74

Microsoft Windows was originally introduced in 1985, ostensibly to compete with the Apple Macintosh, and other computers shipping with graphical shells by that time. However, early versions of Windows were not commercially popular, and MS-DOS remained the primary OS used on PC clones.

This changed rather abruptly in 1990, when Windows was now 5 years old, and Microsoft released the Windows v3.0 update. Windows 3 was the first version to enjoy broad commercial success, and it began to quickly supplant MS-DOS as the dominant OS on PC's.

While I'm certain that Windows 3.0 was a superior product to the earlier versions, it's not clear to me which new feature would have finally catapulted Windows into such new heights of success. Speculatively, I would be more inclined to see this sudden success as being either a culmination of earlier factors, or a confluence of fortunate events occurring around the same time.

What is the best explanation for Windows' sudden rise in the early 1990's with version 3.0, taking into account the relative obscurity of Windows for the five years previous?

  • 6
    I'd be willing to guess it was a combination of the 386 processor making it easier to protect & segment memory for apps, and improved graphics/resolutions leading to a reasonable UI compared to previous versions. – Joe Mar 2 '17 at 21:44
  • 1
    @Joe makes sense, but the 486 processor seems more coincident with 1990 that the 386. – Brian H Mar 2 '17 at 21:49
  • 11
    @BrianH 486s were still very rare in 1990. – Stephen Kitt Mar 2 '17 at 22:30
  • 29
    One word: Minesweeper – Buhb Mar 3 '17 at 15:01
  • 11
    @cbmeeks The computer marketplace in 1990 was a radically different place than today. To put it in perspective, Windows 3.0 sold 4 million copies in 1990 and was considered a huge success. In contrast, the iPhone sold about that many units per week in 2016, which was a slowdown from 2015. 1980's and 90's era computer buyers generally tended to be more avidly interested than today - more serious hobbyists and professional users that'd be more likely to know those kinds of details. (Or at least feel the pinch of the smaller address spaces.) – mschaef Mar 3 '17 at 18:57
54

Stephen Kitt covers the bases well, but I think the majority of the reason relates to fact that Windows 3.0 finally brought 286 protected mode execution to the masses. Even though the 80286 was first released in 1984, Windows 3.0 was the first mainstream platform that actually ran it in protected mode. That made it the first mainstream platform that could present more than 1088K (not 1024K) of address space to a single user process. This may sound like a minor technical issue, but it represents Microsoft effectively addressing a demand from the marketplace that had been only poorly addressed for 6 years prior.)

To give a bit more detail, the 20-bit/1024K 'real mode' address space of the original PC was widely considered to be major limitation, even early on in the PC's life. The original design reserved the top 384K of address space for I/O devices, ROM, etc., leaving 640K of address space for RAM. One of the important software applications of the time was the spreadsheet Lotus 1-2-3, and large spreadsheets could easily hit this limit. This was particularly true when running what were known as TSR's. (There are subtleties, but In modern terms, you can think of TSR's as lightweight background processes.)

In parallel with this, 1984's Macintosh release and the 1985 release of Windows 1.0 started introducing GUI's to the mainstream PC market. These added memory pressure, both in terms of the graphics processing and in terms of the fact that they made it possible to run multiple applications at the same time. (In truth, this only came later to the Mac with Switcher and then MultiFinder.)

So, in the first half of the 1980's into the second half, there came to be a large demand for techniques for addressing more than 640K of RAM, and there were a few common ways to handle the problem. Some machines would let you allocate part of the top 384K over to RAM, allowing for 704K and 768K configurations, but by far the more popular approach was something called LIM EMS.

LIM EMS was essentially a standard bank switching standard developed by Lotus, Intel, and Microsoft (LIM). Lotus wanted to make it possible to work with big spreadsheets and Microsoft wanted to make it possible to run more applications in real mode 20-bit Windows. What LIM would do is take 64K within the top 384K and divide it into four 16K pages. It then presented a separate multi-megabyte address space that could be mapped into those four pages. While the CPU's native address space was still 20-bit and 1024K, this provided a way to use memory operations to access more than what would normally be allowed in a 20-bit address space, and do so in a semi-standard way. Initially, LIM EMS was always implemented in the form of a hardware plugin card that contained the mapping hardware and EMS memory.

