Windows at 30: Microsoft’s biggest OS goofs
In the high-stakes, high-paced world of tech, surviving 30 years is no small feat. But Microsoft Windows has done much more than merely survive. It has enjoyed a remarkably long ride atop the operating system heap, tripping only recently with the ascendancy of mobile OSes.
[Further reading: Windows through the ages: A visual tour]
Yet, like all juggernauts, Microsoft has been its own worst enemy. Time and again, the company has made missteps with its flagship OS that have left even its most ardent supporters shaking their fists.
In a companion slideshow, I outlined the truly great, revolutionary parts of Windows that have kept us cheering for three decades. But there’s more to the story than adulation and fireworks. Saddle up and take a ride with me down the dark side of Windows history.
Windows 1.01: A buggy slug with overblown sales statistics
The computing world loves to idolize the first release of Windows. But in the not-so-good old days, few people paid attention to Windows when it first arrived. Most who did buy it were treated to a glacially slow, buggy, crash-prone operating system, teetering on top of MS-DOS -- which itself was no piece of cake.
The first version of Windows displayed only tiled windows (except for dialog boxes, which could pop on top of the programs below). The number of marginally useful programs it could run could be counted on one hand. Even Microsoft’s Multiplan (see “25 years of Microsoft Office roadkill”) bit the dust in the Windows 1.0 timeframe, in no small part due to its instability.
Peggy Watt, in an April 13, 1987, Computerworld article, explained, “Some vendors claim a love-hate relationship with the Windows concept and its proponents. They complain about pressure from Microsoft to develop for its pet project or say that writing to Windows is too difficult to be done quickly. Even those in the industry who say they like the idea of a universal graphical interface are not certain the winner is Windows. The result is a lull in Windows activity. Infocorp, a market research firm in Cupertino, Calif., estimates 85,000 to 100,000 copies of Windows are actually in users’ hands. That is only 20% of the 500,000 copies Microsoft claims to have shipped, including OEM-licensed copies and runtime versions bundled with Windows applications.”
The more things change ...
Windows 2: Microsoft as black widow
Before Windows 2 appeared, Bill Gates and crew signed a copyright agreement with Apple that let Microsoft use certain aspects of the Lisa and Mac UI -- drop-down menus, tiled windows, mouse support, among others -- in Windows 1.0 and all future versions of Windows.
Windows 2.03’s defining feature, overlapping windows, happened to be a main selling point for Apple’s Lisa and Mac OS. Apple filed a lawsuit in 1988, claiming that Windows 2.03 (and, later, Windows 3.0) stole the “look and feel” of the Apple OSes, and Win2 "embodie[d] and generated a copy of the Macintosh" on PCs. Apple sought damages for 189 infringed patents.
Years later, the judge dropped all but 10 of the copyright claims, ruling that "Apple cannot get patent-like protection for the idea of a graphical user interface, or the idea of a desktop metaphor,” although he found that the trash icon and folder icons were infringing. Most of the copying was determined to fall under the original license agreement, and the copyright allegations generally didn’t apply. Apple lost on appeal (35 F.3d 1435 9th Cir 1994), and a further appeal to the Supreme Court was denied.
Set against that backdrop, Xerox sued Apple in December 1989, noting (accurately!) that Apple had borrowed heavily from Xerox’s PARC Star computer. That suit was largely dismissed in March 1990.
In 1997, Apple and Microsoft buried the hatchet, at least in public, dropping all lawsuits. Apple made Internet Explorer its default browser. Microsoft agreed to keep building Office for the Mac. Microsoft sweetened the deal by buying a $150 million nonvoting block of Apple stock.
Did Microsoft swipe Apple’s design, ideas, or code? You’ll find that question popping up over and over again -- legitimately -- particularly in the Windows milieu.
Windows 3 and the OS/2 connection
No telling of the Windows 3 story is complete without looking at its relationship to IBM’s OS/2. In 1985, IBM signed a “Joint Development Agreement” with Microsoft to build an operating system worthy of IBM’s new Personal System/2 line of computers.
