Judge Judy, please don’t break up Microsoft.
We have them just where we want them, stuck in a corner with wet paint drying around them. Clinton’s antitrust pinstripes may have sought their place in history by making history of Microsoft. But strange as this may sound, an intact Microsoft may well have done more than venture capital to create a continuum of value in the technology world.
I have never cashed a check from Microsoft, never owned the stock, could care less if they like me or hate me. I do believe that the way they have structured the PC and software market, rather than stifling innovation, stimulates it. I also believe that a divided Microsoft is a dangerous, hungry beast, while an intact Microsoft is stuck in a position where it can do little harm and significant good. To prove it, we need to go back in time.
At first, Microsoft just wrote compilers for the BASIC programming language. Bill Gates promised IBM a disk operating system (DOS), pushing aside Digital Research’s version. So Microsoft wrote simple utilities to get programs and files on and off floppy disks. Truth be told, as a 16-year-old “Homebrew” computer nut in the mid ’70s, I did some of the same stuff Gates did, assembling computers based on early Intel and Zilog microprocessors so my friends and I could play computer games. We even formed a company named MicroTek so we could steal chips from trade shows.
Unable to afford a floppy disk, we rigged up a Panasonic cassette player for storage. In a few afternoons at my kitchen table, with just a manual to look up machine code, we wrote a COS, or cassette operating system. We were going to write a BASIC compiler, but found it easier to bootleg one written for the Altair and IMSAI, the first two home computers. Turns out it was Gates’s BASIC compiler (just try to come after me), which he has given away with DOS since then.
I was too busy ogling 18-year-old girls to take off a week or so to write real code. But if a scrawny 16-year-old kid like me could contemplate writing an operating system, many thousands of others out there could have done it. But no one did, for a while anyway, and Microsoft gained a decisive lead.
The magic of Microsoft, though, was not writing software but erecting it as a “platform.” Gates’s basic idea: Keep others out of the operating system business by making it easy for them to use yours. First, Gates had to contrive barriers to entry: added functions such as compression of files, scanning for viruses, utilities for communications, backup, and more. So now an enterprising firm of 16-year-old Gates wannabes would have to build not only a new OS but all these extra functions too, upping considerably the money and brain power required to do battle on Microsoft’s turf. It was easier for the brainiacs to chase after the unoccupied territory down the road.
Also, Microsoft built hooks into their operating system, formally known as an application programming interface or API, so that others could more easily create applications that sit on top of its operating system. Gates made sure developers at other companies don’t have to reinvent the OS wheel, coming up with their own code for handling graphics, getting files off a disk drive, dealing with modems, etc. So DOS and then Windows were wonderful platforms for others to build valuable businesses on, whether by writing applications software like Lotus, databases like Oracle, or pitch-books for IPOs as my erstwhile company Morgan Stanley did. Gates, in short, created an ecosystem, with him at the center, and others busily building out from it, extending his domain.
To protect the platform, the boys in Redmond successfully implemented the Gates Doctrine: “Keep away from my domain—the desktop operating system and anything easy that I will pull into the operating system. Instead, go innovate and move the platform to a higher plane.”
The Gates Doctrine was patterned on the Monroe Doctrine which told the world: “The Americas are our area of influence; everyone should go away and play in some other sandbox.” (Monroe was a little more eloquent.) As the Microsoft Bookshelf CD-ROM Encyclopedia comments, “Although never formally recognized in international law, the doctrine was invoked successfully several times and became important in U.S. foreign policy. As imperialistic tendencies grew, the Monroe Doctrine was viewed with suspicion by Latin American countries, who associated it with the possible extension of U.S. hegemony.”
Gates’s sandbox is the PC operating system, first DOS and then Windows. >From IBM to Novell, Sun and Oracle, and now the Linux gang, others have fought for desktop Operating System real estate. Building it was not that hard. No longer could a 16-year-old at a kitchen table contemplate doing so, but a scrawny 16-person department could. All were rebuffed in the market by Gates, who insisted they go play elsewhere. “Stay in Europe (UNIX), with all its various cultures and languages. The continent (PC) is mine, don’t go there….”
