From rgb@phy.duke.edu Fri May 24 12:49:33 2002 Date: Fri, 24 May 2002 11:35:08 -0400 (EDT) From: Robert G. Brown To: Karen Shaeffer Cc: Robert G. Brown Subject: Re: [off-topic] Re: MS attacking government use of "open source"] > Those of you with an interest might find this article worth a read. > > http://www.neuralscape.com/bmodels/disrupt.html > > I wrote that a year ago, and all that has happened since appears to be > confirming my logic. The next few years are going to be real interesting. Lovely article. Two minor corrections (my contribution to your "open source" effort:-) -- reversible, not reverseable, is the correct spelling. Also the two next-to-last paragraphs both begin with "In contrast" -- you only mean for the second one to. I think. There was another sentence where an inserted "that" in front of a "the" would have improved it a bit, but it is acceptable as is. Regarding your analysis -- in general I agree, and have also been saying and writing this for several years now. In a sense, I think that Microsoft would have been lucky to have been broken up in to baby-microsofts as components of it would have held their value. I view Microsoft's current efforts to influence public policy and law to entrench themselves as the last flailings for a grip before pure economics starts to gut their profit margin. To quote Leia in Star Wars, "the more (they) squeeze, the more ... will slip through their fingers". Their latest lease-only business model will only serve to accelerate the process. In a way, it is a storybook tale of greed and corruption and betrayal among some of the most powerful corporations in the world. IBM singlehandedly "created" Microsoft by using their operating system (instead of, say, CP/M) for the IBM PC. Had they gone with CP/M, Bill Gates would literally be just another dropout hacking around in the tech industry. For many years, Microsoft itself was the "good guys", the linux of its day, empowering new software development just as you describe, where the "bad guys" were the mainframe/cobol/fortran crowd, who were solidly entrenched. "Anybody" could afford a PC, anybody could learn to program in basica, and there was a tremendous gold-rush feeling -- a platform with infinite potential and very few applications, easy to program for. It produced a rush of innovation -- spreadsheets, word processors, terminal programs, and eventually unbelievably cheap ($49!) integrated development compiler/editor/debugger programs, which fueled a >>further<< burst of games and inexpensive applications. The main competitor to the PC was the Apple line. Apple "lost" the war simply because Apple never figured out whether it was a software company or a hardware company. It thought hardware, and was (alas) mistaken. It was, if you like, a much more closed environment than the PC, so that even though it had (and still has) its passionate advocates based on the usability of its SOFTWARE interface, the hardware was overpriced and non-commodity and it never attracted the developer base or intense, price lowering hardware competition of the PC. Microsoft finally started to compete with Apple on the GUI front with Windows 3.1, and was about to attempt real multitasking and virtual memory. Then came the dark betrayal -- IBM and Microsoft, who had basically ridden on each other's coattails to great wealth and market dominance, had a falling out over OS/2 vs WinNT, with Microsoft "betraying" IBM and dooming OS/S (arguably a much better product than NT ever was) to oblivion. In the meantime, a half dozen companies had the opportunity to squash both Microsoft, most notably Sun. Sun had already built an Intel-based Sun workstation -- the 386i (I owned one:-). Alas, they sold it for absurd amounts of money compared to the base cost of the components because Sun, too, viewed itself as both a hardware and a software company -- making the same mistake as Apple but competing in a market where EVERYBODY made that mistake. Just as their native Intel workstation efforts fizzled and they refocused on their SPARC line, they purchased a Unix variant that would run on PC's just about the time that PC's were approaching workstation capability -- the 80486 integrated the formerly separate 8086 and 8087 instructions, had a flat memory address mode (although it also preserved a segmented architecture so that it could easily run the immense base of legacy DOS applications). Then there was the Pentium, waiting in the wings. Had Sun turned around and repackaged Unix for Intel in a way that would have run on modestly equipped 486's (easily enough done) and sold it for $50, Linux would basically never have happened, and both WinNT and OS/2 would also never have happened (or would have come out drastically cheaper and aggressively driven by competition) and the world would be a totally different place today. Once again, Sun was "the" platform of choice of the same group of programmers that today write for linux -- there was a whole core of Unix-based open source software that was developed primarily on Suns (although it was often ported to e.g. Ultrix, Irix, etc. as well) -- all Sun provided was the unix core and compilers, and