[ps2000]

If I didn't overlook something it only states something about hardware market shares. Which are much easier to measure than software market share. (Although even in the hw market shares there might be a significant uncertainty due to DYI built computers which were quite common during the 90s.)

[streptomycin]

By the mid to late 90s, I don't think DIY computers would have made much of a dent compared to prebuilt computers, especially since most of them would have ended up running Windows.

I wasn't able to find web usage stats that are very old. http://www.geek.com/wp-content/uploads/2009/05/os_marketshar... is the oldest usage-based data I could find, which puts Windows up at around 97% in 2005. I can't imagine it was much lower in the late 90s, given that Macs of that era were pretty shitty and even less people used Linux back then.

[ps2000]

During this time I still went to school and half of my friends had self-build computers. The other half had computer from very small shops.

And this was also a time when none-Linux Unixes were still still popular, so I would expect some share from them. Linux got a lot of momentum around '98. (Yes, and Macs existed too.) Moreover there was OS/2, BeOS, Irix and whatever... Linux however lost a lot of momentum in the early 2000s when people realized it sucks as a Desktop OS for non-geeks.

90% is well possible I think, but 98%? Never ever.

[streptomycin]

Well, I linked to some data showing it up around 97%-98%. If you think OS/2, BeOS, Irix, or something else was making a significant dent in the late 90s OS market, then... well, frankly I don't know what to say to something so ridiculous.