Bundling all the easy pieces, Microsoft contrived a more complicated and functional, if not more robust, operating system. Then the company sent its minions to bully PC manufacturers to use the Microsoft OS and only its OS, all or none.
But Microsoft is not content with only one continent, you say; they have shown “imperialistic tendencies,” raising “suspicions” among rivals who “associated [the Gates doctrine] with the possible extension of [Microsoft] hegemony.” It appears to many techies, and at least one judge, that Microsoft is an imperial predator, slowly but surely pulling other software applications into its OS monopoly. The fact that Windows includes TCP/IP networking, enables Internet access, has a browser, imparts virus protection, offers a kitchen sink and a full set of appliances, including virtual toasters for rival programs, is fuel for the government’s claim that Microsoft’s hegemony over the rest of the software industry is so extensive that it crushes newcomers and their potential innovations.
The reality is almost the opposite. To begin with, Microsoft’s policy of sucking new and relatively simple functionalities into the operating system is driven at least as much by Moore’s Law—ordaining a doubling of the number of transistors on a microprocessor every 18 months—as by the Gates Doctrine. That incessant doubling of processing power impels a constant expansion of the functionality of the central processing unit that the operating system serves. To bar the inclusion of new functions in the OS is by implication to ban Moore’s Law itself as a form of illegal bundling.
More important, by sucking the easy stuff in, Gates forces others to innovate more deeply. Even with 10,000-plus programmers, Gates can’t make everything himself. So he practices what he calls embrace and extend. Find what he needs, embrace it as his own, and then extend the functionality throughout his software. Microsoft uses this as a tough, competitor-crushing strategy, but the consequence is to push potential rivals to more difficult (and therefore useful) creative feats. So because of Microsoft, Sun focused on servers for databases rather than desktops, and was ready when the market for Internet servers exploded. And AOL built their interface on top of Windows, rather than reinventing an OS, and focused on developing original content. Meanwhile, with no potent competitors for its PC operating system (Apple is irrelevant), Microsoft gained complete ownership of the desktop platform business.
Microsoft commanded an ever-rising stock price, which it used to print more currency to lure more programmers and employees. “Be a Microsoft millionaire” proved an even better recruiting slogan than “Be all that you can be.” Although no company can control its own stock price, Microsoft got good at manipulating it.
I made several trips to Redmond for analyst meetings. I chiefly followed chip companies like Intel, whose fortunes were intertwined with Microsoft’s, so in a way I was slumming, but all the Microsoft analysts were there, hanging on every word.
Just before issuing stock options to new recruits, key employees, and of course, management, Microsoft would contrive to trash its own stock. I vividly remember Jon Shirley, then Microsoft president, starting his talk in 1989 or 1990 on a grave cautionary note: “You analysts who have the high earnings estimates, and you know who you are, need to bring them down, they’re too high.” I was among the few who heard the rest of the presentation, because all the analysts skulked out to call their trading desk.
Within a few minutes, Microsoft’s stock dropped six points, which at the time meant a couple of billion dollars of market cap gone up in the smoke of Shirley’s prose. When the analysts returned, the room was glum. A few minutes later, though, I encountered Bill Gates and Frank Gaudette in the hall, laughing uproariously about the stock price dip and the gullibility of moron analysts.
It was not until the early- to mid-1990s that Microsoft, now led by Steve Ballmer, got greedy. No longer content with 20 percent sales growth, and unable to raise prices enough to reach 40 percent, Microsoft extended the Gates Doctrine to include desktop applications.
The world didn’t need ten word processors or 15 spreadsheet programs. Microsoft bundled Word and Excel into Office Suites, squeezing the air out of the market for stand-alone office programs such as Borland’s Quattro Pro spreadsheet, which Phillippe Kahn had introduced at a threateningly low $89 price point. Borland later tried to hide in databases. Lotus went after “Groupware,” or software enabling groups of people easily to share documents and processes. By focusing on “Notes” Lotus saved itself, eventually setting up a $3 billion acquisition by IBM. Novell, which owned the file and print networking business with Netware, never did recover when Microsoft finally fathomed networking, and pulled those functions inside its sphere of influence. Novell’s then-CEO Ray Noorda was obsessed with Microsoft. He tried to do battle with a compatible OS, DR-DOS from Digital Research. He bought WordPerfect and Quattro Pro to create a suite, but in the end, Novell managed only to create two anchors to pull the company down faster. They later dumped the suite on a Canadian company called Corel, which also suffered the anchor effect.