universities generally built everything else. If they'd provided cheap multitasking multiuser VM unix for PC's, complete with compilers and a functional GUI, in the first half of the 90's, there would have been a rush of developers as everybody hated the fact that Win3.x (and later Win95 and Win98 and WinME) had unbelievably poor multitasking and virtual memory, let alone anything like a security layer or networking. But they didn't. IBM kept trying to develop OS/2 and found a limited market, but Microsoft basically slaughtered them with NT which, although it sucked compared to Unix and was expensive, was much cheaper than any flavor of Unix, ran on OTC PC's, and was sort-of-compatible with companies' large Win9X base and indeed ran some key applications (the burgeoning Office, in the process of destroying WP and Lotus and the slew of OTC office suites and tools). IBM's anger over this literally smouldered for five or six years. It's not NICE to bite the hand the lifted you up out of the dust and fed you for years and ultimately made you the world's wealthiest single human. In the meantime, into the vacuum conveniently left by the shortsightedness of Sun and the various other Unices came Linus Torvalds and his toy, accompanied on the side by the maturation of a fifteen year old effort to develop an unencumbered BSD-style unix. The linux kernel ran fine on Intel hardware -- it was developed for it -- and my first impression of Linux back in 1994 or 1995 when I first saw it was that it was an almost perfect functional clone of SunOS, which was (at the time) the world's best operating system and running on all of my Sparc boxes (and I had a lot of them:-). It's GUI support kind of sucked -- Free X then was still a bit of a disaster -- but in console mode or running simple twm/xterm/xclock GUI's it worked marvelously and even then was pretty darn stable if you had the right hardware. The next five years saw linux literally explode. In the very late 90's, just as the Mainframe Wars were winding down (yes, they were STILL being fought -- indeed, there may still be a skirmish or two being fought in some organizations) IBM finally started to see linux as the vehicle that OS/2 had never become. It was rock-stable and fully featured, had a vast and rapidly growing user and developer base in Universities and small companies, and several companies were actually making a go of "reselling" the absolutely free software. IBM, OTOH, was faced with competition from NT that promised to permanently destroy their entire small business marketplace within a very few years. It was war. IBM embraced linux, put away all its old/legacy operating systems and started porting, planning to destroy Microsoft (while at the same time retaking a lot of its business back -- remember, IBM's business has ALWAYS really been support, not hardware or software). A number of other Unix-based vendors, reading the writing on the wall, jumped on the band wagon, although a few (like Sun) continue to hedge and rely on a dwindling base of users dependent on their high-end platforms and still-unported tools. By this time, nearly all the hardware manufacturers from Intel on down intensely disliked Microsoft because of its rapacious business practices, and a remarkable number of the most innovative developers recognized that developing a killer app to run on a Microsoft-based platform was just asking for Microsoft to come in and take away 70% of your business the moment the market was proven. Perhaps the most famous case of this, although far from the first, is the Explorer-Netscape collision. The Web had emerged from the truly innovative UNIX (TCP/IP) side of computing as the next real "killer-app space". The Internet had come of age. Microsoft wasted no time at all in moving into the space. Unfortunately, the Internet was entirely based on open protocols, which was inconsistent with Microsoft's operational plan (which relies on locking a customer in with a proprietary format for all their critical data). It had to take over the Internet as it had taken over, in turn, spreadsheets, word processors, development suites, various games, dozens of small tools, educational software, and more, all of which were originally developed by others. Microsoft had discovered a remarkable fact. If you own the operating system and the compilers, and have a huge shop employing some of the best software engineers in the business, and "automatically" get basically all the hardware in the world for free to test on, then >>every<< developer who built for Microsoft's OS was really developing for Microsoft. After all, give any decent programmer a fifteen minute look at an application (long enough to get a decent idea of its purpose and underlying data structures) and, given a bit of domain-specific support as required (which Microsoft could ALWAYS afford to hire on a per-project basis) they could hack together a clone of the application to compete with it, generally in six months to a year. Even better, as the owners of the code base and sole possessors of the source code, they could ALWAYS integrate their products successfully with the OS and its underlying toolset. Their