo bumper sticker. Yeah, there was that one guy who got a Zune tattoo, but he subsequently had it covered up.

But on to the meat of the issue: I don’t think Ballmer had anything to do with Microsoft’s downfall. I think Microsoft was doomed before he got the job.

Around 1998, I read some market research about Microsoft. A company had surveyed business IT decision makers, and one of the questions they had asked was: “If you could choose one company to never again do business with, which one would you choose?”

I forget the exact number who had said “Microsoft”, but I recall it being between two thirds and three quarters. Well over half, though.

I don’t have an MBA, but I’m pretty sure that when more than half the people you do business with would love more than anything else to never touch another of your products, you’ve got a big problem.

(No, I don’t have a copy of the survey. I wish I did.)

The reason Microsoft didn’t seem to be in trouble in the 90s was that Apple was failing to come up with an updated OS. The classic MacOS had fundamental architectural issues that meant it was unreliable — one bad piece of software could crash your entire system. Most business software was half-heartedly ported from Windows, which meant that running Adobe products or Microsoft Office was a bit like playing Russian roulette with anything else that was running at the time. So the Mac didn’t get used in business.

Since people had Windows at work, they tended to buy Windows for their home computers too. Partly because they figured they’d only have to be familiar with one type of computer that way, and partly so they could pirate all the software they wanted. Meanwhile Linux at the time was a disaster, and the web was in its infancy. So Microsoft basically had no competition.

Once Apple had a solid Unix-based OS, Mac sales started to pick up again. Then as more and more things started to be doable online, most people realized they didn’t actually want a general purpose personal computer with an OS they would have to babysit. Schools certainly didn’t. So now the top-selling laptop OS is Linux, in the form of ChromeOS. It outsells Windows and Mac. More people go online via their phones than via personal computers; I have friends who don’t own a computer, the phone is all they need or want for e-mail and light web browsing.

Sure, Microsoft fell flat in every area after 2000. But they fell flat in every area before 2000, except two businesses: office software and operating systems. That was their guilty secret — no other Microsoft division turned a profit. Windows CE was a failure, their online services were money pits, attempts to expand beyond the x86 platform failed, it was only the massive profits from Office licenses and OS bundling that kept the company profitable.

The Windows OS was completely rewritten by a team led by DEC engineers who had worked on VMS; the result was NT, which gradually took over from the creaky Windows 9x series of OSs. But the organizational structures weren’t changed, and the new OS quickly started to accrue the kind of technical debt that had crippled the old one. The choice of C++ as primary development language had looked good in 1989, but by the time Ballmer took over he was faced with an OS that was getting to be a nightmare to develop for. Hence Longhorn and .NET were needed because of decisions on Gates’ watch.

Longhorn failed because of technical debt. .NET would have provided a way out of that, but Microsoft were unable to drag developers to a totally new set of APIs. Apple had done it with Cocoa, forcibly ending support for the older Carbon APIs, but Microsoft has always seen back-compatibility as something to maintain at all costs — so pre-.NET Windows has remained, complete with DLL hell and all the other ugliness.

One thing the article is right about, though, is that Microsoft crippled themselves by insisting on everything being bolted to Windows. Some Windows CE mobile phones were quite nice, but if you didn’t run Microsoft Windows and Microsoft Office they were basically dumbphones with big screens. And nobody wanted to be even more strongly wedded to Microsoft. (Ironically it’s now Apple that are making the same mistake. The Apple Watch is useless unless you have an iPhone. Notice how Apple aren’t releasing Apple Watch sales figures?)

Another thing the article gets is that Microsoft has always been known for terrible user experience. When their competition was IBM, that wasn’t a problem. Against Apple and Google, however, it’s more of an issue.

Take Bing. It started out as a Google clone, but look at it now — chum boxes along the bottom, a “personalize your homepage” message harking back to the portal mania of the 1990s, an ad for a rewards program, a random quiz I didn’t ask for, and a ton of menus. Contrast it with the spartan Google home page and is it any wonder they have to pay people to use it?