Microsoft’s classic embrace and extend: Find an existing market, embrace it as its own, and then extend the functionality beyond its original capabilities, into the OS or into a Suite. If it is easy to do (the 16-year-old kid test), Microsoft is going to do it and put you out of business, unless you move up, in degree of difficulty and functionality. I can’t tell you how many companies I visit that explain they don’t compete with Microsoft, but they insist, with a nervous tick, they are always on the lookout.
Since the days of chariot building, most industries follow predictable phases of creation, rapid growth, and proliferation of entrants doing minor variations on the original concept. Overcrowding follows and leads to consolidation. The result is an industry of two to five large players, as in autos or airlines.
Despite Microsoft’s size and strength, or even because of it, the opposite has happened in tech, even in software. New entrants flood the field. Hundreds of significant technology IPOs have been issued just since 1995. These new players are not doing slightly different spreadsheets, they are doing supply chain management, Web caching, self-service applications, and an ever-expanding array of communications products. At least until this year, the stock market, recognizing the creation of new value, has made it easier, not harder, to create a company with a $1 billion market capitalization.
But technology life cycles are shorter than the life spans of the stocks that value them, so there is no rest for the weary. Milk a supposed franchise and Microsoft duplicates your product, embraces your space and squeezes you out. With stand-alone products mostly untenable and flash-in-the pan IPOs getting crushed, the intelligent entrepreneur must seek out just the kind of long-run, secular-growth path that normally emerges only from a fundamental innovation.
In the mid-1990s, many observers believed that the Internet browser constituted such an innovation. But Gates missed it. In Redmond, where all compatibility issues were solved by mandating Windows, the Microsoft networking division has always emitted a giant sucking sound, with a capital S. Every year Gates and Shirley and their team would visit big corporate accounts and pitch their LANMAN networking products. And every year, corporations using Novell would yawn. It seemed that by the mid ’90s, Microsoft had finally given up.
And then little signs emerged that the PC platform could do real networking (not just the lame file and print sharing that Novell offered).
At the University of Illinois, Mark Andreessen (or was it his friends?) cobbled together a browser that worked using the TCP/IP protocol. An underfed little firm named NetManage wrote a driver to do TCP/IP for Windows, selling them like hotcakes through trade magazine ads. Also building on the Microsoft platform, Andreessen (with Jim Clark’s money) created Netscape, which produced a stable browser for Windows and presented it as a new platform.
Once Gates realized how easy this technology was, Microsoft went out for a new episode of embrace and extend.
Netscape’s browser proved no match for the Gates Doctrine. With only 10,000-plus lines of code, Microsoft did not break a sweat duplicating its features. In the Justice Department’s gimlet eye, Microsoft crossed the line by integrating the browser with the operating system. But Netscape’s mistake, which became Microsoft’s mistake—and eventually Justice’s mistake as well—was to think the browser was another platform. It wasn’t. It was just a feature.
Why? Any company developing new Internet applications or services, from Amazon to GoneInaFlash.com, does all its creative work on servers. Developers only need to know that a browser exists, they could care less which browser it is. While Netscape threatened to turn its browser into an OS, it turned out to be an empty threat.
In the end, Microsoft did not kill Netscape. Netscape killed Netscape…by invading Russia, aka IBM. IBM/Lotus was doing everything it could to help Netscape fight Microsoft, bundling its browsers with IBM PCs, making sure code was interoperable, cooperating on new cross-platform Java programming tools.
Taking the company public, however, Netscape CEO Jim Barksdale gained a high-priced asset. Inspired by Cisco Systems’s success, he wanted to wield it to build an empire of acquisitions. The temptation was Napoleonic and fatal.