At the time David Weise and Murray Sargent cracked the 640KB memory barrier in 1989, Microsoft was building IBM’s OS/2 -- viewed by many as the next big thing in operating systems, the rightful heir to DOS. The two companies together built OS/2 versions 1.0 (1987), 1.1 (1988), and 1.2 (1989).
OS/2 was a huge, pivotal contract for Microsoft, but in the end, Microsoft wanted to create its own operating system, not build one for IBM. It ditched IBM and went its own way.
Details of the death dance remain muddied, but at one point in the early 1990s, Microsoft and IBM announced that IBM would develop OS/2 2.0 as the successor to OS/2 1.3 and Windows 3.0. At the same time, Microsoft would build OS/2 3.0, destined to replace OS/2 2.0. Yes, the plan sounded as harebrained then as it does now.
None of that ever came to pass. IBM published OS/2 2.0 in 1992, but Microsoft had already moved forward with Windows 3.1 -- and started a small project to rewrite Windows, named Windows NT. Under the terms of the divorce, both Microsoft and IBM were allowed to use DOS and Windows code -- likely another impetus for Microsoft to create a new version of Windows from the ground up.
I still haven’t decided if that was a Microsoft misstep or a smart pivot.
Cairo: Leading developers astray
In the early 1990s, Microsoft worked Windows nerds (like me) into a frenzy of anticipation by leaking tons of information about “Cairo,” a loosely defined set of features that was supposed to turn into a product. Alas, Cairo never materialized. Some of its pieces -- defined, undefined, speculated, regurgitated -- appeared in other products, but Cairo itself faded into the desert dust.
All this happened in the soon-to-be-explosive atmosphere between Jim Allchin and Brad Silverberg (see next slide). Gates hired Allchin in 1990 to take the reins of Windows NT 3.5, then he was put in charge of Cairo. Gates hired Silverberg in 1990 to lead the DOS and Windows efforts. After Win95 shipped, with an eye to the cloud, he led the newly created Internet Platform and Tools Division.
At the Win32 Professional Developer’s Conference in July 1992, Allchin gave a presentation about future Microsoft OS Cairo and how it would fulfill Gates’ vision of “information at your fingertips.” (Unfortunately, I can’t find a transcript of his talk, but Bill Gates’ keynote, which mentions Allchin’s presentation is available on YouTube.) In mid-1992, Cairo was viewed as a mainstream operating system, apparently a modified version of Windows NT. At the Dec 1993 Cairo/Win95 Professional Developer’s Conference, Allchin showed off a demo running Cairo. But it never came to pass.
Cairo was many things to many people, but the glue that held much of it together was the Cairo Object File System, later renamed “WinFS” for “Windows File System.” Jon Udell wrote a Sept. 7, 2005, column in InfoWorld that goes into many of the details. At one point, Bill Gates was quoted as saying that Windows Media Player, Photo Gallery, Office, and Outlook would use WinFS. Even though Gates continued whipping up the troops a decade later -- his 2003 Professional Developer’s Conference announcement is a prime example -- WinFS died a painful, public death, although (as you will see) it was resurrected for project Longhorn. Developers who lived through the spectacle and the anticipation never viewed Microsoft the same way.
In the early 1990s, many of us thought that NT 4.0 would be (or resemble) Cairo. NT 4.0 arrived and it looked a lot like the Win95 UI on top of NT 3.5. We figured NT 5 (Windows 2000) would be Cairo -- nope. Windows XP and Server 2003 didn’t look anything like Cairo. There was an official mention of Cairo’s demise on the Channel 9 MSDN channel, “A few words about WinFS: The project is CLOSED,” posted in Decemeber 2006, but even that post has been pulled.
Allchin vs. Silverberg, dirty laundry, and who was right?
