Windows 7, the new version of Microsoft Windows and the successor to Windows Vista, is officially released in two days time. On his blog, my good friend Patrick Crozier has asked a possibly not very important question, specifically
I’ve heard of Windows 3.1. I am about to a lot about Windows 7. But I’ve never heard a peep about Windows 4, 5 or 6. Were they, by any chance, really good versions of Windows that we never got to hear about because the praise for them was drowned out by complaints about 95, 98, 2000, Millenium and Vista?
I think we should be told.
The simple answer is of course that 95, 98, 2000, Millenium and Vista were Windows 4, 5 and 6, although not necessarily in that order. Microsoft decided in 1995 to abandon product numbers on many of its products, and replace them with names consisting of the years in which the product was released. Since then, they have released products with names consisting of product numbers, years, two letter codes that might or might not means something, and words that might or might not have anything to do with what the product is supposed to do, with versions of the same product seemingly seldom ever using the same convention twice.
If you ask Microsoft’s PR department, they will tell you that Windows 95/98/ME were Windows 4, Windows 2000 was Windows 5.0, Windows XP was Windows 5.1, and Windows Vista was Windows 6.0, which it appears to make a certain amount of sense to follow with Windows 7.
However, it is of course more complex than that, and I am going to attempt to explain it. Reading the rest of this post is unlikely to improve your life in any way, although it will teach you something about the mindset of Microsoft and/or that of nerds in general. Madness may lie at the end of it. However, here we go anyway.
Microsoft has always used version numbers internally. You can find out the version number (“build number”) of any version of Windows by looking in the registry for a key named “CurrentVersion”, and the numbers given by Microsoft PR above are approximately those you will find if you check this for those versions of Windows.
Windows 1 and 2 were products that Microsoft sold before Windows 3, and they didn’t work very well and were not very successful. It was and perhaps still is a golden rule that Microsoft products didn’t work properly until the third major release, and that certainly was the case with Windows.
Windows 3.0 came in around 1990 and was very successful and was quickly updated to 3.1, which was the first widely used version of Windows. Windows 3.11 had more networking features, and Windows 3.2 existed, but was only available in Chinese.
For the next major release, Microsoft decided to name the product after the year rather than give it a version number, and thus we got Windows 95. This was version number 4.0. (Actually, 4.0.950, but everything after the second decimal point is really only of interest to the developers and I will ignore those from here on unless it is helpful not to). Two and a half years later Windows 95 was replaced with Windows 98, and this was Windows 4.10.1998, followed by Windows 98SE, which was Windows 4.10.2222. One must remember that in those days, the practice of software regularly upgrading itself over the internet had not arrived yet, so 98SE is similar to what would later have been called a service pack and not got a new build number at all.
The final version of DOS based Windows was Windows ME, which was Windows 4.90 and was released in 2000. At this point, Microsoft decided to sometimes but not always replace years with two letter names. Lord only knows why.
To understand what happened next, we need to go back a bit. In the early 1990s, Microsoft decided that its existing codebase, which was ultimately based on something a guy named Tim Paterson knocked up in a weekend in 1980, was not sophisticated enough for server applications, and decided that a more robust system was necessary. Microsoft hired some operating system experts from DEC, and they developed another operating system that was theoretically for servers, although Microsoft undoubtedly always had ambitions that its reach would be wider than that. This was Windows NT. The first version of this was ready for release in July 1993. Microsoft did not want their server OS to have a lower version number than their then current client OS, and so the first available version of Windows NT was given version number 3.1. It may also be that they wanted to try to avoid the “Microsoft products don’t work until version 3” problem, so for this product they dispensed with versions 1 and 2. NT version 3 was not widely used, but was followed by steadily improving versions 3.5 and 3.51, and by Windows NT 4.0 in 1996, which was designed to have a similar look and feel to Windows 98. This was used in quite a few professional environments, often for client machines as well as servers, and was a clearly better product than Windows 98.
At this point, Microsoft realised that the old DOS based Windows was not good enough for clients in the long run either, and decided that it would base its future client Operating Systems on Windows NT well. Therefore, the next version of Windows was based upon Windows NT. This was Windows 2000, which came in both a server and a client version (Windows 2000 Server, and Windows 2000 Professional), and is Windows NT 5.0. The 5.0 actually makes it the successor to Windows NT 4, not classic Windows 4. However, both versions of Windows very deliberately had the same version number at the time, so this point is often brushed over.
