Microsoft has decided to reinvent the square wheel once more. The next version of Windows, currently code-named Chicago (apparently it was Jaguar before that), will not be called Windows 4.0 as one might expect, but will instead be called Windows95. Microsoft claims customers have trouble figuring out which recent version of Windows is the latest, though I can't figure out how anyone could get stuck because no matter how you parse the numbers, they proceed sequentially from 3.0 to 3.1 to 3.11. Evidently, the problem is more that, given a single version number, many users have no idea if it's the latest one, which is at least conceivable. Windows95 would in fact solve that problem, albeit incompletely and temporarily.
(As an aside, the main naming convention that causes problems uses a d for development, since development versions are usually followed by beta versions, which use a b, so in fact NewsWatcher 2.0d17 came before NewsWatcher 2.0b13, the latest version.)
Lest we think Microsoft invented the concept of giving a software product name the name of the current year, think about the history of Adobe Illustrator. First Adobe released the aptly named Illustrator 1.0, then 1.1, and then, in an apoplectic stroke in 1988, Illustrator 88, which made sense at the time because it was pretty clear that Illustrator 88 was the latest version. But Illustrator 88 hung around for several years, and started looking seriously dated (and confusing people, when there was no Illustrator 89 or Illustrator 90), so Adobe came to its senses and next released Illustrator 3.0, skipping 2.0. They later skipped over 4.0 to 5.0 (probably because of the Windows versions of Illustrator). Illustrator 5.0 synchronized the Macintosh and Windows versions.
As yet another example, a friend reminded me of a magazine that started out life as Science 80, and every year changed its name to match the year, Science 81, Science 82, and so on. My friend knew someone who worked at the magazine, and this person said they found dealing with the name change every year nightmarish, since business cards and letterhead had to be redone, ISSN registration resubmitted, and so on.
So although Microsoft claims it will become easier for users to determine the latest version of Windows, the argument is flawed. Although they don't necessarily plan to release a new version of Windows every year, many people will want to know where Windows96 is once we get to that year (and it's unlikely Microsoft will release a major upgrade in 1996 if past history is any indication). Also, what happens if Microsoft needs to release two upgrades in one year? Should they increase the year number for a minor bug fix, and what if there are two bug fixes in a single year? Will we see Windows95a, or Windows95 1.1, or perhaps the ever-popular Windows-October95? And, as Tonya pointed out, such a naming convention makes no long term sense. What happens in a few short years when we hit the year 2000? Windows00 is going to look stupid, so they'd have to go for Windows2000, which should confuse customers who figure Windows2000 is older than Windows 3.0, given that two is less than three. Perhaps the Microsoft marketers have too much time to twiddle their thumbs and come up with wacky marketing ideas, given the delay in shipping the product.
The only useful piece of information that comes out of this official name change is that we can be certain Chicago won't ship until next year. Reports place the realistic ship date in the range of March to August of next year. So perhaps Apple can move past System 7.5 by then - who knows?
The new name has already prompted many tongue-in-cheek comments about how 95 stands for the percentage that will be complete at ship, the number floppies it will ship on, the number of megabytes of hard disk space required, or perhaps the number of minutes to install.
A more serious problem for Windows users is the fact that Windows apparently uses two version numbers internally, major and minor revision numbers, where both numbers are stored as decimals. Thus, Windows 3.1 was major revision 3, minor revision 1. The problem appears when some program checks to be sure the major revision is greater than 3 and the minor revision is greater than 1 before proceeding. Programs that check in this way will fail if the internal version numbers go to 4 and 0 (since the 0 is smaller than 1), as one might expect them to. If these programs actually formed the full decimal, it wouldn't be a problem, of course, but since some programs, perhaps many, don't do this, it becomes a real question. This entire issue predates the new Windows95 name, so now the question is what those internal version numbers will be in the shipping version of Windows95.