The downside to LIM EMS was that 1) programs had to be specially written to use the memory and 2) it was a bit of a pain to use. Microsoft Windows (as far back as 1.0) addressed these issues at least in part through the use of a handle based memory manager. Unlike traditional malloc/free, which returned a pointer, the Windows memory manager returned block handles. Developers only got the pointer to the block when the indicated to the OS that they actually wanted to use it. This let Windows do things like automatically move blocks around, including to EMS, when they were not in use. (Keep this in mind, it becomes important later on.)

Of course, the real solution to the address space issue was to select a CPU that had a bigger address space. This happened in 1984 when IBM picked the 80286 for the PC AT. The 80286 introduced a 24-bit address space, which was more than even a real-mode EMS 3.2 machine could manage... and it was native to the chip.

The catch was that to get the 24-bit address space, you had to adopt 80286 protected mode, rather than real mode. Unfortunately, DOS didn't support protected mode, and protected mode changed the rules enough that it wasn't necessarily easy to make the switch. Existing applications couldn't run in protected mode, and once a 286 was in protected mode, it was difficult to get it back to real mode. Still, around this time, there were continual rumors of an Advanced DOS (aka 286-DOS or DOS 4.0) that would somehow enable protected mode operation. What this wound up being was the IBM/Microsoft joint effort - OS/2.

OS/2 brought protected mode to the PC, but due to limitations of the 286, it was itself limited. While it could multitask OS/2 applications, running a DOS application suspended all the OS/2 apps, and was subject to all the usual DOS limitations. It was also expensive, only partially complete when it shipped, and had high memory requirements. It wound up not doing very well, while the industry stayed with DOS, etc.

At this point, it's worth pointing out that there are also 80386 machines. The 80386 is expensive, but it has a couple things the 286 doesn't. The most notable (at least for the late 1980's) is what is known as V86 mode. The 80386 has hardware support for running multiple-real mode processes. This is used by Windows/386 and DesqView/386 to run multiple real mode processes at the same time. The 80386 is also powerful enough to emulate LIM EMS with its internal MMU. The 80386 is truly revolutionary. (Note, though, that even under Windows/386, the Windows environment itself runs in real mode with a 20-bit address space...)

There's also Windows/286 around this time. One of the implementation details of the 80286 is that it would actually generate physical addresses as high as 1088K-16bytes. Windows/286 allowed access to this, so-called, high memory area beyond 1MB.

This answer is turning into an epic, but this is the state of affairs when Windows 3.0 is about to be released. 80286 machines have been out for 6 years and 80386 machines for 4. While there are hacks and third party solutions to use the capabilities of these machines, there's nothing that works all that well, and nothing that easily ships out of the box. It's really rather a mess.

This is when a Microsoft Windows developer named David Weise figured out how to run normal, mostly unmodified Windows applications in protected mode:

https://blogs.msdn.microsoft.com/larryosterman/2005/02/02/farewell-to-one-of-the-great-ones/

This is a huge win for Windows, because there finally was a cheap path for people to take their existing hardware and software, and mostly cheaply get to use the capabilities of their system. (The reason this was possible at all is that handle based memory manager I mentioned several paragraphs ago. If using protected mode means following a different set of rules about memory usage, it turned out Windows applications were already mostly following those rules.)

You also have to keep in mind that this represented a huge turning point for Microsoft. They immediately knew it would undermine OS/2 (largely useful for protected mode access) and irreparably damage its relationship with IBM. They of course did it anyway, and made a huge success out of it.

As Stephen points out, there were a number of other significant and very positive changes to Windows, but I think the biggest is that it finally provided a decent answer to the memory question.