Despite having version number 5.0, Windows 2000 was actually the third version of NT, and became the first version of Windows NT to find itself in extremely wide use, perhaps confirming Microsoft’s “Third time gets it right” reputation, or perhaps not, given that the earlier versions of NT were much better products than the equivalent versions of classic Windows.
At the time Windows 2000 was released, it was considered ready for professional environments, but not for consumer environments, as there was not yet enough driver support in environments containing lots of diverse hardware, and because it required more hardware resources than many consumers were typically using. Therefore, there was not consumer version of Windows 2000, and Windows ME was released at the same time as the last version of DOS based Windows, although considering how awful it was, this may not have been a wise move. However, this gave Microsoft time to make their next NT based product more consumer friendly and for drivers to be developed for more hardware. This happened, and Windows XP came out in both professional and consumer versions in late 2001. This was Windows version 5.1. (All versions of WIndows after this are Windows NT based, but I will from here stop putting this before every version number to save myself the typing). Windows XP is actually extremely similar to Windows 2000 internally, although it has a different visual appearance.
Microsoft did not have a new version of the server version of Windows ready in 2001, and didn’t really need one, as driver support and a different visual appearance are much less of an issue with servers anyway. Microsoft did not get a new version of Windows server ready until 2003. Possibly because it was a little later than the consumer XP, this was given the name “Windows Server 2003” rather than a letter based name. In any event, this is considered by many to be the most stable version of Windows ever produced, and it has version number 5.2. Around this time, Microsoft developed 64 bit versions of Windows XP (based firstly on the ill fated Itanium processor, and then on the x86-64 architecture). These were based on Windows Server 2003, and both also have version number 5.2.
The next version of Windows, which was codenamed “Longhorn” internally at Microsoft and ultimately became Vista, had a troubled life which began early. Initially it was intended to be released around 2003, and Microsoft started with the standard XP codebase and developed from there. After a couple of years this project had evolved into an almighty mess, and Microsoft made the decision to abandon most of the code that had been written and use the Windows 2003 server codebase as a starting point from which to try again. After many more further problems with this project, Microsoft finally released Vista in 2007. This was Windows 6.0. Vista wasn’t really finished at the time and had horribly lacking driver support, and developed a bad reputation. A lot of people didn’t feel like upgrading hardware for the new OS, and for various reasons new categories of PC with relatively simple hardware became popular (netbooks, most notably). Thus there was great reluctance to upgrade to Vista, Microsoft is still selling XP in some categories, and Microsoft was determined not to make the same mistakes again.
Rather than attempting to put a lot of new features in the next Windows, Microsoft decided to fix the problems with Vista, wait for better driver support, and slim the code down so that it would run better on current hardware. They wanted drivers that had been developed for Vista to work on their new system, so as to not be hit by driver incompatibility problems again.
They largely did this. Windows 7 is largely “A slimmer Vista that Works”. In defence of Vista, after two service packs and some, Vista pretty much works at this point, too. However, due to the bad reputation of Vista, Microsoft wanted to sell the new product as something new that it would dissociate from Vista. So, they did the same thing that had been done to disassociate XP from Windows 2000, which is that they gave the default screen lots of different colours from the previous release. And they gave it a brand new name that sounded like a version number of a major release.
Except, of course, if you go look in the registry, there is a dark secret.
That’s right. If you ask Windows 7 what its version number is, it answers that it is version 6.1. “Windows 7” is not actually a version number. It is a version name.
There is something very Microsoft about this.
There is actually a good reason for this. When drivers are being installed, the registry is examined to check the version number of Windows to see if they are compatible with the operating system. Usually, they decide if they are compatible or not based on the major version number – ie the number before the point. If this is different from what they expect they may refuse to install. If the registry was set to 7.0 then many drivers that were previously designed for Vista would not install, even though Windows 7 might well work fine with them in all other ways. This suggests something slightly askew in the overall design of Windows driver support model, but we shall let this pass for now. Given, though, that Microsoft has gone to some trouble to minimise changes from Vista so that Vista drivers will continue to work on the new system, it probably is reasonable to conclude that from a technical perspective, 6.1 is a more reasonable version number than 7.0.
