Random version numbers, a UI full of mismatched colors and a phone OS that has gone so far from its roots its just confusing. I mean sure they seem to be getting more dev friendly, but my god their end products are getting so far off track.
The first version of Windows NT was numbered 3.1, and since then we've had: 3.5, 4.0, 2000, XP, Vista, 7, 8, 10
So "7" and "8" were anomalies in that they actually matched the indices in this sequence.
Windows 2000 kind of makes sense too, since we had Windows 95 and 98. Or at least 2000 would have made sense if ME didn't exist.
I always thought XP was an odd name though, but 10 is definitely the worst thus far because, rightly or wrongly, people have certain expectations with version numbering and Windows 10 break that (just look at how damning people were towards KDE because they had high exceptions from version 4.0 despite developers repeatedly warning users that it was still a beta and that the version number wasn't a statement about KDE's stability)
NT 3.1 was the rename of OS/2 NT after MS and IBM couldn't agree on what to do with OS/2. But it was a parallel release to the Windows 3.1.
> Windows 2000 kind of makes sense too, since we had Windows 95 and 98. Or at least 2000 would have made sense if ME didn't exist.
Me was the DOS-based successor to the Win 3.1 -> Win 98 SE line, 2000 was the NT-based successor to the Win NT 3.1 -> Win NT 4.0 line.
Since 2000 wasn't a successor to 95/98, it can't really make sense because of them -- to the extent that the name indicates its the next in that line, that illustrates why it doesn't make sense rather than indicates that it does (though, IIRC, that was part of the intent -- 2000 did support more of the consumer features than NT 4.0 did and there was some initial indication that MS hoped that 2000 would be the convergence OS that XP became, and that Me was a result of MS realizing belated that that wouldn't work and that they needed a new consumer OS as a stopgap before the convergence OS landed.)
And FYI, ME (nor 9x) weren't DOS-based. They used DOS bootloaders, but their drivers bypassed DOS entirely. I'm not sure whether 3.x did the same (architecturally I think it did, but OS/2 could also run Win 3.x inside a DOS emulator)
DOS to Windows 9x was like the read only kernel environment, initramfs, in Linux. Except that DOS is obviously a fully functioning interactive environment in itself (actually, initramfs might be too. I think you can get a Bourne Shell prompt if it goes tits-up)
While it's true that Windows 1 and 2 were literally just front ends to DOS (Windows applications would often call DOS functions) Windows 3.x was an OS in it's own right (albeit still heavily dependant on DOS) and Windows 95+ basically just kept DOS about as a bootloader and for DOS compatibility (games etc)
So it wasn't just "some Windows facilities" in Win95 that bypassed DOS; it was every single Windows application (it had it's own binary executable format, it's own kernel, it's own ABIs and APIs, drivers, memory management, etc.
But it's fun to bash Windows 9x as being "DOS-based" because it's derogatory and, let's be honest, there wasn't a whole lot right with that whole lineage of Windows.
Microsoft Windows [Version 6.3.9600]
(c) 2013 Microsoft Corporation. All rights reserved.
When it comes to Windows, Microsoft appears to have only two values: strict adherence to backwards compatibility and changing everything else as much possible.
Getting Windows to function across so many platforms can only cubicly raise the difficulty of maintaining compatibility. If that's not the yawning abyss of a long-term tech support hell, I don't know what is.