  • 1
    @BrianH Yeah... anything with a 68K was automatically in much better shape. Not only was there a wider address space, it wasn't segmented. MacOS had some self-inflicted issues going from 24 to 32 bit pointers, but nearly as bad as the growing pains from real mode x86. (Developers on pre-32 bit MacOS systems would sometimes store extra information in the top byte of a pointer.) – mschaef Mar 3 '17 at 20:20
  • 3
    These are the answers that make this site awesome. This is why I am addicted to this site. – maple_shaft Mar 3 '17 at 21:42
  • 2
    @mschaef Wonderful job telling this story! And of course thanks for the kind mention. You're right, the FixDS technique worked in real mode. That's what I originally wrote it for, long before Windows 3.0. In fact, it would have worked fine with Windows 1.0 apps! The entire EXPORTS/MakeProcInstance() misadventure really never did need to happen. Raymond Chen has an article on it here, and this page on calling conventions is interesting too. – Michael Geary May 21 '17 at 7:00
  • 1
    @MichaelGeary Thanks so much for the note (as well as all you've done over the years to make the profession better.) – mschaef May 22 '17 at 20:38
  • 3
    ...but used 32-bit segment registers, such a mode could allow programs written in Java or .NET to access up to 64GiB of address space while using 32-bit object references (which could be more efficient than using 64-bit references everywhere). – supercat Dec 1 '17 at 21:40
76

There were a number of factors involved.

  • Windows 3.0 introduced a more refined user interface than available in Windows 2.0: more colours, proportional fonts everywhere, smaller icons, and MDI windows (multiple document windows inside an application window)... This made it "obviously" better than plain DOS to many users. Reviewers back in the day perceived this as Windows "growing up", since its new interface was modelled after OS/2's Presentation Manager (and OS/2 was still what a grown-up operating system was supposed to be like back then).
  • Windows 3.0 included better 386 support than previous versions (including Windows/386). This didn't make a huge difference for Windows applications yet (compared to running them on a 286), but it meant that users could run multiple DOS applications at the same time, displaying them in windows on-screen. Most users still had DOS applications they needed, so running DOS applications was very important.
  • While it could make better use of 386s, it still ran well on the very common 286s, and was usable on 8086s (but you probably wouldn't want to really). In 1990, 486s were jaw-droppingly expensive (the realm of servers effectively), 386s were the high end and many people still bought 286s.
  • As pointed out by john_e in his answer (upvote it!), on both 286s and 386s, Windows 3.0 ran Windows applications in protected mode, allowing them to use all the installed memory (without any change to the applications themselves). On 386s, Windows 3.0 could additionally use a swapfile to provide more memory than was physically installed. mschaef's answer gives a lot more detail, as do Andrew Schulman’s two PC Magazine articles on the subject, Windows 3.0: All That Memory, All Those Modes and The Programming Challenge of Windows Protected Mode.
  • Windows 3.0 was much easier to use than previous versions, in particular thanks to its new File Manager and Program Manager (which allowed programs to be logically grouped). In previous versions you started applications from the Executive which was a rather primitive file manager.
  • Windows 3.0 included a few simple but usable "productivity" applications, which were actually good enough for many people: Write in version 3.0 was a passable word processor (fine for letters, homework etc.), and Cardfile was a neat little information organiser (both of these were available in previous versions of Windows). Solitaire made its appearance in version 3...
  • Microsoft managed to get extensive press coverage for the Windows 3.0 launch, far more than previous versions; a quick look through the Byte archives shows a cover story for the launch in June 1990, extensive coverage in the July 1990 issue with a number of in-depth articles, and continued coverage thereafter as more and more software was released for Windows 3.0.
  • Around 1990, Windows software was starting to become usable, with enough features to convince people to switch to Windows to be able to run specific pieces of software: Word for Windows 1.1, Excel 3.0, PageMaker...
  • Last in this list but not least, Windows 3.0 was the first version to be widely factory-installed on PCs. Before 1990, when you bought a PC you'd get a version of MS-DOS, and that would pretty much be it. After Windows 3.0's release, many manufacturers started bundling MS-DOS and Windows with their PCs.