As I said, there is something very Microsoft about this.
Ah, version numbers.
I was once trying to get a DEC switch working and the support guy asked me for the version number of the backplane firmware.
“V7.2.3.2-10-97.17:50” I replied.
“Let me check my notes.” he said.
(OK, I don’t actually remember the exact number, but you get my drift.)
Please do some fact checking before posting stuff like this. There WAS a product called Windows NT4 (both server and workstation). THAT was Windows v4.
Er, I said that. About a third of the way down. The end of the paragraph that starts with “To understand what happened next….”. That was the only product that was officially released with “Windows version 4” as part of its name, but the whole Windows 95/98/ME family was also “Windows 4” in the sense that the build numbers all started with 4.x, and the software internally identified itself as Windows 4 point something. The point being that when Microsoft sold both DOS and NT based versions of Windows, they generally attempted to have the two products have roughly the same version number at the same time. This didn’t matter once we got to NT5/2000/XP, as after that the NT variants were all we had.
And yes, client versions of NT did exist right from version 3.1, so Microsoft had clearly considered the possibility of moving its client OS base to NT pretty much from the beginning. However, it was selling NT as a server solution at the time. (And let’s not even go into the lip service that was still some of the time being paid to OS/2). Microsoft was clearly preparing for such a move when it made the user interface of NT4 very similar to that of Windows 98, but the big push only came with Windows 2000.
I feel ashamed.
That was a really interesting read.
As I said, I feel ashamed
I take it English is not your first language. Or did you just not read the whole article?
Something slightly akin to Windows 7 pretending to only be 6.1 internally happened with the Opera Browser. The latest version is 10, and no other browser has been around long enough to reach this. After discovering that too many sites couldn’t have two-digit version numbers, the final release pretends to actually be Opera 9.80. Maybe they hope to stall until IE reaches version 10 or something and sites get fixed.
What have I done?
As a user of free and open-source operating systems, I am obliged to make the following crack. If I don’t, Richard Stallman’s ninja army will hunt me down and sing some of his songs. I apologise in advance.
So just like all the other ones, then?
Windows 1.01 was just a Interface UI for DOS
Windows 2 was a bit better and allowed Word and Excel to run, plus third party stuff too
Windows 3/3.1/WFWG 3.11and NT speaks for themselves
Windows 4 is commonly acknowledged as Windows 95
Windows 5 was Windows 2000
Windows 5.1 was Windows XP
Windows 6 was Windows Longhorn/Avalon was which became MS’ greatest ever calamity, i.e. Vista
Windows 7 is the hope that y’all forget about Vista 🙂
Hope this makes it clear.
I wear my geek hat proudly (I’m almost 60) and thoroughly enjoyed your article.
Now I understand why I needed Vista 64 bit drivers to make my printer operate with Windows 7.
Most enlightening!
Still, I’d always always rather buy something called “Windows 7” than bleedin “Snow Leopard”.
The real irony is that I now have to run my daughter’s Ipod on an old Windows XP laptop because my Mac, being nearly four years old, cannot load a version of Itunes recent enough to connect to it.
Still, I realise that there are few things more foolish on an Interweb forum than to dare criticise a Mac, so I’ll leave it there.
Many large organizations have the name confusions.
Many times documents will have a note as to the version of it. Then, they may do a redo of it, giving it another type of name altogether. This has gone back before computers.
Cosmetic changes, such as document names, are something people may do, that shows they’re doing something, of a nature that will not cause objections.
I once set up the microfilming of bank records going back to 1816. From there it progressed to microfiche,
then to digital. This was in the 1980’s.
Some documents would have 4 different names to index
This is often a problem in research of old records.
Most of the cosmetic name changes started and became more often, in the 1940’s and 50’s as corporate
and government organizations grew in size.
