Not really. If you were following the company in the 90s, two quick examples are Windows 95 and Internet Explorer 4 (Remember when IE was actually _praised_ for better standards-compliance and developer-features?)
I think Windows NT is an even better example of a case where Microsoft was genuinely leading the way. NT was a workstation-class operating system for the average PC -- or to put it another way, it was "desktop VMS" eight years before Apple shipped the first real "desktop Unix".
Without the 98->XP growth path afforded by NT, Windows would have stalled ten years ago.
It might appear that way if Microsoft was the only company one was following back in the 90s. While they were freezing the market with promises of what Chicago would be, NeXT had already brought the future - and that codebase still feels like the future today.
Except that a NeXTCube cost $10k at the same time that a decent windows PC cost $1.5-2k. Even though both NeXTStep and NT represented full, proper 32-bit "modern" operating systems and were both invented in the early 90s it took until 2001/2002 for both to come of age as consumer grade operating systems (NeXTStep as OS X, NT as Windows XP). In the meantime MS put forth an interim solution (Windows 9x) that was still suitable for consumer-grade PC hardware, resulting in a continuation of Windows market dominance and billions of dollars in profits for Microsoft. In contrast, NeXT took the intellectual high road, stubbornly refusing to compromise its vision of the future, resulting in massive financial losses, being forced to lay off 2/3 of its employees, and a long, slow slide into bankruptcy and irrelevancy before being acquired by Apple (which had stretched the decrepit Mac OS to its breaking point).
Had NeXT maintained a little more pragmatism and a little less stubborn pride it's possible that they could have become the next Apple or the next Microsoft, rather than merely a juicy intellectual property morsel for the real players in the industry to pick up and take advantage of.
Sure, I couldn't afford a NeXTCube in 1990. But by 1993 I was running NeXTstep on my generic intel hardware, a full two years prior to the release of Windows 95. (Incidentally, I also had the pleasure of running it on a SPARC laptop made by Tadpole, and on one of the HP PA-RISC "gecko" machines, but those were owned by employers. I could afford to run it on a PC, though.) But, you're right, people continued to believe that NeXTstep required a $10000 machine long after it did not.
Reading the article, I don't think the author meant what you think - "Microsoft never developed a true system for innovation. Some of my former colleagues argue that it actually developed a system to thwart innovation." He says they always sucked in a moderately polite fashion.
MSR is a very powerful research org, agreed. In, say, machine learning and computer vision, this has been so for over 15 years.
But the point, and it is a valid one, is that this research does not seem to be turned in to actual products.
What is the systemic problem? Why has Google, which historically has had nowhere near the research depth, been able to innovate demonstrably, but not MSR?
One hypothesis is that research innovation starts with a small idea. It's easier to deploy a small idea on the web than as part of shrink-wrap software. Google can field the small idea, and iterate it on the web.
But to be included in the kind of software MS ships, you have to do things like (a) convince a project like Office to take it on, or (b) pitch it as part of a "big concept future system". Avenue (a) is very hard. And avenue (b) tends to produce vaporware. Sound familiar?
Maybe the scientists there are less ambitious than scientists elsewhere. I have read some good papers from MSR - I am interested in concurrency and how to express it more clearly in code - but MSR doesn't particularly stand out. They have a couple good scientists there doing fine research, but thinking of MSR as an organization that somewhat leads the way is somewhat of an exaggeration.
Sun Microsystems was spending more than 10% of its revenues on R&D... that's not a recipe for sure success, I would say, if you're not able to produce something sellable from time to time. Ah, SUNW used also to have a "CTOs council"...
Many friends of mine have tried to explain me what's so great about LINQ, but it never seemed that much impressive. The same with F#. Transactional NTFS flew completely under my radar, but it, too, doesn't look impressive the way, say, ZFS does.
I may sound harsh, but there is so much more going on that their contributions really feel underwhelming.
From what I can see, the transactions in ZFS and Transactional NTFS are completely different kind of things. The Transactional NTFS kind are distributed transactions of the kind you normally associate with database server and messaging systems like MQ - so your file IO can enlist in a transaction that touches a lot of different things.
Not saying it is a good idea - abuse of distributed transactions is a terrible thing, but I do think it is fairly innovative.
Being a bad idea kind of reduces the value of whatever innovation led to it. When you find something obvious nobody tried before, you should get suspicious.
Don't get me wrong. I am an engineer and I frequently answer a "why?" with "because it's cool" or "because it could be done", but with years of experience come a strong sense of what's a good idea and what's not and what you describe looks like a very bad one. And not even a lot of fun to implement.
Microsoft never did that. Author has to be a press release junkie to not know that.