Before Windows 3.0, people who used Windows did so because they had to, typically to run Excel or PageMaker: they'd start Windows, start the software they were really interested in, do whatever they needed to do, then exit the software and exit Windows. (The same was true of other desktop environments, e.g. GEM for running Ventura.) Micros were single-tasking and people just ran one program after another. With the features listed above, it suddenly made sense to run Windows and stay there, even to run DOS software (apart from games). On a 386 with its virtual memory support, Windows could even be a better DOS than DOS... With Windows 3.0, people started adding win at the end of their autoexec.bat, and PCs really became Windows PCs, no longer just DOS PCs.

In following years, memory prices fell, and PC prices fell too; by 1992 a 386DX or 486SX with 4 MiB or even 8 MiB of memory was affordable (at least in North America and Western Europe), and Windows 3.0 was very comfortable to use on such a system. Windows 3.1 improved things further with its Truetype engine and built-in multimedia support in particular, OLE brought composite documents and made multi-tasking indispensable, and Windows for Workgroups brought networking to the (office) masses (easy networking was a killer feature in the early nineties). Eventually games started appearing too (in particular Myst on Windows 3), and the rest is history.

Here's a quote from the June 1990 issue of Byte, to give you an idea of Windows 3.0's reception:

Users of DOS PCs rejoice! Windows 3.0 will breathe new life into your machine. Microsoft has bundled a graphical environment, a suite of desktop applications, a DOS program switcher, a Windows multitasking executive, a V86 multitasker, and a virtual memory manager into a single package.

Windows has long aspired to change the face of DOS computing — not just for Excel or PageMaker addicts, but for the rest of us. A year ago, that transformation seemed unlikely. Today it appears inevitable. Windows 3.0 finally consummates the Windows/DOS marriage. That is good news for the 386 crowd, and it's great news for the silent majority: 286 owners who've lately been made to feel like they have bought an Edsel.

  • 6
    IMO the biggest factor was your last ("but not least"). Having a new machine boot up into Win 3.0 on its first power-up was a game change. If OS/2 had managed that there would be few who remembered Microsoft Windows as anything more than an also-ran, like Microsoft Bob. (Remember, when Win 3 came out, Microsoft was still telling people that OS/2 was the future.) – RichF Mar 2 '17 at 22:36
  • 2
    I agree (and see my first point for OS/2). I put that point last though because the others explain why Windows was finally an attractive proposition for manufacturers to bundle... – Stephen Kitt Mar 2 '17 at 22:40
  • Good answer, but one nitpick: Civ2 is 1996. Myst is early 90s and part of the "multimedia" CD-ROM boom, so I agree there. – Coxy Mar 3 '17 at 4:06
  • 6
    +1 for Solitaire -- I would not underestimate this as a factor! ;-) – jkf Mar 3 '17 at 5:41
  • 1
    @MichaelKjörling I'm not sure about "many"; Excel 2.0 included a run-time version of Windows 2.0 which it would use if you didn't have the full Windows 2.0. I'm not aware of any other application which did this, although I dare say there were some... I do know people who configured Windows 3.0 (and 3.1) so that it ran a single application (instead of Program Manager), and exited when that application exited, achieving the same effect. – Stephen Kitt Mar 5 '17 at 8:38
22

One reason that Windows 3.0 was popular with software authors: it included a DOS extender, meaning that on 286 / 386 processors Windows programs could run in protected mode and access as much memory as the computer had, rather than the 640k allowed by PCDOS.

  • This was already true of Windows 2.1 (which could even use EMS on 8086 computers). – Stephen Kitt Mar 3 '17 at 10:33
  • 1
    As I understand it, the 286/386 versions of Windows 2.x could run in protected mode, but programs were still limited to 640k. That's how it's described at virtuallyfun.superglobalmegacorp.com/2011/06/01/windows-3-0 for example. – john_e Mar 3 '17 at 11:05
  • Ah yes, you're right, the kernel ran in protected mode but applications still ran in real mode. – Stephen Kitt Mar 3 '17 at 11:12
  • 2
    Windows/286 was real mode with the A20 gate disabled. This gave access to a paragraph less than 1088K, through the use of the HMA. Windows/386 had a protected mode VMM sitting underneath a real mode Windows kernel. You can think of it was DesqView/386, but with Windows as the GUI.. the main benefit was background processing of DOS apps. – mschaef Mar 3 '17 at 15:33
9

For me, the main advantage of Windows 2.0 over MS-DOS was that I could configure my printer just once and then all Windows programs used the same configuration. Also it printed graphics better than most DOS applications. But overall, Windows 2.0 was basically a technology preview of things to come...