products were more likely, in some cases MUCH more likely, to work well and reliably than the very products they were cloning. Who knows whether or not they ever actually jerked the operating system around for the PURPOSE of quietly killing off a competitor? They didn't really have to -- the changes introduced in an upgrade of completely benign intent are often sufficient to break a complex application even today in the open source world where EVERYBODY has access to the source and has a clear description of the changes. In the open source world, most applications get fixed up in a matter of weeks to months, dictating the cycle of "distribution upgrades" from e.g Red Hat and and Mandrake and others -- collections of applications built on top of a consistent snapshot of the kernel and library base. In the Microsoft world, only Microsoft developers had complete and unrestrained access both to the new operating system but to its source and its developers. For them, the weeks or months were over before the actual release of the OS and the few remaining bugs could be quickly squashed when the program was in mass use. For their competitors, it might take six months to a year, with customers growing increasingly frustrated. This produced an enormous economic force in favor of Microsoft's NON-operating system software offerings, which in turn produced enormous revenue. Microsoft's nominal "value" swelled to a truly absurd price to earnings ratio (one that persists today) as it was and continues to be viewed as a growth stock. For the first time, however, someone struck back. Faced with immediate extinction following its heady ascent into wealth, Netscape (founded by some of the INVENTORS of "the web") turned back to its open source/Unix roots on the one side ("giving" its product away to compete with Microsoft, which was doing the same as a loss-leader intended to drive Netscape out of business altogether so it alone could dictate the future of the web) and it >>sued<< Microsoft, joined by various states and the federal government. Did this really matter? I personally don't think so, but it may have. By delaying Microsoft, the Internet has had time to stabilize on open standards, and although Microsoft plans to make another go at taking it all with .net, it will now encounter substantial resistance (which it is, as usual, trying to overcome by "giving away" tools that it will undoubtedly sell for large sums as soon as all the competitors are destroyed). In my opinion, Microsoft has met its match anyway, regardless of what it manages with .net and the antitrust suit. Your analysis is basically right on the money -- it's about MONEY. A paradigm shift, with a significant economic advantage on one side, has occurred and all that is left is waiting for the war to be over as pure economics rewards those that use Linux and punishes those that use Microsoft. This will take a while -- the Mainframe war lasted for nearly twenty years, although some of that was due to legacy hardware, which is no longer an issue in the commodity-driven world of today. It was also extended by legacy applications with extremely long "useful" lifetimes -- all that cobol code base performing mission critical tasks. Microsoft completely lacks the first protection, and lacks a lot of the second. Yes, there is a large Microsoft application base, but it has to be reengineered every 2-3 years ANYWAY to keep up with Microsoft's own bait-and-switch evolution of the core operating system. Furthermore, what Microsoft discovered to be true for its own code base is even more true for the developers of the Linux code base -- any developer is potentially a linux developer in the sense that a new idea can rapidly be cloned and implemented. Here Linux enjoys a huge advantage over Microsoft. ALL developers have access to ALL the source. Development tools are all free. The very kids in garages that gave birth to a lot of the killer applications running on Microsoft today cannot afford Microsoft code development tools, which might well cost more than their hardware, but they ALL can afford linux. XFree86 is not without its clunkiness (pixel based GUI geometry being evil, evil, evil) but it is still stable, well supported on nearly all hardware, and (with Gtk and MESA and SDK and various other libraries and toolsets) very easy to program and getting easier all the time. I WRITE code for linux, and single-handedly I can engineer a project from start to finish with a GUI fronting a complex application in six months of part time work. With Windows I'd still be figuring out the code development environment and how Windows handles VM and dynamic libraries, and I'd have shelled out literally thousands of dollars for software to match the linux-based development shop I have in my home today. Linux, being free and visibly more powerful and more stable, simply cannot lose against Microsoft. Every person Microsoft "loses" to Linux will never go back. Every company that tries Linux will rapidly learn that it is adequate to meet their needs and much more scalable and cheaper and THEY will never go back. Now linux is being pushed by IBM, and we're starting to see actual