Schools taught that computers would reduce paperwork
In the institutions, printers became so fast, they needed
sound shrouds to protect hearing in the same room.
manuel: A four year old Mac should be running at least Tiger, which is compatible with iTunes 9, which should support any model of iPod. A five year old Mac might be running Panther, but this should be upgradeable (for a fee) at least to Leopard, which again should run iTunes 9 and support any iPod fine. I am therefore curious as to how you have hit that particular problem.
Where does Microsoft Bob fit into all of this? 😉
Michael – I have OSX 10.3.9. I have tried loading a version of Itunes capable of supporting the Ipod Touch and I can’t. You may be right – it could be more than 4 years old, but still, it’s unimpressive.
I become relaxed and happy just thinking about the day (quite soon) when I will reprogramme it with a large axe.
I’ve actually written code for NT 3.5. After being brought up on Unix and Mac systems programming, it was a very rude shock. It was appallingly badly designed (and as for WinSock, I still quake at the memory).
Ask the developer who invented it. Ask him why he can’t sit down in comfort.
That’s where MS Bob ended up fitting.
I want to find the guy who invented Clippy. Oh, yes, he and I need to speak. I got some sea bass with fricking laser beams set aside in my bathtub just waiting for him.
Please, everyone; get out more!
I will give up XP 64-bit when they pry my cold, shriveled fingers off of the keyboard.
Or when driver support dries up, whichever comes first.
Wake me when “Windows DeathKernel 9: A New Beginning” hits the big screen.
I really shouldn’t have read that. Can someone please pass the aspirin? My brain hurts.
You did warn us though.
@Spectre765: You found driver support for XP x64? Where?
One of the few things that Vista got right from Day 1 was better driver support in the 64 bit version. XP x64 has quite possibly the worst driver support of any version of Windows other than NT 3.1/3.51
From a professional viewpoint (and I’ve had to deploy and support damn near all those versions of windows, servers included), the only decent one of the lot is Windows 2000 Pro and Server. Server 2003 was ok, but Server 2008 is a bit of joke and don’t get me started on IIS7.0.
Vista is the biggest mistake MS made since Windows ME which 7.0 has rectified somewhat, but they really need to bin the whole codebase and start again. Unfortunately, Linux is going the same way – it’s gone from a lightweight run-from-floppy OS into a bloated and slow liability.
Of course, if you start from source and compile your own, Linux is lovely 🙂
“my Mac, being nearly four years old, cannot load a version of Itunes recent enough to connect to [an iPod]”
I call bullshit.
Even assuming you have an iPod that absolutely requires the latest iTunes 9, that will run on any G4, G5, or Intel machine that can run at least OS X 10.4.x.
The official requirement for 10.4.x is a G3 or better and built in firewire, which includes every iBook with 366+ Mhz (inroduced Sep 2000), every iMac with 400+ MHz (July 2000), and the Pismo Powerbook (Feb 2000).
Assuming that iTunes 9 really does need a G4 not a G3 (plausible as it may well have code that requires Altivec) the very last G3’s sold were the 900 MHz iBooks in October 2003, and the 900 MHz G3 iMacs in March 2003. Other than missing a vector unit (which MP3 & AAC decoding will benefit from for sure), they are much faster than many G4’s, which started at 400 MHz.
In summary: unless I missed something obscure, anything sold in the last six years can run the latest iTunes. Something “almost four years old” most certainly can.
I don’t know about XP, but NT means simple New Technology. XP probably is something as inane.
Eric,
Switch to Greek. XP is chi rho. Cairo. An internal Microsoft designation.
Just a minor nit: the “ill fated” Itanium processor is alive and well and powering big-iron servers all over the place. Which may not be what was originally hoped for it, but it’s still a large and profitable niche.
That’s another interesting and entirely separate discussion that it probably wouldn’t b a good idea to get into just now, namely: commodity Linux/Intel based hardware certainly is eating its way up from the low end of the server market, and already controls certain niches with special requirements in the high end – basically, huge websites – but the complete eclipse of proprietary big iron from IBM, HP and Sun/Oracle isn’t happening as fast as was generally expected a few years ago.
(See also: the last non-Intel “G5” Macs, which were basically a valiant but failed attempt to take an excellent high end server processor design and use it in desktops, for which its power and cooling requirements turned out to be too overwhelming.)