Windows 3.0 let you run MS-DOS sessions in their own window, making it more user-friendly for multitasking than Desqview. Also it supported 256-color VGA.

Windows 3.1 was where Windows really took off. It added support for sound cards and had much nicer fonts.

  • 4
    It's hard to overstate how much the printer support mattered... particularly if you had an oddball printer (like my family did in its Toshiba P321). – mschaef Mar 3 '17 at 15:31
  • 1
    It was the combination of the printer driver and proportional fonts that convinced me Windows was the future. – Mark Ransom Mar 3 '17 at 21:39
  • 1
    My recollection is that Windows 1.0 (yes, I actually used it) was basically a really bad graphical shell for DOS. Windows 2.0 was marginally useful, and solved the memory and printing issues. Windows 3.0 was good enough to use, and Windows 3.1 was good enough that you didn't want to use anything else. I switched to OS/2 for a short period of time because it seemed like it would be a better platform - then switched back to Win 3.1 when it turned out it wasn't. – Bob Jarvis Mar 5 '17 at 4:10
  • OS/2 had the problem that by the time 2.0 rolled around, they were advertising it as a better Windows than Windows. For a while, they could actually achieve some of this promise, because the IBM/Microsoft joint development agreement gave them access to the Windows source. This let IBM bundle OS/2 with a custom version of Windows optimized to run under OS/2. Of course, the agreement ran out, and all IBM had really achieved was to undermine any motivation developers might have had to target OS/2. I think all those OS/2 versions accomplished was to advertise for Windows. – mschaef Mar 5 '17 at 22:56
  • (I should point out that OS/2 did have a long life in certain niche applications... ATM's and dev workstations in some IBM mainframe development shops come to mind.) – mschaef Mar 5 '17 at 22:57
7

TL;DR: The success of Microsoft Office, proprietary data formats and the ensuing marketing strategies thereof, undercutting the competition, collusion cooperation with OEMs to lease MS software preinstalled on their products, and probably most importantly: landing deals with IBM and their longstanding contracts with them while retaining the ownership of certain IPs giving them market control over products that they sold to IBM and other companies ("Merchandising!"), and to a lessor extent: ease of use via the company who first brought us GUIs.


Windows is what most early PC computers came with, is what you needed installed to run other Microsoft products, and historically is what other users needed to also be running to facilitate the useful exchange of documents which were written with their software.

"This program requires Microsoft Windows to run."

There was nothing sudden about Microsoft's "aggressive marketing" and the predatory strategies (United States v. Microsoft Corp.) that they use(d) leading to their domination of the market. Windows 3.0 and Microsoft's success as a whole is indeed a culmination of earlier factors. This answer focuses on why everyone had it, as opposed to what made it distinctly better than other operating systems.

The confluence was the fortunate event of the explosion of the PC market, where OEMs began to sell computers with preinstalled operating systems, which due to Microsoft's marketing strategies was largely corned by versions of DOS and Windows. The company's perpetual existence and continued success, and the budding PC market in general, owes its greatest dues almost entirely to contracts that they and other companies had with IBM.

IMO, the was no "relative obscurity" of Windows Microsoft software in the years previous - they held an iron fist over the entire PC industry until only very recently.

One year prior to the release of W3.0, they released Microsoft Office, which today is used by approximately one seventh of the world's population, and at the time of its initial release required a Windows operating system to run. [citation needed] And at the very least, until 2007, required recipients of Office data to be running Office to be read.

Microsoft Office prior to Office 2007 used proprietary file formats based on the OLE Compound File Binary Format. This forced users who share data to adopt the same software platform.