I attended an investment bankers’ conference where Barksdale declared: “We are going to buy a company a quarter, and then when we get good at it, we are going to buy one a month.” He got a standing ovation. Most things he bought were stupid, but harmless. Then Netscape bought Collabra. Run by Eric Hahn, who had sold cc/mail to Lotus, Collabra gave Netscape a choice new corporate product in the Groupware business. But Barksdale forgot his ally IBM had shelled out $3 billion to gain Lotus’s Groupware trade. When Netscape became a threat to Lotus/IBM, Netscape soon learned the lessons of Napoleon. Next thing you know, IBM is bundling with its PCs not Netscape Navigator but Microsoft Explorer. Having invaded Russia, Netscape underwent a long winter retreat before finally falling exhausted into the arms of AOL. I guess Barksdale never did get “good at” acquisitions.
For all their high-profile drama, the browser wars were always peripheral to industry evolution. The real work on the Internet takes place behind the scenes, in data centers, in server farms, where the Internet wizards work. The server side, as it is known, is where Web pages are created and served, where databases reside, where petabytes (10 to the 15th) of disk storage spin away.
Since Microsoft has been found guilty of monopoly abuses, it now enters these new markets with a choke collar on. Microsoft owns the desktop platform, and I have a feeling it is stuck with it.
Innovation has moved elsewhere. New platforms are being created at an unprecedented rate, with Microsoft mostly on the sidelines. And the funny thing is, while none of these new platforms were easy enough for a scrawny 16-year-old, or even a 16-person firm, to create, I’ll bet that most of them were initially done with less than $16 million of investment capital. Microsoft has $23 billion in the bank. Netscape came close with a couple of million from Jim Clark and five million from Kleiner Perkins. These new platforms weren’t all that hard to do.
Once started, a new platform gains momentum as developers write applications for it. Switching costs are high. What worked for Microsoft with DOS and Windows now works against it on these new platforms. These are not point products that Microsoft can just embrace and extend. They represent a decade or more of upgrades and new functions.
Web servers are more important than browsers, and when you pull up a Web page today, that page likely does not come off a Microsoft-based server. According to SecuritySpace, as of December 2000, 58 percent of the time Web pages come off a freeware Apache server; it works best, especially for the price. Microsoft IIS server is used only 28 percent of the time; Netscape’s server, 4 percent. These are still Pentium-based machines, PCs with big disk drives, but without Microsoft code.
I took an early tour, circa 1996, of Exodus Communications, a Web data center hosting company: Think big air-conditioned rooms with fiber and Ethernet cables running to servers in strange metal cages, one for every new dot-com that came down the pike. Exodus provides the central office space for the new Internet phone companies. Microsoft has only been a marginal player on the server side. In what has telco-like requirements, there is a desperate need for scalability, reliability and cost effectiveness. Microsoft won on the last point only, and now even that is questionable.
The cages were locked, to prevent competitors from messing with each other’s machines. What caught my attention were the pool cues apparently hanging everywhere. “What’s with the pointers?” I asked. The answer was NT. Huh? Windows NT was Microsoft’s solution for a server-side operating system. Unfortunately, it was as reliable as a Yugo. Turns out that Exodus would get 20 calls a day from customers to reset their crashed NT servers, which meant getting the combination, unlocking the cage, pushing the power button first off, then on. With pointers, you could do that through the cage. (Imagine a guy on roller skates cruising around resetting NT boxes with the pointer.) Down servers equal lost sales. Guess how long it would take to replace those NT machines with anything reliable? Enter Linux.
Running on the same Pentium PCs, Linux was cost effective, but also reliable. Even better, due to its inherent networked architecture, Linux was scalable, meaning not only could you toss your NT boxes, but even your expensive Sun servers could be replaced with a rack of Linux machines. Linux was developed by an open-source community. Rather than one company doing product development, Linux development is broken up into pieces, with code worked on by lots of people, in their homes, in Starbucks, in a café in Athens—and the best code is used to create the ever-improving product.
Once you go Linux, you never go back. Apache is on its fifth or sixth iteration. Linux is probably on its tenth significant release. So an entire server-side computing architecture has moved quietly away from mother Microsoft.
And now those servers are serving up Web pages and information not only to PCs, but to millions of PDAs and pagers, and a billion or so cell phones, and hundreds of millions of set-top boxes. Or to billions of voice phones, by using Internet Protocol and digital speech processing. While PC to PC connections are still vital, the proliferation of other Internet access devices diminishes the PC’s stranglehold on consumers and Microsoft’s stranglehold on software.