Through most of the 1990s, the ongoing battles between Jim Allchin and Brad Silverberg rocked the foundations of Microsoft -- as well they should. In a nutshell, Allchin pushed for Windows NT to stay ultrareliable, as it extended from corporate servers onto consumer machines. Silverberg was more interested in pushing a consumer-friendly Windows into the cloud -- which is to say, making parts of Windows that people want to use run on Microsoft’s computers. The two goals weren’t mutually exclusive, but they offered lots of contention.
A big part of David Bank’s book "Breaking Windows: How Bill Gates Fumbled the Future of Microsoft" discusses the ways Gates, Allchin, and Silverberg bobbed and weaved through the ’90s: “To the public, Microsoft appeared to be competing with rivals such as Apple, IBM, Novell, and Sun Microsystems. But for much of the 1990s, the biggest competition was inside Microsoft itself, between the operating system that commanded more than 90 percent of the market and the one seeking to displace it.”
Allchin went on to play a key role in the US v Microsoft antitrust trial (see next slide).
Internal conflicts are nothing new, but this rivalry riled developers, both inside and outside the company. In the end, Allchin won, with Silverberg leaving on sabbatical in 1997. Allchin and Brian Valentine went on to ship Windows 2000 -- a troubled consumer version based on Win NT -- then the disastrous Windows ME, which was still based on DOS. The first realization of Allchin’s NT-founded consumer-friendly version of Windows was XP, which shipped in October 2001 to wide acclaim. Allchin left in September 2005.
Many developers and pundits (me included) favored Silverberg’s approach. Seeing him maneuvered out of the driver’s seat left many of us upset.
US v Microsoft: Cringeworthy courtroom drama
The Department of Justice filed an antitrust case against Microsoft in May 1998, claiming that Microsoft had turned Windows into a monopoly and, using that monopoly, had spread Internet Explorer illegally, crushing other browsers such as Netscape Navigator and Opera. The case kept hundreds of lawyers and their staffs fully employed until September 2001, and stringing along until the final ruling in June 2004.
Microsoft’s mistake was to stick IE into Windows, pretending as if IE couldn’t be deployed independently, in an effort to hobble Netscape. Oh, and leaving behind a trail of embarrassing emails and providing very poor performances -- yes, performances -- on the stand didn’t help. Not to mention lying in court and getting caught at it ... twice.
Guilty, your honor.
A whole cottage industry of published analysis of the case ensued. My take is pretty simple: Microsoft was found to have a monopoly in Windows, which isn’t a crime. Microsoft used that monopoly to crush Netscape. Microsoft execs stupidly documented every step of the process in emails that ended up in court. None of the Microsoft senior executives came across well on the stand. The doctored “demo” video (actually, videos) was an amateurish attempt that still embarrasses me, personally.
Microsoft got off the hook in 2001 because the new Bush administration’s DOJ didn’t want to pursue the case.
The effects of the antitrust suit are hard to underestimate. Microsoft got its fingers burned, badly, and apparently refrained (at least publicly) from all sorts of anticompetitive behavior for quite some time, changing the course of Windows history.
Internet Explorer: The kludge that keeps on kludging
Quite possibly Microsoft’s greatest flub of all time, Internet Explorer deserves a public flogging and a quick demise -- one that appears to be in progress, with the ascendancy of Microsoft Edge. (Yes, I’m biased. I’ve been recommending that people ditch IE since the early days of Windows XP.)
Internet Explorer 1 arrived in August 1995 as part of the Microsoft Plus Add-on for Windows 95. But it wasn’t until Version 3, when Microsoft started down the slippery slope of building a better Internet, unilaterally, that the pig started to show beneath the lipstick. IE3 included ActiveX technology (which became, arguably, the single largest source of Windows infections), the third-party infection harboring frames, and support for helper applications.
Version 4, in September 1997, shipped in Windows 95 OSR 2.5, and with it came the antitrust lawsuit. (See the previous slide.)