(What I cannot find is the number of W3.0 boxed copies sold to end users versus all the licenses of it granted to OEMs, though I suspect the latter to dwarf the former - I've never even seen a boxed copy of W3.0, and it's the only version of Windows that I've never installed save for Vista and W10)


DOS (Disk Operating System) was the operating system that brought the company its first real success. On August 12, 1981, after negotiations with Digital Research failed, IBM awarded a contract to Microsoft to provide a version of the CP/M operating system for use in its new IBM Personal Computer (PC). Microsoft purchased a CP/M clone OS called 86-DOS (originally known as QDOS for "Quick and Dirty Operating System") from Seattle Computer Products, which IBM renamed to PC-DOS. Around 1983, Microsoft collaborated with several companies to create a home computer system, MSX, which contained its own version of the DOS operating system, entitled MSX-DOS; this became relatively popular in Japan, Europe and South America. After Columbia Data Products successfully cloned the IBM BIOS, quickly followed by Eagle Computer and Compaq, PCs manufactured by other companies flooded the market.

Its arrangement with IBM allowed Microsoft to have control of its own QDOS derivative, MS-DOS, and through aggressive marketing of the operating system to other manufacturers of PCs, Microsoft became one of the major software vendors in the home computer industry. Microsoft continued to expand its product line in other markets with the release of the Microsoft Mouse on May 2, 1983. Microsoft Press, a book publishing division, debuted on July 11 the same year with two titles: Exploring the IBM PCjr Home Computer, by Peter Norton; and The Apple Macintosh Book, by Cary Lu.

In August 1985, Microsoft and IBM partnered in the development of a different operating system called OS/2. On November 20, 1985, Microsoft released its first retail version of Microsoft Windows, originally a graphical layer on top of its MS-DOS operating system. In 1987, Microsoft released its first version of OS/2 to original equipment manufacturers (OEMs).

In 1989, Microsoft introduced its flagship office software suite, Microsoft Office, a bundle of separate office productivity applications, such as Microsoft Word and Microsoft Excel. On May 22, 1990 Microsoft launched Windows 3.0, a new version of its operating system boasting features such as streamlined user interface graphics and improved protected mode capability for the Intel 386 processor; it sold over 100,000 copies in two weeks. Windows generated more revenue for Microsoft than OS/2, and the company decided to move more resources from OS/2 to Windows. In the ensuing years, the popularity of OS/2 declined, and Windows quickly became the favored PC platform.

During the transition from MS-DOS to Windows, the success of Microsoft Office allowed the company to outpace its competitors in applications software, such as WordPerfect and Lotus 1-2-3. Eventually, Microsoft Office became the dominant business suite, with a market share far exceeding that of its competitors.

Microsoft, newworldencyclopedia.org (headers and citations removed, emphasis added)


Undercut the Competition (and unnecessarily* incorporate planned obsolesce under the shroud of the prevention of software piracy)

I further suspect that almost all versions of Windows were or still are offered to OEMs at a(n optional cost) loss, just like Microsoft's game consoles, as you could either buy a free standing copy of their operating systems for a few hundred dollars, or basically get a free computer at the same or less price point by buying a computer that came with it preinstalled.

Aside from software piracy, that's why certain versions of Windows don't like it when you drastically change the hardware in your computer. How would they sell you "The White Album" again if it just still worked when you clicked play (or in terms of any given OS, simply migrated or installed it) on a new device?

[*] Unless you are a gamer or a designer using graphically intensive programs, there hasn't been a need for a "better" computer or operating system for at least the last 20 years.

Functionally, from a UI standpoint, there isn't much difference between Windows 3.0 and Windows 10. Both have a point-and-click GUI. The next step IMO (which I thought would have been motion control, like the Wii) is likely to be thought or eye controlled. So until we invent a better mouse, Microsoft is the company who got it right first and continues to do so, offering us the tools (we don't really need) to satisfy our own 'two-foot-ides'.

Long story short in a single word, be it software and/or hardware: bundling. An extent to which Apple has taken to a new level with their proprietary software and hardware and why their market share has increased so much in the recent past: they took a page out of Microsoft's book. The same one Gorge Lucas was so apt to read: retain the rights to your IP, become a company giant, and use that power that you now have to enslave the universe through keen marketing of the merchandise that you sell, but somehow still own, because your lawyers are awesome.