Great article, Michael. I too have used virtually every version of Windows (with the blessed exception of 1 and 2), and not always of my own chosing… thus is the corporate world defined.
IMHO, Win ME is undeniably the worst version of the lot. It tries to be flashy, but falls short in so many ways. 2000 Pro was good, but, like Vista, took several service packs to get there (about 4 if I recall correctly).
I’ve been using Vista 64bit for about a year, and have found it surprisingly very stable… must be the AMD dual CPU and Gigabyte motherboard 😉 But then I’ve never had much truck with Intel since that crappy Celeron CPU I got stuck with in my first home PC.
“my Mac, being nearly four years old, cannot load a version of Itunes recent enough to connect to [an iPod]”
I call bullshit. Posted by Bruce Hoult
Er, thanks Bruce.
Itunes capable of supporting the Ipod Touch doesn’t run on 10.3.9 (at least that’s what Apple’s own upgrade process tells me).
Your “bullshit” call appears to be saying that my machine is theoretically capable of running it with a different operating system. Right, thanks. “I call pointless”, as you might say.
I suppose this is one advantage of Microsoft taking so long between product releases after XP, and then the final product (Vista) being so poorly received that they had to keep XP on sale in some channels. Pretty much nobody who has bought a PC since 1991 needs to pay to upgrade their operating system in order to run recent software (or hardware for that matter).
Of course, Apple does basically operate an honesty system with respect to Operating System upgrades. There are no product keys, compulsory registration requirements, or any checking of any kind that you have paid for it. I would think they are particularly unlikely to care in the event that you were to upgrade from one obsolete system that they no longer sell (10.3, say) to another obsolete system that they no longer sell (10.4, say), which would be enough of an upgrade to run iTunes 9.
Not that I would ever endorse doing such a thing.
I haven’t touched Vista yet and stayed with XP. Now that 7 is out, is it better than XP, or should it first go through the service-packs routine?
Excellently informative article, more like this please!
I used Windows 2.0 (must have been?) in an HP XT or similar in, i think, 1988. It wasn’t bad really: considering what else was available apart from on mackintoshs and things like that, and it was “available” and not _too_ expensive. you could do spreadsheets on Excel without having to go things like ” /wgdd ” and the like.
Everybody likes to hate Microsoft: I think it’s because it did a Terence Conran – it “took all the [geeks’] precious things, and it gave them to bloody everyone”.
DOS and Windows brought much of humnaity, the very very big bit that just wants computers to mostly work most of the time, fairly well, out of the Darkness of unDOSness, and into the dawn of this age of little machines.
I liked WIN98 best I think, although I was younger then and didn’t want to do so much so fast, and I think XP is more stable. That’s what this is on. The Boy, however, favours Windows 7 and has been using it entirely without mishap, doing complex things, on a old Dell laptop, for some months, and is even able to upload heavy stuff to and from school.
Good exposition.
I didn’t know about Windows 3.2, thanks for that.
Pity you never mentioned the OS/2 fiasco, it’s an interesting part of the Microsoft story and reveals a lot about their psychology at the time, and possibly now.
I forgot to thank Michael for the post, which, strangely enough, I also found interesting (as well as some of the comments. BTW Ben, which machines run AMD dual CPU?)
Yeah, and for some strange reason we are always told that this works the other way around:-)
Andrew: There is one possible connection between OS/2 and the purported subject of the post – version numbers of Windows.
The team of developers Microsoft brought from DEC were supposedly to bring new technology to the development of OS/2 version 3.0. At one point this was known as NT OS/2 3.0. At some point, either due to the preferences of the engineers or the preferences of Microsoft management, they abandoned the OS/2 heritage, and developed a new OS separate from it. This became Windows NT, and OS/2 version 3.0 was developed by IBM. However, the OS/2 heritage is perhaps another reason why the first version of Windows NT was version 3, although this does not explain the .1.
The whole OS/2 saga is so big that it would have swallowed the whole post if I had attempted to bring it in, and in truth it is very hard to know the whole story. The outline is pretty clear. Microsoft and IBM agreed to work together to build a successor to DOS. At the same time Microsoft was developing this graphical “Windows” project to provide a GUI on top of DOS. Microsoft managed to get Windows to provide more features working in Windows than had been thought likely, the product was a huge success, and Microsoft abandoned the joint project with IBM, eventually developed its own next generation OS (NT) that it ultimately sold as the successor to Windows.