PDAs, or personal digital assistants, were an early target for Microsoft. Microsoft came out with a consumer electronics version of Windows, known as WinCE, which was supposed to make competitors wince. But instead consumers did. WinCE has yet to happen. No one wants to run Excel on a PDA. As of December 2000, according to NPD Intelect, Palm commanded 67.5 percent of the market; Handspring (using the Palm OS) had 16 percent, and Microsoft WinCE platforms, mainly Casio, held the dwindling remainder.
Apparently an easy target for Microsoft, Palm is most likely the new millennium’s Apple, with hubris and packaging more important than technology. Licensing the Palm OS to Handspring, Sony, and others, Palm did reverse one key Apple error. Unfortunately, connectivity is a problem (since they must use a sideband of the cellular network for data), size is a problem, stylus input is a problem, and they are still beating the pants off Microsoft.
Pagers became an interesting platform in the last year or so. Since they always delivered data, as soon as they became two-way, they became an excellent form factor for e-mail, which is mainly text. Canadian RIM (Research In Motion) came up with a two-way e-mail pager, and rich executives on expense accounts fell in love with it. The operating system inside is RIM’s. They expanded to the Blackberry 957, which looks like a PDA but is really a dressed up pager. Again, the OS was developed internally.
Philippe Kahn, the guy who started the PC software price war with Quattro, went on to create Starfish. Starfish launched an operating system for credit card-sized PDAs. He once told me that he was interested in doing operating systems for a system where an AA battery looks like a nuclear device. The REX was literally a PCMCIA PC card, which fits in a shirt pocket with room for pens, and you could access phone numbers while driving and downshifting from 3rd to 2nd gear when traffic stopped on the 101. He was set to add connectivity when he sold the firm to Motorola, which still has plans to use the operating system and PDA features in a next-generation cell phone. It just keeps being the future next generation.
Cell phones went digital in the late 1990s, mostly to squeeze more users on over-extended cellular systems, but once digital, there was room for an operating system and applications. Phone.com, now OpenWave, created a Microbrowser for cell phones and freely licensed their WAP (or wireless application protocol) to 30-plus phone makers so they could sell server-side systems and services to 60-plus operators. By the summer of 2000, they touched 80 percent of Web phones. In the end, WAP was a dud. It had no good applications and was too cumbersome for anything more than stock quotes. A consortium of the major cell phone vendors, Nokia, Motorola, Ericsson, and Matsushita, was organized to create a real Web phone operating system. Based in Cambridge, UK, Symbian is working feverishly to launch an industry standard. But, of course, so is each consortium member in their own labs. Microsoft? They are watching with interest.
Set-top boxes, those old ugly analog things used for cable TV, are quickly converting to digital cable and satellite receivers for DBS—direct broadcast satellite. DBS is already digital. Here is a market that Microsoft has a chance in. Because Microsoft has a great product for it? Heck no. Because it has cash. The cable business is run on the premise that you load up on debt and keep buying more systems, so you never have to show a profit or pay taxes. But cable constantly needs money, especially for anything that smells of improvement over basic video. Microsoft bought a design win for the WinCE operating system in digital set-top boxes at the mostly old and tired TCI and Media One cable systems now owned by AT&T. In May of 1999, Microsoft invested $5 billion for AT&T shares (at $55-$60 a share, oops) for the rights to sell 2-1/2 to 5 million copies of WinCE on Motorola/General Instru-ments DCT5000 digital set-top box, on top of an existing 5 million unit order from 1998, for a potential total of 10 million copies. Even if they charge $50 a copy ($5 is more like it), that’s $500 million in revenue. This is not a good use of $5 billion, since they have already lost $2.5 billion in value, or $250 per set-top box. A few companies such as OpenTV and Liberate have a chance with real technology, if only they had a couple of billion each to buy design wins.
