That's not really related to the organizational dynamics that come into play for planning products and features. That tends to be complicated, involving a lot of people, meetings, and debate. (I'm not defending it, just saying that's how it is.) That 14 steps isn't about a code change; it's about the end-to-end design, implementation, and shipping of a feature, expressed in the most detailed, far-reaching way possible, to make a point about how things operate in a huge organization.
Both points of view are accurate, IMHO. When you're strictly coding, things are pretty efficient. Not too surprisingly, those were the times I enjoyed the most. But it's equally true that at MS, coding is only a small piece of a larger, somewhat slow-moving process. And sure, all that other overhead can get tedious, even exasperating at times. It's just a separate subject from what the OP is writing about here. The coding part is typically quite fun and autonomous.
However, his intro blurb specifically says that we are wrong in assuming that writing software is a "soul-crushing bureaucratic exercise." Which, as you point out, the post never really says anything more about.
Consider that Windows ships in over a hundred languages to hundreds of millions of people. I'm not sure how many languages Linux ships in (I'd bet quite a few), but Linux contributors don't have to worry about getting arrested for drawing map lines wrong: http://blogs.msdn.com/b/alexbarn/archive/2004/08/20/217602.a...
Do we ship software without testing? No. Do we ship software that isn't globalized? No. Do we just hack something together, check it in, and call it a day? No. These aren't the kinds of practices that lead to a stable platform.
Mistakes get made (at all levels of the management chain), but they tend to be the exception in my experience.
For instance, the build system is extremely "complicated". But all you really need to do is to make sure your code compiles, you pass all your local tests, and the BVTs, and you are good to go (oh, and a code review; anyone who objects to a code review is probably shouldn't be coding).
When, you, as a developer, satisfy those local conditions, the system automatically works at the global level.
Don't get me wrong, complexity is an evil that I'd rather avoid, but it is a necessary evil for creating systems that are useful in the real world; and it is an evil that can be tamed with automation.
[Full disclosure: as a former MSFT employee, my thoughts are heavily influenced by the company.]
Effectively preventing it from ever being simplified. Like you said, complexity is evil and has to be addressed because no amount of automation will correct it.
See the 14 steps which turn a 5 minute code change into a multi-man-month-mammoth: http://blogs.msdn.com/b/ericlippert/archive/2003/10/28/53298...
Or the now infamous Windows Shutdown "crapfest": http://moishelettvin.blogspot.com/2006/11/windows-shutdown-c...
There's more. And I have no problem with the author stating how great coding is (because he has an office and a nice computer!). However, while your right hand is telling us how great it is and we shouldn't assume otherwise, your left hand is telling us how horrible it is.