  • 3
    While it definitely was an important factor in Microsoft's later dominance of the PC software market in general, I don't think Microsoft Office was much of a factor in the success of Windows 3.0; Office only really became important with version 4.0, which was indeed bundled with lots of PCs. (Although the cut-down Word 6.0 / Works bundle was quite popular too.) In 1990, PCs were only starting to be bundled with Windows, and not many had extra software. – Stephen Kitt Mar 4 '17 at 21:17
  • 1
    I'd also disagree with the novelty of bundling PCs with operating systems; all non-kit micro computers came with some form of operating system. PCs were unusual in that there was a choice, although that didn't stop Microsoft from monopolising the operating system market (and then the general productivity software market). – Stephen Kitt Mar 4 '17 at 21:20
  • 2
    It's easy to forget, but both Lotus and WordPerfect dropped the ball on Windows versions of their software and Ashton-Tate essentially dropped the ball completely. At the time Windows 3.0 was released, there was very little in the way of competition for either Excel or Word. (You can argue that this is because of Microsoft's misdirection of the industry towards OS/2, but for single product companies like Lotus and WP, they should've been on Windows as a safeguard, if nothing else.) – mschaef Mar 4 '17 at 22:05
  • 2
    The bundling argument has always been a bit odd to me. Around the time of Windows, it was easily possible to have a DOS, Windows, a memory manager (QEMM), disk utility (Norton), networking stack (Netware), drag drop enviornment (PCTools or NewWave),all on the machine as separately bought and paid for licenses. By the time Windows95 rolled around, it was essentially all bundled into the core OS product for about a quarter the total price. Bundling eliminated some market opportunities, but was wholly better for the consumers.) – mschaef Mar 4 '17 at 22:10
  • 1
    I don't see how a 2000s anti-trust case has anything to do with the sales of Windows 3.0 (or why everyone else was free to bundle any software they wanted, but not MS, but I digress). All around, your answer reads like a rant, which isn't a very nice way to answer questions on the SE network :) If you want to show that Microsoft used dirty practices to become the software leader it did, show the alternatives. Who didn't use proprietary formats? Who didn't bundle software? And are you seriously saying that contracts are evil? Office run with bundled Windows and on Apple PCs. – Luaan Mar 6 '17 at 14:06
3

DOS programs all had unique interfaces, each had their own graphics and device drivers and each had their own fonts. Cut and paste did not exist. Each new program had to take all these into consideration. Windows alleviated this overhead for the developer and at the same time gave the user experience continuity.

Even though you could launch older DOS programs, it just felt dirty and primitive. Better hope your printer was supported and you really liked one of the ten fonts offered ( I'm looking at you Printshop ).

I would have to say that the icing on the cake was NETWORKING. Network setup in DOS was dicey at best and expensive (cough-Novel-cough). With Windows you just bought a cheap network card and it just worked. In my own experience this lead to network gaming.

I guess my take is that not unlike the iPhone paradigm, in it's own way, it created a better ecosystem for both developers and users.

  • Welcome to Retrocomputing Stack Exchange. Thanks for the answer; it explains in detail some of the reasons that people would have prefered Windows. You might want to take a look at the tour to learn a bit about how the site works; it's a little different from forums and Brandname Answers™ sites. – wizzwizz4 Mar 3 '17 at 7:21
  • 2
    Windows 3.0 didn't include networking. That came later, with Windows for Workgroups 3.1. With Win 3.0, you still used LANTastic, Netware, or whatever you wewre using in DOS. – Greenstone Walker Mar 3 '17 at 9:58
  • 1
    @GreenstoneWalker - you are correct, but I would be interested to see statistics about whether Windows "took off" at 3.0 or at Workgroups - my own recollection was that in my company it was at Workgroups... Prior to that only bleeding edgers used Windows, after it, everyone. – Grimxn Mar 3 '17 at 16:57
  • @Grimxn It didn't matter much for homes - most people didn't have multiple PCs. But for a business, it was a huge money saver - not only was it much cheaper than the competition (Novell and friends were very expensive), it also didn't build on dedicated/hired network admins/consultants (which was commonly seen as a great source of revenue at companies like Novell, IBM or Oracle). And when those people decided to buy a home computer, why wouldn't they buy the one they use at work? They already know how to use it :) – Luaan Mar 6 '17 at 14:23
1