But the real pain came when IE6 shipped with Windows XP in August 2001. IE6 may have been Microsoft’s biggest misstep of all time. With it, Microsoft lost an enormous amount of goodwill, as users began to understand that their computer was at risk because of a bad piece of Microsoft software. (Don’t get me started on what developers think of IE6.)
From my point of view, ever since IE6, Microsoft has blatantly put its own financial interests ahead of its customers’ security, for about a decade and a half.
Longhorn: The Ballmer brahma
In late 2013, Mary Jo Foley at ZDNet ran an insightful interview with departing Microsoft CEO Steve Ballmer. In it, Ballmer identifies project Longhorn as his biggest regret.
Foley quotes Ballmer as saying, "When I look at it and I say, OK, what's the thing that I did that I feel -- that I regret the most, not just in my CEO-ship but my whole time here, it's absolutely 'Longhorn becomes Vista.' That was the single biggest mistake I made… Why? Not only because the product wasn't a great product, but remember it took us five or six years to ship it. Then we had to sort of fix it. That was what I might call Windows 7. And what we wound up with [was] a period of let's say seven or eight years where we had the A-team -- not all of the A-team but a bunch of our best people -- tied up not driving. We did not make years progress in eight years, and there were other things those people could have been working on, [like] phones… It wasn't Bill [Gates'] thing and it wasn't Jim [Allchin]'s thing and I didn't get it.”
That was precisely the problem. Longhorn was a massive time sink, basically impossible, and Ballmer didn’t get it.
That didn’t keep him from dangling Longhorn in front of the Windows peanut gallery. We were told, starting in July 2001, that Longhorn would be the know-all, be-all upgrade to Windows XP, Server, and Office -- basically, a feast on a technologically superior silver platter. Longhorn was supposed to include WinFS (see previous Cairo vaporware slide), with a new UI, 3D video (gulp!), vastly improved security, multimedia galore, automatic bug reporting, and a new set of APIs that would replace the Win32 API, called WinFX.
Microsoft didn’t do any of it. Many of us got suckered -- I was posting about Longhorn as late as April 2005. Instead of Longhorn, Microsoft left us with the biggest pile of bull pucky since Windows ME: Windows Vista.
The mess was so complete that Ballmer brought in the big guns to clean it up. Steve Sinofsky, basking in the limelight of a remarkable string of successful Office versions, moved to the big Windows chair. He and a large handful of talent took the mess that was Vista and, in three years, gave us a far better product in Windows 7.
Windows 8 and Windows RT: Killing the Windows brand
No discussion of Windows ire would be complete without a tongue lashing in the direction of Windows 8 and its not-really-Windows henchman, Windows RT.
Steve Sinofsky and crew, fresh off Windows 7 and a string of Office cash cows, decided to create a new interface for Windows. What you see is what we got.
How badly did we hate it? In an InfoWorld article, I brought together 14 well-known Windows book authors and asked them to tell the truth about Windows 8. A few of them said nice things. Most of them, though, came up with the same reaction: What could Microsoft possibly be thinking?
No doubt you’ve come to your own conclusions about Windows 8. Sales of Windows and Windows PCs were already headed down. We’re lucky Win8 didn’t sink them entirely. Time will tell if Windows 10 is strong enough to stanch the downward trend.
As if killing Windows wasn’t punishment enough, there’s Windows RT. Why would anyone in their right mind name an operating system “Windows RT,” knowing full well that it won’t run Windows programs? Beavis, meet Butthead, and a billion-dollar write-off.
Satya Nadella has seen a meteoric rise, in a few short years, from the head of Bing to CEO. Terry Myerson, who was in charge of Windows Mobile -- an enclave outside of the direct Sinofsky empire -- is now in charge of Windows and hardware. People who weren’t as well known in the Sinofsky era have taken on new roles and performed brilliantly. Microsoft has opened up its development process to the public at large, without the excesses of Cairo and Longhorn. Things are looking much better now than they were in the days of Windows 8.
Guess it’s time for the next big flub.
Today's Best Tech Deals
Picked by PCWorld's Editors