Maybe from 30 years as a devloper and dealing with their sub-par stuff. Also have books written by MS employees detailing a corporate culture of dead line driven releases as opposed to releasing when ready. What is missing in any of these books is a bit about meeting customer needs and/or providing value to the end user. Maybe someone can refute that: one can't know everything.
To be fair from what I have seen it takes three to five years to get an operating system right. Have tested a version of Unix and 5 flavors of Linux in the last 4 months. Also have XP and Win7 machines to test with. Suffered through Dos 4-6, win 3.1, 3.11 where viruses started a big presence, win 95, mostly driver issues, 98 which wasn't so bad but then IE three came and immediately completely crashed the system into a no choice reformat and redo, (that initial release was later described as more virus than program), then NT good stuff, Millennium (PU), 2000 a very good effort, then XP which after some initial tweak is still in my opinion the best they ever made, Vista, a total P.0.S until the service packs and then win7 which did nothing new for all practical purposes. Most people don't encrypt their drives, use specific Internet filters, but do use email of which win7 had no native email client. Oh yeah, but even with a bounty on virus producers that MS initiated, virus protection still takes up a chunk of cpu and power.
After 30 years of dealing with MS, seeing first hand from hands on practical application, to include 16 years of field work supporting their products, having done web development since 1998, having read a minimum of 5 Microsoft books cover to cover and browsed maybe 15-20 more, having worked on more than 1,000 differnet machines, as in not 100 instances of the same machine in an in house environment, engineered from scratch as in spec the parts, order the parts, and assemble from pieces over 200 machines, used SQL server and others so I have some basis for comparison, engineered dozens of networks and implemented them to include doing all the cabling, and done who knows how many web apps, I have to think that I'm in a better position than most to evaluate a product at least from a practical application side.
My opinion would be that unless you have a good reason, let the suckers test the new stuff. That is in keeping with the principals of logic. Bye the bye, your logic is flawed. Classic informal logic error argumentun populum. If the number of people who believed something was true or the number who did the same thing made something true then the world would still be flat.
I don't test do testing for multi-billion dollar corporations or the government although I have taken parts of their products improved them and earned a fee from pleased customers. I don't see them testing my applications. Maybe for a small house or friend. I take pride in my work. Don't think MS or the Government does. I know from some of the books I've read, some Microsoft employees are embarrassed by their company. I would be.
My current vote for best operating system and most improved over the last 4 years is Ubuntu. Free C++ compiler, graphics, vector graphics, video editing, IDE, on and on. You do have to learn the OS to feel good about it though.
The biggest problem with MS is the black box of it, okay maybe after the virus thing. If you learn Unix or Linux properly you learn how a computer works. If you learn MS all you learn is how MS works. And then of course MS is not very stable as defined by: not crashing (MS doesn't crash any more, well maybe viruses but that is not relevant for this comparison as it is only their fault by negligence, you could argue willful negligence, not commission) or changing to where you spend hours relearning the system. MS fails on that part of the definition of stability. However again to be fair, things need to change as hardware improves. Still waiting for programs that actually use all the cores at the same time when they run instead of load balancing though.
Keep Up with TechRepublic