However, the point at which Microsoft decided on this, and how deliberately they decided on this rather than just going with market events as they happened really is hard to know. There was far too much spin and lying in which a company said it supported something while really doing something else to know what was actually going on, and when it happened.
You lot remind me of a bunch of Corvette enthusiasts, arguing about the number of strands in the horn wire of the early ’58’s.
IOW, wrangling about stuff that means absolutely nothing of consequence to 99.99987% of product users.
Guess what? Microsoft isn’t selling to people to whom this sort of things is either a) comprehensible or b) meaningful – just as Chevrolet is not selling cars to people who understand the part numbering system for the exhaust rocker arms. You’re all discussing all this stuff as though MS was supposed to plan product designations according to some laid-down law of the Geek Priesthood. But you folks are not their target market.
That’s why I always laugh when the propellor-heads around here open another joyful bout of Microsoft Bashing – it’s A Fun Game For All The Family! They’re right – MS products can be awfully bad, sometimes, but that doesn’t change the fact that the market loves them and buys them by the boatload. Technical superiority does not always map to market favour, in fact, it often doesn’t. Deal with it.
BTW, I still have in captivity a PC with Windows 3.1 installed and running (quite well) – to support a CAD software version which we need alive for a support contract commitment. For what it was, it really wasn’t bad at all.
llater,
llamas
I couldn’t agree more with Llamas.
I work on mainframes ( you know the REAL computers that were supposed to be replaced with PC’s 20 years ago because At&T was going to run payroll on a PC – yeah, right !).
Most Windows users, myself included couldn’t care less about which bits Microsoft didn’t twiddle the way some of you think they should have.
Windows does what I need it to do and I don’t waste my time trying to find out how it was
designed/put together.
If it is SO bad why don’t one or more of you whiz-kids design, build and market a BETTER system according to your superior grandiose approach.
Someone said earlier, and I agree, a LOT of you need to get out more.
That was the single worst explanation of Windows version numbers I have ever seen. Bar none. Please stick to what you’re good at.
Editor’s note: as you don’t say why, safe to say you can be ignored
Michael, I think these people are absolutely right, and you should refund their subscription money immediately.
BTW, to tell Michael, of all people, to get out more, one has to be really new to this place:-)
Hehe, he is so well travelled he is a one man global warming machine…Yes indeed, epic fail!
@ Adam:
I have a somewhat dated dual core AMD machine with ATI graphics, and Logitech webcam. I use it mostly for web browsing and have had few problems finding 64-bit drivers at the manufacturers’ websites. The hardware list:
AMD Athlon x2 5000+
Biostar A760G mobo
Sapphire Radeon HD4650 video
Kingston 2Gb memory
WD 500Gb hard drive
Logitech webcam
No hardware problems so far.
I hate it when a political blogger does a geeky post better than the geek bloggers. I’m a geek blogger. I wish I had explained the issue half as poignantly as this.
kelly H: Why? I mean, sure, Michael Jennings did a good job. But I could have done the same, and so could a few hundred thousand others. There’s no inherent reason why a political blogger can’t do a technology post just as well as (or better than) a technoblogger.
Vista is now a capable OS. The original release sucked hard, especially the networking stack. I could hardly get my broadband to cooperate with it. But after two SPs, it’s settled down and is now rock solid. Of course, the audio stack still sucks somewhat. But yes, I’m looking forwards to Win7.
My recollection of birth of NT is a bit different.
IBM and Microsoft were jointing working on a new OS called OS/2. It was a very difficult marriage, and eventually lead to a messy divorce. Both IBM and Microsoft gained joint custody of OS/2.
After the divorce, IBM took the OS/2 code, made the changes Microsoft would not let it make, and the result was called Warp.
After the divorce, Microsoft took the OS/2 code, made the changes that IBM would not let it make, and the results was called NT.
I’m delighted that I have observed this weblog. Finally anything not a junk, which we go through incredibly frequently. The website is lovingly serviced and kept up to date. So it need to be, thank you for sharing this with us.