Hardware with more RAM could have been a big practical issue. I remember buying a cheap second-hand 80286 PC around that time, with 1Mb of RAM and Windows 2.0 installed - though its previous owner warned me that Windows wasn't useable.

Win 2.0 itself would just about run, but trying to load any application resulted in literally minutes of memory paging before enough of the OS got out of the way to make room for the app.

But of course a 1Mb 80286 PC running MSDOS was a perfectly useable computer - and with a 20MHz clock it was pretty fast for those days as well!

  • 1
    Unfortunately this doesn't quite answer the question. It actually implies that Windows 3 was worse than MS-DOS. Please make it explicit that you're comparing pre-Windows 3 machines to post-Windows 3 machines, and not Windows 2 to MS-DOS. – wizzwizz4 Mar 3 '17 at 16:58
  • 2
    20 MHz 80286? That must have been a pretty late production of the CPU then. (I remember an 80386 that ran at 16 MHz.) And frankly, "literally minutes of memory paging" sounds like an exaggeration given a 1 MB RAM system. Contemporary HDDs at the time might have managed high single-digits megabytes per second throughput; even if you add seek delays due to fragmentation, I have a hard time seeing swapping taking more than maybe a second. And I'm trying hard to remember if Windows 2.0 even supported swapping to disk (as opposed to on-demand page loading and eviction), though it seems doubtful... – a CVn Mar 5 '17 at 12:08
  • 2
    There were 80286 chips clocked as high as 25MHz shipping in mainstream PC's . The Dell System 220 is a good example of a relatively mainstream 20MHz model. These fast 286's all post dated the beginning of the 386sx/dx era, and were mainly positioned as less expensive but still fast DOS PC's for people without the need for 386-specific support. IIRC, a fast 286 could outperform a more expensive 386sx machine. (It was also the case that Windows 3.0 in standard mode could be faster than in 386 Enhanced mode.) – mschaef Mar 5 '17 at 18:57
  • A company named Wietek (or something like that) made those fast 286's. I remember wanting one. – Scott Goodgame Jun 29 '17 at 8:10
1

Many interesting answers here but I would say that the number one reason(s) without any shadow of a doubt is Microsoft Word for Windows and Microsoft Excel for Windows in combination with the fact that Windows 3 was the first really usable version of Windows.

That's probably all there is to it. We can argue about technology and stuff forever, but what sold Windows 3 is a combination of a UI and application software that people found easier to use than the competition.

1

TL;DR application support

I worked for Microsoft in 1991, on their own version of OS/2. OS/2 was designed for the future, and was seen as the successor to DOS. It was never designed to be back compatible with DOS, as it ran in 286 protected mode and DOS ran in 286 realmode. By design, you couldn't mix and match. Eventually OS/2 gained a single DOS box, which kludged the 286 back into realmode, but was clearly a temporary measure. OS/2 was never designed to leverage the virtual 86 functions of the 386. And once IBM designed something, they stuck to it!

By contrast Windows 3 could run in 386 mode and cater for multiple DOS sessions, each running a full application and OS. For a business this was a much better return on investment. Businesses didnt want to dump all their DOS applications and replace them with OS/2 or Windows equivalents, but they did want to run 123 in one window, Word Perfect in another and switch between them. So OS/2 was designed for a future without DOS and Windows was designed to be a better DOS than DOS. The market chose, and were happy to buy 386 machines, for which Windows 3.0 provided more support, in order to do this.

My impression was that MS were slightly caught by surprise and promptly jumped ship on OS/2 when they could see which way the wind was blowing.

  • I thought OS/2 Warp supported the 80386's virtual-86 mode, did it not? – supercat Apr 29 at 21:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.