PVR, or personal video recorders, are what set-top boxes will eventually become. Pioneered by Tivo and Replay, PVRs make great products. However, Tivo/Replay still are questionable businesses. The TV model of prime-time viewing was fine when that was the only thing to do (after sex and reading a book), but e-mail and Web surfing have eaten into viewing. PVRs allow users to record shows and view them at their leisure and zap commercials. With my Tivo, I record the one-hour Letterman show every night and watch it the next day in 20 minutes. You can’t believe how much dead time is lost to laughter and applause. Microsoft does play in this market, almost, with their Ultimate TV (basically a PVR with Web access from the WebTV purchase in 1997). Better late than never, I suppose. Sony, Panasonic, Phillips, the Tivo and Replay licensees (and probably owners if they go under) will be the market.
Residential gateways will become the terminating point for digital subscriber lines (DSL) and cable modem connections. The high-speed connection running the TCP/IP protocol will enable not just Internet access for PCs, but four or more voice channels over IP phone lines, a few video channels, and whatever new applications come about. These under-$300 boxes can control the whole house network, and the operating systems inside are still home grown, made by companies like 2wire and Telocity.
Voice User Interface, or VUI, brings the rest of the world into the Internet. While a juicy 100 million PCs are sold a year, there are billions of phones around the world. Though optimized for person-to-person voice communications and occasional interactive voice response (IVR—i.e., press 1 if you hate being on hold), a phone can be turned into an Internet-access device via voice recognition for input and text-to-speech for output. That day has arrived, and the voice user interface can do for the phone what the GUI (graphical user interface such as Windows) did for the PC. The two leaders in voice processing are Nuance and Speechworks. Add to which companies such as General Magic that reinvented themselves to implement VUIs for phone users, and for in-car service via General Motors’s OnStar system.
Other digital platforms where Microsoft is virtually nowhere to be found: camcorders, DVD players, MP3 players, digital TV, digital still cameras, video game consoles (although the big, expensive X-Box game player launch will test Microsoft’s ability to crack new markets).
Add to these platforms databases, customer relationship management software, content delivery, streaming media, encryption, soft switches, and more. Microsoft is no longer the dominant player, period.
This doesn’t mean that Microsoft is not trying its hardest to penetrate these platforms. Microsoft has taken a page out of the IBM playbook and is today FUDing the market—spreading Fear, Uncertainty, and Doubt—mostly by pre-announcing products and architectures to freeze the market over. Microsoft’s .NET initiative is a great example: embracing the industry standard XML database and promising to extend it to solve all the Internet’s problems. The .NET Hailstorm service promises to seamlessly communicate with any device, anywhere. Its slogan, “Making the OS the online service,” is as blatant as running through Virginia yelling “We are going to kill AOL.”
Hey, maybe this will all work and Microsoft will dominate the next generation of platforms. But the devil is in the details. Nothing ships until 2002 at the earliest. Digging deeper reveals all sorts of products that run on PCs and begin with the name Windows—same as it ever was.
The PC market is flat to down. Microsoft is scrambling to embrace or define next-generation platforms, but is losing. And the very visible Justice Department trial means Microsoft can’t coerce its way into these new markets and platforms, and since embrace and extend won’t work, Microsoft has to innovate for a change.
The Gates Doctrine has almost too successfully sparked innovation away from Microsoft. As long as Microsoft remains intact, it has an unavoidable, built-in incentive to milk the existing PC platform as long as it can. Developing a new platform would slaughter the easy PC cash cow: the pure profit derived from selling operating systems onto 100 million PCs. If the government breaks up Microsoft into an OS company and an applications company as proposed, the new MS application company would have an incentive to sell not only to PC makers (a declining business it would abandon quickly), but to makers of cell phones, servers, set-top boxes, etc. With deep pockets and unhampered by Justice supervision, this new Microsoft would be a vicious competitor, running rampant over the hundreds of smaller, more innovative companies that now develop for non-PC platforms.
A united Microsoft, on the other hand, has little incentive to compete with itself by expanding and marketing non-PC platforms. And the Feds and the industry now know Microsoft’s method of operation—bundling, tying, all-or-none licensing. The company cannot use these tactics to enter new markets or it will get slapped down. Its stock price no longer doubles every year, making it hard to recruit or retain smart people. It is Microstuck.
Comments