TV advertisements touting the incredible cost-savings associated with the conversion. Only the legacy applications and Microsoft's illegal entrenchment in PC stores stand in Linux's way. The most common reasons I hear for people to choose Microsoft over Linux are: a) (from the vendors) Microsoft has us locked in so that we cannot afford to offer a linux installation. b) (from the corporate types) We need to run Office. c) (from the kids) We want to play Diablo II (or some other game). The first one really needs attention from the attorney general as it is clearly anti-competition. Microsoft should not be allowed to reprice its products on the basis of exclusivity. In the long run it won't matter -- Linux is already trivial to install over the network; as bandwidth to the household and corporation continues to increase, reinstalling with Linux will be quick, easy, and free. The second and third ones are still an issue. Many office suites exist for linux, and they do a "decent" job of importing office documents, but few of them really do a good job at writing/exporting office documents and all of them have feature creep. Office is likely to be to Linux vs Microsoft war as the cobol code base was to the Mainframe vs PC war. It will be long fought and won a system at a time, but as long as Office costs $200 or more on top of the $100 OS vs a half dozen office suites for free on top of a free OS, there will be a steady bleed of systems over to Linux side. Tools like DOSEMU, Wine, Win4Lin, VMware also provide an interpolation point where one can run Linux but still run Windows applications. Games are similarly an important sticking point. I experience considerable pressure from the kids to run Windows on some boxes at home just so they can play games. Unfortunately, very few game CD's sold in stores contain both linux and winxx versions of the game. When this changes (especially if and when it inverts, and a killer game is released only for LINUX), the war will be all but over. A corporate type might be able to resist converting from Office and argue that it is better for him to spend $150 a year on average to keep using it than it is to spend a half a day learning to use one of the free alternatives, but he will not be able to resist his own children. What he or she use at home, they'll use at the office. My own personal prediction is that we'll see a significant impact of Linux on Microsoft's entire product line before the end of the year. Microsoft's products are legendary for being bleeding wounds from the security point of view, and viruses are a constant source of corporate irritation. Linux is all but immune to viruses in general and is far more fundamentally secure in the first place. Linux is cheaper by far and is far more scalable. Linux has full coverage of all critical applications from the desktop through massive client/server deployment -- any Linux installation is more powerful than Windows 2000 server with unlimited client licenses, the full suite of compilers, and much more. The last few major distributions have even seen windows emulation come of age to the point where it sometimes works as well or better than Windows itself, and for the first time one can run at least some very modern games under emulation within linux. Once the shrinkage starts, only federal protection of one sort or another can save Microsoft. Its inflated P/E ratio is extremely vulnerable and has plunged once in the recent past (and has still not recovered anything like its peak value). Its net revenue is really not all that large. If it EVER stops being perceived of as a growth stock, its P/E could drop to something optimistic but sane or even something pessimistic, and the confidence of the market in Microsoft would be shattered. Of course Microsoft is vulnerable. Look at the once-mighty Digital Equipment, I mean Compaq, oops, I meant Hewlett-Packard. Where is Lotus? What has Borland or Corel released lately (for Windows, that is)? Even once-invulnerable companies like Oracle and Sybase are learning that once Microsoft builds a database to compete, well, they wrote the operating system and have all the sales channels and can vertically integrate their own software in any market. All developers are Microsoft developers. They just THINK they're in business for themselves. The computer market is fickle and short memoried and driven by cost-benefit. Every improvement in Linux is irreversible. Every new linux desktop is also irreversible. Every corporate conversion is irreversible. Microsoft has no way to go but down, as it cannot match the profit margins charged by Linux vendors (and retain all but a few percent of its current revenue), and it cannot maintain any sort of real edge in added value. Right now its efforts to CONVINCE the government and increasingly cynical users of its added value ring hollow. A year ago I gave Microsoft two years to lose half its value. I think I was pessimistic. A year from now, I think it will be in free fall. rgb -- Robert G. Brown http://www.phy.duke.edu/~rgb/ Duke University Dept. of Physics, Box 90305 Durham, N.C. 27708-0305 Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu