OS / 2 a quarter of a century later: why IBM lost, and Microsoft won
Twenty-five years ago, IBM presented a master plan for regaining control of the PC market. In November 1987, the first diskettes with OS / 2 version 1.0 went to the stores. Microsoft developed it together with IBM and in the case of successful implementation of these plans, the world would have been completely different. And the world has already begun to change.
Now OS / 2 is usually remembered only in connection with the role that it played in the industrial war, which ended with the triumphal victory of Microsoft. At the time of release of OS / 2, Microsoft employed 1,800 people, less than it currently works in the Liverpool TV shop QVC. Microsoft was not even the largest maker of PC software. But just a few years later, the company became not only the largest player in the industry, but also one of the most expensive companies in the world; just one rumor about Microsoft entering a new niche caused panic among existing players.
According to the traditions of our publication, I brought the old programs back to life and appreciated OS / 2 from a modern point of view. But it is much more interesting to ask another question: could IBM have won? If it was OS / 2 that caused the Blue Giant to be defeated, could they stop Microsoft with more cunning plans? Let's go back to the times when the world did not yet know the words “platform” and “ecosystem” and recall what the information technology industry looked like in the mid-1980s.
')
Primary Soup: The World in 1987
It was a divided universe. Serious money was made on the old hardware, vertically integrated systems. IBM was the mistress of this universe, having just settled an antitrust case that lasted ten years. In 1985, it employed 405,000 people - 12 Google.
IBM's closest competitor, Digital Equipment , which owned half the minicomputer market, employed 100,000 people. The Information Industry Center was located in the Boston and New York area, reflecting the impact of Cold War government orders and MIT , the largest technical university in the world, located in Boston, Massachusetts. The just begun penetration of microprocessors on the market led to a stormy discussion of " open systems ". Over the next years, “open systems” became the main marketing slogan, almost equaling in popularity with the current “clouds”. Open systems meant Unix, and Unix meant Sun Microsystems - which became the engine of competition in the new network of workstations.
That was the business market. Games and cool multimedia lived on powerful 16-bit microcomputers, including niches for audio and (later) video processing of the Commodore Amiga and Atari ST line . Against the background of their success, Apple Macintosh looked like an overpriced toy. In fact, the Mac was just another microprocessor based on the Motorola 68000 family of microprocessors — very well designed — but technically inferior to the much more complex Amiga and too strange for business. For a while, it seemed that Mac would sink Apple, but the Postscript language, the Apple LaserWriter printer and PageMaker created a computer layout and gave Apple an unshakable support in publishing.
There was also an IBM PC. Rough, little-featured, unfriendly: PC managed to collect in itself the worst of all worlds. As a business computer, it was significantly inferior to Sun workstations. And what idiot could have thought of buying a PC home? It was pointless to use it as a multimedia machine, there was not even a sound card, only a primitive squeak was available to it.
But the PC had its advantages. Offices and offices of large companies could buy their own PCs or its clones and much faster, easier and cheaper to start working with databases or spreadsheets. The alternative was bureaucratic procedures for requesting working hours on the mainframe and writing or modifying relevant programs. The PC had a wide selection of varied hardware, as numerous companies sold clones or “compatible” machines and a thriving dealer, training and support market. PC meant Novell for networks, SCO Unix (or a dozen forgotten alternatives) for vertical integration, and for everything else - the largest companies in this market, Lotus Corporation and Ashton Tate . Ubiquitous PC business applications — dBase , Lotus 1-2-3, WordStar, and WordPerfect — had their own “ecosystems” made up of trained users, macros, extensions, and even compatible serious scheduling and reporting programs.
You probably noticed that something is missing. Nobody knew the very idea of ​​an “API” or an ecosystem — the multilateral market — because at that time, the “platform” on the PC meant the application itself. DOS did not provide any API, only allowing something to break the interrupt. There was nothing in the DOS goal that could help a novice developer. If you were developing a new application for DOS, you had to support the entire periphery yourself. If the user had an HP LaserJet printer or a fancy Hercules video card, you had to write drivers for them yourself. In 1989, WordPerfect dismissed WordStar from the place of the most popular text editor for the PC, thanks in large part to its excellent drivers.
DOS was not a “operating system”, but a commonplace bootloader. DOS had a primitive 8 + 3 file system. There was no API - the applications did everything they wanted and were capable of. And he was single-tasking, for multitasking it was necessary to install hacks from third-party companies.
Thus, by the mid-1980s, IBM faced two challenges. The first is to repel the onslaught of open systems, strengthen the reputation of the PC as a business machine, and return it under the parent wing. The second - to bring the PC to the modern level, which meant multitasking, network support and a graphical interface - users of other systems already enjoyed all of this. IBM planned to achieve this in three ways - a new software platform, a new iron architecture and a new strategy. IBM made the PS / 2 system bus closed, hoping to sell licenses to manufacturers of clones and expansion cards. But no one wanted to buy them. Instead of making the rest of the world incompatible with IBM and therefore unattractive to customers, PS / 2 made IBM computers incompatible with the rest of the world. PS / 2 sold only IBM and eventually the company gave up.
The new strategy was a grandiose, all-embracing, marketing nonsense called Systems Application Architecture and a very IBM-based answer to open systems. SAA promised a huge set of standards that combine IBM’s developer interfaces, user interfaces, and communications protocols. But IBM has long ceased to create reasonable standards. For example, IBM systems did not use the industry standard for text encoding, ASCII - they used their own, called EBCDIC - which existed in at least six different versions. SAA did not make IBM products "open systems".
New PC operating system
The third focus of IBM’s efforts was a new operating system for the PC — OS / 2. It was a new challenge for the company. For many years, programs were just a part of a “computer” bought by a client; selling software separately from IBM computers was only forced by a 1968 Ministry of Justice decision. Most of the software continued to be written for a particular hardware and was tightly tied to it. IBM has never before sold software in an open and mature market, where users make the choice themselves. These were uncharted waters in which extremely aggressive creatures hid.
With its multitasking, polished multithreading, inter-process utilities like channels and queues (IPC goodies like pipes and queues), the graphical interface and the new OS / 2 file system, it was as good as its competitors. The kernel was very well designed and able to simultaneously run almost any application. The OS / 2 API also traced forward-thinking system planning with long-term use.
IBM led the development of OS / 2 together with the tiny company Microsoft, and despite many disagreements, they somehow reached the joint release of the system.
OS / 2 was a truly new “platform”. But it was unusual for users and at the time of launch did not have any applications. OS / 2 needed what we now call an “ecosystem.” And what to do in this case?
What is a platform?
IBM at that time did not have any idea about the idea of ​​a hardware-independent platform. In the literal sense of the word, she did not know what she did not know. But how could she know that? Customers usually bought what IBM advised them to ask, questions only arose regarding the timing of the purchase. OS / 2 required a market for developers and support from iron manufacturers so that OS / 2 could work on a wide range of video cards, printers and other peripherals. It was a classic chicken and egg problem. Why allocate funds to conquer the market, which is not yet?
With all its technical correctness, OS / 2 offered very few benefits, and its implementation was a major headache for corporate support services.
Multitasking comes to PC: OS / 2 version 1.0
The size and prestige of IBM forced several major developers to make versions of their OS / 2 DOS programs, Ashton Tate, for example, transferred dBase, but moving applications did not mean moving the ecosystem. Driver development was hellishly difficult, after three years it was extremely difficult to type something from OS / 2. Nowadays, platform manufacturers understand that they need to support, persuade, and even bribe application developers and drivers. Microsoft knew this, IBM did not.
OS / 2 backward compatibility was also not impressive. Originally designed by 16-bit, the new system did not use the new features of 32-bit Intel 386 processors to launch DOS virtual machines. The “DOS Compatibility Tool” on 16-bit OS / 2 was slow and unreliable. Without attractive applications and iron support, the only advantage of OS / 2 remained the ability to use more than 640 kilobytes of memory in applications. But in the summer of 1988 there was a jump in the price of memory, which made the 2 megabytes necessary for OS / 2 extremely expensive. The graphical user interface was even more demanding, Presentation Manager made most of the existing PCs too weak for OS / 2.
The third party patches for DOS, adding multitasking and expanding the available memory, have become the most reasonable way out for users.
The first version of OS / 2 entered the market in 1987, with multitasking and a text interface. The GUI version came out only a year later. It still did not have a modern HFS file system, with support for partitions and large files.
OS / 2 failed
Different workstation ideas
In the summer of 1988, one project changed the relationship between partners. DOS could not use more than 640 kilobytes of RAM, which gave rise to a thriving market of memory expanders, which with the help of various tricks sought allocation of a larger volume. One of the undoubted advantages of OS / 2 was the mechanism of working with memory, which does not impose any restrictions of up to 15 megabytes. But this summer, one of the Redmond trainees found a way to run DOS in Intel 286 protected mode, giving it more memory.
The head of the trainee showed his work to Steve Ballmer and the Microsoft management thought about the possibility of using the Windows shell as a memory extender for DOS. A PC could load DOS, start Windows, and gradually replace DOS components at a lower level, similar to replacing wheels on a moving car without stopping it. The intricate way. But Windows was full of similar hacks - in the technical sense of the word it was self-modifying code.
The “shell plus memory expander” did not have the rich OS / 2 APIs — and the system would never be stable. It did not even have long file names. DOS used an archaic 8 + 3 system (“LTTR2BOB.WPD”) a decade ago. It was not an operating system at all. But it had a graphical interface and with the help of smoke, mirrors and a fair amount of faith from users, it could be presented as a new operating system.
With a light stretch, Windows could be advertised as a new “platform.” The new platform meant that all existing applications continue to work and soon new, radiant and beautiful will join them. At that time, no graphical shell over DOS had a significant advantage. The probability of victory was very high.
Microsoft had some very serious advantages. First of all, she had a very good position for the development of the “ecosystem”, there were development tools, she could sell them very cheaply and knew how to advertise Windows as a “new era”, something with ever-increasing acceleration. Microsoft didn’t have conservative units to pacify.
Secondly, she had a very good relationship with PC makers. They paid for MS-DOS regardless of whether they installed it on a PC. Buyers needed something that could run dBase and Lotus 1-2-3. Microsoft could easily add Windows to all of these computers, and the first run of Windows was easy and painless.
Thirdly, the press and experts were very keen on the success of Microsoft and the failure of IBM. Not because they loved Microsoft very much. But the clones created a thriving market and no one wanted IBM to once again control both hardware and software. Nobody wanted to return to the world under the rule of one huge, vertically integrated giant. This was not obvious in the summer of 1990, three years after the announcement of OS / 2, when Windows 3.0, the first version with support for protected mode, was released. Almost all of the press considered Windows 3.0 another rival of the GEM shell and the Quarterdeck memory expander.
Divorce
For many months, before the release of Windows 3.0 in the summer of 1990, Microsoft distributed copies of it that all people in the industry seemed to have. It was clear that Microsoft sees in Windows something more than a flower on the grave of MS-DOS and a farewell to pre-OS / 2 world, which remains somewhere behind. Nothing like this. Microsoft has done a lot of work to increase the attractiveness of Windows and over-actively advertised it among developers. It was a “pseudo-platform”, not actually a new OS, but allowing developers to use a graphical interface. Fed up with the meager sales of products for OS / 2 developers reacted very positively. At the same time, Microsoft was obliged to continue demonstrating the 32-bit OS / 2 to the press, and there was all that was lacking in the 16-bit version.
The pseudo-mutant platform Microsoft Windows 3.0 was launched with a lot of noise and began to sell well. It became clear that during 1989 at least some of the developers who were studying OS / 2 Presentation Manager actually turned towards Windows. Windows gave them a lot of opportunities and turned into a chance to get the market, even if it was modest and short-lived. But at least something was better than nothing. Soon, more and more applications began to appear under Windows, after which they began to flow, as major application vendors plugged holes in their portfolio. In just one month, Microsoft sold more copies of Windows than IBM copies of OS / 2 in two years.
During 1988 and 1989, IBM tried several times to convince Microsoft to shut down Windows, but Bill Gates refused. After that, the head of the PC division at IBM, Jim Cannavino (Jim Cannavino), made an unintentional but catastrophic mistake. Aware of Microsoft’s growing enthusiasm for Windows, he couldn’t recognize that they were right and give the system any chance of breaking the agreement with Microsoft. As a result, Windows received a cool endorsement - the opportunity to be born and the space for growth. Despite the complete incompatibility of the Windows API, it was positioned as a preliminary version of OS / 2. IBM had many opportunities to get rid of Microsoft and protect OS / 2, but this error was the most destructive.
This was in September 1989. IBM’s most terrible mistake was maintaining partnerships with Microsoft and refusing to suppress the development of Windows. The relationship between them quickly deteriorated, as Windows was sold in huge quantities.
The ecosystem benefits Ashton Tate and Lotus enjoyed quickly began to evaporate. Windows provided drivers for devices. New skills boiled down to the ability to handle Windows. Knowing the intricate sequences of nonstandard hotkeys was no longer an advantage. WordPerfect developers desperately tried to hold positions by transferring hot keys from DOS to the first version under Windows, but did not succeed.
Marsh on industrial systems
In the business field, Windows PCs have begun to use clumsy end-user applications with forms. In May 1991, Microsoft launched Visual Basic, a GUI-based scripting language that is primitive but capable of calling routines from compiled libraries. Windows also gave a huge advantage to the Microsoft application development divisions. In the DOS world, the market share of Microsoft Multiplan and Word was small. But the positions of their Windows heirs, Excel and Word 2.0, were much stronger. In addition, Microsoft began selling a bundle of its office applications with an Access database and an e-mail program that is noticeably cheaper than its competitors. The entire “Office for Wndows” cost less than one dBase or Lotus 1-2-3.
Two years IBM and Microsoft waged a PR war of attrition. IBM could not kill Windows and had to get rid of Microsoft.The inevitable divorce occurred in April 1991. Under the final agreement, IBM independently took up the development of a 32-bit version of OS / 2 and obtained a license to include the Windows code. Microsoft immediately stopped developing 16-bit OS / 2, although for several years it continued to sell the Microsoft OS / 2 LAN Manager, all reducing the font size of the word OS / 2 on the box.
Paul Carroll, in his book Big Blues: The Unmaking of IBM, notes that at the time of public divorce, Microsoft sold 13 million Windows 3.0 during the year, while IBM sold only 600,000 OS / 2 copies in three years - of which only 300,000 were actual sales, much less was actually used. In the wake of Windows sales and demand for Microsoft Windows-based applications, the company grew from 1,800 employees to 10,000.
But in spite of all this, power was not on her side. As Gates himself later wrote
And the worst thing was that in the 32-bit version of OS / 2 there was everything that 16-bit lacked. It was a very impressive system, it was just not completed yet.
The Empire Strikes Back
A new, 32-bit version of OS / 2 fully utilized the capabilities of the Intel 386 processor: a flat memory model and the ability to virtualize DOS sessions while maintaining the connections between their processes. The 32-bit version also supported unreliable DOS memory extenders, giving DOS applications fast 32-bit memory access so that they themselves did not notice this and running DOS applications running in fall-protected virtual sessions even better than earlier. IBM and Microsoft were equally proud of this. IBM took Microsoft Flight Simulator, one of the most demanding graphics applications for DOS - and ran one copy after another, until a whole dozen started to work at the same time.
Suddenly, OS / 2 began to look much better than its competitors, solving many of the problems of DOS business users with whom they had been tormented for many years in a row. Even Windows programs worked in multiple virtualized sessions faster than in DOS. OS / 2 combined the best features of all worlds: old DOS applications with their trained users and third-party add-ons, a new wave of Windows applications and native OS / 2 applications.
However, the system was delayed. IBM decided to add OS / 2 shine with the radically new Workplace Shell. It was not just a user interface, but a collection of object-oriented class libraries that could be overridden or extended by developers. Workplace Shell has added many Mac features and new features to OS / 2. Even nowadays, she looks very good. Oddly enough, OS / 2 has become increasingly popular among enthusiasts, thanks in large part to the stunning demonstrations of David Barnes, one of the great demo gods in our industry. Here is Barnes in all its splendor in 1993 .
Microsoft needed a beautiful story and they found it. They had a new, multi-user and portable operating system on different processors at an early stage of development. It turned out to be fateful. OS / 2 3.0 or OS / 2 NT has become one of the largest vapourware in the history of the industry. In 1990, it was only able to boot the core, only three years later someone was able to buy it, and only six years later it began to have a noticeable impact on the market. But she radically changed the public opinion.
RISC business and Unix killer
At that time, the ability to work on different processor architectures seemed very important. The common wisdom said that the microprocessor architecture from Intel is exhausted and the future for RISC. The RISC processors from MIPS and Sun (SPARC) dominated the workstation market, followed by DEC (Alpha) and IBM (Power). It was difficult to understand which of them would win in the end.
The new dummy platform also promised something that Unix manufacturers had to do, but could not: a single platform with rich APIs. In the world of Unix reigned the tyranny of small differences. The developers fought for the smallest details that had no meaning for most users. They made complicated plans and plots, merged and diverged, threw out most of the attractive changes (like the NeWS Display Postscript UIfrom Sun) for the sake of fruitless attempts to find a compromise.
Unix was supposed to be united, but IBM joined the all-except-Sun alliance called OSF, and Microsoft’s enterprising marketers were able to take advantage of the resulting vacuum. NT was a dummy, but Microsoft promised help in porting it to MIPS, Alpha, and even IBM / Motorola PowerPC. The company promised that 16-bit OS / 2 and POSIX Unix libraries will work in NT.Microsoft also promised a uniform Win32 API that allows it to run programs on both NT and Windows DOS-based machines. In 1993, the participants of Unix-wars realized the threat and signed the single " Spec 1170 ", but this didn’t bother anyone.
A non-existent platform allowed Microsoft to retain customers. Microsoft offered customers to wait for a 32-bit version of Windows, Chicago, which was supposed to be released in 1993. It was still based on DOS, but offered the user many of the advantages of OS / 2, such as long file names - and had to work on 4 megabytes. They also had to prepare for the Unix killer, Windows NT. NT was supposed to not only kill Unix, but eventually turn into a comprehensive, all-powerful, object-oriented Cairo system .
Promises went beyond the clouds. IBM realized that it had the last chance to return the market and at least part of the multi-billion dollar cost of developing OS / 2. In the fall of 1994, a fairly optimized version of OS / 2 3.0, Warp, was launched, followed by IBM-wide mass television advertising. It was supplied in two versions - one included Windows, the other used Windows already installed on the hard disk. IBM much earlier than Microsoft realized the potential of the Internet - it was the first OS that came with a browser. Microsoft at that time was busy with an ambitious project of a carefully fenced garden, Compuserve Killer, which was supposed to enter Chicago, and did not even include the TCP / IP stack in Windows.
For a short time, OS / 2 captured the imagination of a small part of customers, but was never able to change the situation on the market - largely because of all the major PC makers, only IBM decided to sell computers with OS / 2 pre-installed. IBM support did not know how to work with unprepared users. New users of OS / 2 met the time-consuming installation process, poor support for hardware, a small number of native applications. OS / 2 supported Windows applications so well that in the absence of dangerous competitors among native OS / 2 applications, all serious developers focused on the much more profitable Windows market.
As a result, Chicago came out only in 1995 under the name Windows 95 and did not work on 4 megabytes. NT was too big and slow for the market, it lacked office applications - the promise of portability Win 32 was never fulfilled. NT came out a year later and users demanded the Windows 95 shell. By that time, the system architecture had changed so much that most of the promised features were never implemented. Microsoft imposed restrictions on the number of network connections for workstations (which could be used as departmental servers for this) and moved the graphics drivers to the core, sacrificing reliability for the sake of performance. About Cairo quietly forgotten.
Last spurt to glory
In the late summer of 1996, IBM admitted defeat and transferred most of the programmers to other tasks. She released another version of OS / 2 in 1996, incorporating voice recognition and a more familiar interface. Support for this system has continued for many years. But the war was over - Microsoft won. OS / 2 IBM CEO Lou Gerstner finally convinced the death of native OS / 2 applications in the senselessness of further development. This was partly due to IBM itself. The famous strategy “Workplace OS” implied the launch of personalized versions of the OS on top of the microkernel - but failed to attract customers and was never completed. More than 2000 developers were engaged in a portable version of OS / 2, mainly designed for IBM workstations with RISC processors, but never brought it to a stable state.Apple's joint object-oriented OS projectTaligent ended in nothing. IBM could not buy Apple. Microsoft Win32's tenacious strategy has conquered the market.
WPS - WTF?!?
Workplace Shell first appeared in OS / 2 version 2.0, released in April 1992 — simultaneously with Windows 3.1. There is no task pane, no program directory, no system menu, no shutdown menu ... all applications are located in the “System” directory. Minimized applications go to the folder. OS / 2 had a radically new interface that lacked many of the elements of modern UI. Where did he come from?
He was born from IBM’s plans to create an office of the future - literally object-oriented functionality was invented in research laboratories - this is in the print ads of 1990 that preceded Workplace Shell.
AboveOfficeVision employed 1,500 programmers, but the project was never completed. To keep the promise made to customers about developing a collaborative environment, IBM bought Lotus in 1996.
Such office automation tools were difficult to sell, but IBM emphasized their flexibility. They were so flexible that you could change the font and background image of each window and each window element individually. Workplace Shell had more useful features. Like Mac, shortcuts (links) to files tracked their movement. Template directories could contain applications and quickly be cloned. Closing the directory minimized all applications and documents that were associated with it. There were other oddities. By default, the icons could not be moved with the mouse, although this quickly changed in the settings.
The latest version of OS / 2 included much more familiar elements. The system menu gave quick access to each file, and Mac icon designer Susan Kare, who also made icons for Windows 3.0, worked on the appearance of the icons.
OS / 2 Warp 3.0 in the fall of 1994. Under pressure from customers, a hurriedly written taskbar was added to OS / 2.
Last battle: OS / 2 Warp 4.0 in 1996.
For some reason, IBM made deafening sounds for each mouse action (for example, dragging).
Why did IBM lose?
Around the rise of Microsoft to the top of the PC industry, passions are still raging. If in 1987 you had said that Intel and Microsoft would determine the future of the industry for 10 years, they would simply laugh at you. Intel could not create a competitive RISC chip, and Microsoft never wrote any operating systems. But the questions "Why did Microsoft win?" And "Why did IBM lose?" Have completely different answers.
IBM has never before faced with so enterprising and ruthless competitors like Bill Gates. Gates fought desperately and IBM could not get rid of it. But regardless of the presence of OS / 2, IBM could replay it or even trite it with the masses if it had developed a coherent strategy that gives customers good reasons to spend money on IBM.
IBM could have realized that in 1988, OS / 2 was too revolutionary for customers. She could take control of DOS and offer clients a smoother transition to the world of graphical interfaces. In fact, IBM tried to do just that with a project called PM for DOS - which was never completed. It was too late. IBM could buy Apple with the most attractive user interface and put it on a solid technological foundation. As an even more revolutionary solution, IBM could recognize the real value of software and move the competition to a higher level — by creating a free platform like Linux.
Many alternative history options can be offered - but most of them will remain a game of imagination. After Microsoft gained strength and captured the minds of the developers of the OS / 2 single platform, it would be very difficult to get them back. More interesting is another question - why did Microsoft win?
First of all, Microsoft ruthlessly followed the requirements of the market and cared only about the number of copies of Windows sold, it was not concerned about the attitude of others and the theoretical correctness of the actions taken. The market did not like Microsoft, but Microsoft brought money to the market. Software makers, hardware, peripherals - they all earned more with Microsoft. The programs infuriated you every day, but the “ecosystem” as a whole looked much healthier. IBM had no idea how to build new ecosystems. The high authorities believed that if they made the system and it would be good (and it eventually became so), then there will be buyers - simply because of the magical power of the three letters IBM. The IBM sales team has mastered many dark methods, including the use of other IBM products and services,but they did not help much in open knife fights with Bill Gates.
Microsoft was completely focused on one task. She did not make compromises in order to appease the units occupied by other products. Gates was not going to, as it seemed to many then, to limit himself to writing applications and tools for foreign platforms. He wanted his platform. Microsoft did a great job on Windows. Needless to say, Microsoft’s huge advantage was the extremely rigid terms of licensing agreements. PC makers had to pay Microsoft for DOS regardless of whether they installed it on computers or not - so why should they install alternatives? In the end, Microsoft lost the antitrust lawsuit and was forced to give in to the requirements of Novell, Be and Sun. But do not forget that two are needed for tango. Partly to blame and short-sighted PC makers.
OEMs did not protest against Microsoft’s similar policies and later it hit them themselves. After the death of competitors, Microsoft raised prices - this is one of the advantages of a monopoly position. As the price of computers fell, the larger part of the price was leaving Microsoft. Sometime high OEM profits plummeted. None of them could stand out from the crowd, Microsoft did not allow it - she insisted on a uniform "Windows experience". Sometimes it is more profitable for a business to sacrifice current profits in order to maintain competition and preserve future profits. Many are actively engaged in this - but not the PC industry.
In the OS proceedings following the end of the war, many OS / 2 fans blamed the computer press for total hostility to the system. In many ways, they were right - journalists who used and praised OS / 2 can be counted on the fingers of one hand. One of them was John Lettice, the founder of The Register - the other your humble servant. The beautiful OS / 2 2.0 souvenir from the illustration above is still in our office. But the press simply reflected the then generally accepted opinions and economic decisions. She also did not want a world in which IBM defines open system standards.
Everyone was so busy fighting the outgoing war that they did not notice the approach of a new one. In 1995, when the OS war was still raging on Usenet (although it was clear to everyone who won), Guy Kewney summed up his annoyance:
More on OS / 2 - the memories of a person who at that time worked at IBM (if they are interesting, they can also be translated)
IBM insider: How I caught my wife while bug-hunting on OS / 2
Where were the bullet holes on OS / 2's corpse? Its head ... or foot?
An alternative overview of the history of OS / 2 on Habré with detailed descriptions of technical issues
Now OS / 2 is usually remembered only in connection with the role that it played in the industrial war, which ended with the triumphal victory of Microsoft. At the time of release of OS / 2, Microsoft employed 1,800 people, less than it currently works in the Liverpool TV shop QVC. Microsoft was not even the largest maker of PC software. But just a few years later, the company became not only the largest player in the industry, but also one of the most expensive companies in the world; just one rumor about Microsoft entering a new niche caused panic among existing players.
According to the traditions of our publication, I brought the old programs back to life and appreciated OS / 2 from a modern point of view. But it is much more interesting to ask another question: could IBM have won? If it was OS / 2 that caused the Blue Giant to be defeated, could they stop Microsoft with more cunning plans? Let's go back to the times when the world did not yet know the words “platform” and “ecosystem” and recall what the information technology industry looked like in the mid-1980s.
')
Primary Soup: The World in 1987
It was a divided universe. Serious money was made on the old hardware, vertically integrated systems. IBM was the mistress of this universe, having just settled an antitrust case that lasted ten years. In 1985, it employed 405,000 people - 12 Google.
IBM's closest competitor, Digital Equipment , which owned half the minicomputer market, employed 100,000 people. The Information Industry Center was located in the Boston and New York area, reflecting the impact of Cold War government orders and MIT , the largest technical university in the world, located in Boston, Massachusetts. The just begun penetration of microprocessors on the market led to a stormy discussion of " open systems ". Over the next years, “open systems” became the main marketing slogan, almost equaling in popularity with the current “clouds”. Open systems meant Unix, and Unix meant Sun Microsystems - which became the engine of competition in the new network of workstations.
That was the business market. Games and cool multimedia lived on powerful 16-bit microcomputers, including niches for audio and (later) video processing of the Commodore Amiga and Atari ST line . Against the background of their success, Apple Macintosh looked like an overpriced toy. In fact, the Mac was just another microprocessor based on the Motorola 68000 family of microprocessors — very well designed — but technically inferior to the much more complex Amiga and too strange for business. For a while, it seemed that Mac would sink Apple, but the Postscript language, the Apple LaserWriter printer and PageMaker created a computer layout and gave Apple an unshakable support in publishing.
There was also an IBM PC. Rough, little-featured, unfriendly: PC managed to collect in itself the worst of all worlds. As a business computer, it was significantly inferior to Sun workstations. And what idiot could have thought of buying a PC home? It was pointless to use it as a multimedia machine, there was not even a sound card, only a primitive squeak was available to it.
But the PC had its advantages. Offices and offices of large companies could buy their own PCs or its clones and much faster, easier and cheaper to start working with databases or spreadsheets. The alternative was bureaucratic procedures for requesting working hours on the mainframe and writing or modifying relevant programs. The PC had a wide selection of varied hardware, as numerous companies sold clones or “compatible” machines and a thriving dealer, training and support market. PC meant Novell for networks, SCO Unix (or a dozen forgotten alternatives) for vertical integration, and for everything else - the largest companies in this market, Lotus Corporation and Ashton Tate . Ubiquitous PC business applications — dBase , Lotus 1-2-3, WordStar, and WordPerfect — had their own “ecosystems” made up of trained users, macros, extensions, and even compatible serious scheduling and reporting programs.
You probably noticed that something is missing. Nobody knew the very idea of ​​an “API” or an ecosystem — the multilateral market — because at that time, the “platform” on the PC meant the application itself. DOS did not provide any API, only allowing something to break the interrupt. There was nothing in the DOS goal that could help a novice developer. If you were developing a new application for DOS, you had to support the entire periphery yourself. If the user had an HP LaserJet printer or a fancy Hercules video card, you had to write drivers for them yourself. In 1989, WordPerfect dismissed WordStar from the place of the most popular text editor for the PC, thanks in large part to its excellent drivers.
DOS was not a “operating system”, but a commonplace bootloader. DOS had a primitive 8 + 3 file system. There was no API - the applications did everything they wanted and were capable of. And he was single-tasking, for multitasking it was necessary to install hacks from third-party companies.
Thus, by the mid-1980s, IBM faced two challenges. The first is to repel the onslaught of open systems, strengthen the reputation of the PC as a business machine, and return it under the parent wing. The second - to bring the PC to the modern level, which meant multitasking, network support and a graphical interface - users of other systems already enjoyed all of this. IBM planned to achieve this in three ways - a new software platform, a new iron architecture and a new strategy. IBM made the PS / 2 system bus closed, hoping to sell licenses to manufacturers of clones and expansion cards. But no one wanted to buy them. Instead of making the rest of the world incompatible with IBM and therefore unattractive to customers, PS / 2 made IBM computers incompatible with the rest of the world. PS / 2 sold only IBM and eventually the company gave up.
The new strategy was a grandiose, all-embracing, marketing nonsense called Systems Application Architecture and a very IBM-based answer to open systems. SAA promised a huge set of standards that combine IBM’s developer interfaces, user interfaces, and communications protocols. But IBM has long ceased to create reasonable standards. For example, IBM systems did not use the industry standard for text encoding, ASCII - they used their own, called EBCDIC - which existed in at least six different versions. SAA did not make IBM products "open systems".
New PC operating system
The third focus of IBM’s efforts was a new operating system for the PC — OS / 2. It was a new challenge for the company. For many years, programs were just a part of a “computer” bought by a client; selling software separately from IBM computers was only forced by a 1968 Ministry of Justice decision. Most of the software continued to be written for a particular hardware and was tightly tied to it. IBM has never before sold software in an open and mature market, where users make the choice themselves. These were uncharted waters in which extremely aggressive creatures hid.
With its multitasking, polished multithreading, inter-process utilities like channels and queues (IPC goodies like pipes and queues), the graphical interface and the new OS / 2 file system, it was as good as its competitors. The kernel was very well designed and able to simultaneously run almost any application. The OS / 2 API also traced forward-thinking system planning with long-term use.
IBM led the development of OS / 2 together with the tiny company Microsoft, and despite many disagreements, they somehow reached the joint release of the system.
OS / 2 was a truly new “platform”. But it was unusual for users and at the time of launch did not have any applications. OS / 2 needed what we now call an “ecosystem.” And what to do in this case?
What is a platform?
IBM at that time did not have any idea about the idea of ​​a hardware-independent platform. In the literal sense of the word, she did not know what she did not know. But how could she know that? Customers usually bought what IBM advised them to ask, questions only arose regarding the timing of the purchase. OS / 2 required a market for developers and support from iron manufacturers so that OS / 2 could work on a wide range of video cards, printers and other peripherals. It was a classic chicken and egg problem. Why allocate funds to conquer the market, which is not yet?
With all its technical correctness, OS / 2 offered very few benefits, and its implementation was a major headache for corporate support services.
Multitasking comes to PC: OS / 2 version 1.0
The size and prestige of IBM forced several major developers to make versions of their OS / 2 DOS programs, Ashton Tate, for example, transferred dBase, but moving applications did not mean moving the ecosystem. Driver development was hellishly difficult, after three years it was extremely difficult to type something from OS / 2. Nowadays, platform manufacturers understand that they need to support, persuade, and even bribe application developers and drivers. Microsoft knew this, IBM did not.
OS / 2 backward compatibility was also not impressive. Originally designed by 16-bit, the new system did not use the new features of 32-bit Intel 386 processors to launch DOS virtual machines. The “DOS Compatibility Tool” on 16-bit OS / 2 was slow and unreliable. Without attractive applications and iron support, the only advantage of OS / 2 remained the ability to use more than 640 kilobytes of memory in applications. But in the summer of 1988 there was a jump in the price of memory, which made the 2 megabytes necessary for OS / 2 extremely expensive. The graphical user interface was even more demanding, Presentation Manager made most of the existing PCs too weak for OS / 2.
The third party patches for DOS, adding multitasking and expanding the available memory, have become the most reasonable way out for users.
The first version of OS / 2 entered the market in 1987, with multitasking and a text interface. The GUI version came out only a year later. It still did not have a modern HFS file system, with support for partitions and large files.
OS / 2 failed
Different workstation ideas
In the summer of 1988, one project changed the relationship between partners. DOS could not use more than 640 kilobytes of RAM, which gave rise to a thriving market of memory expanders, which with the help of various tricks sought allocation of a larger volume. One of the undoubted advantages of OS / 2 was the mechanism of working with memory, which does not impose any restrictions of up to 15 megabytes. But this summer, one of the Redmond trainees found a way to run DOS in Intel 286 protected mode, giving it more memory.
The head of the trainee showed his work to Steve Ballmer and the Microsoft management thought about the possibility of using the Windows shell as a memory extender for DOS. A PC could load DOS, start Windows, and gradually replace DOS components at a lower level, similar to replacing wheels on a moving car without stopping it. The intricate way. But Windows was full of similar hacks - in the technical sense of the word it was self-modifying code.
The “shell plus memory expander” did not have the rich OS / 2 APIs — and the system would never be stable. It did not even have long file names. DOS used an archaic 8 + 3 system (“LTTR2BOB.WPD”) a decade ago. It was not an operating system at all. But it had a graphical interface and with the help of smoke, mirrors and a fair amount of faith from users, it could be presented as a new operating system.
With a light stretch, Windows could be advertised as a new “platform.” The new platform meant that all existing applications continue to work and soon new, radiant and beautiful will join them. At that time, no graphical shell over DOS had a significant advantage. The probability of victory was very high.
Microsoft had some very serious advantages. First of all, she had a very good position for the development of the “ecosystem”, there were development tools, she could sell them very cheaply and knew how to advertise Windows as a “new era”, something with ever-increasing acceleration. Microsoft didn’t have conservative units to pacify.
Secondly, she had a very good relationship with PC makers. They paid for MS-DOS regardless of whether they installed it on a PC. Buyers needed something that could run dBase and Lotus 1-2-3. Microsoft could easily add Windows to all of these computers, and the first run of Windows was easy and painless.
Thirdly, the press and experts were very keen on the success of Microsoft and the failure of IBM. Not because they loved Microsoft very much. But the clones created a thriving market and no one wanted IBM to once again control both hardware and software. Nobody wanted to return to the world under the rule of one huge, vertically integrated giant. This was not obvious in the summer of 1990, three years after the announcement of OS / 2, when Windows 3.0, the first version with support for protected mode, was released. Almost all of the press considered Windows 3.0 another rival of the GEM shell and the Quarterdeck memory expander.
Divorce
For many months, before the release of Windows 3.0 in the summer of 1990, Microsoft distributed copies of it that all people in the industry seemed to have. It was clear that Microsoft sees in Windows something more than a flower on the grave of MS-DOS and a farewell to pre-OS / 2 world, which remains somewhere behind. Nothing like this. Microsoft has done a lot of work to increase the attractiveness of Windows and over-actively advertised it among developers. It was a “pseudo-platform”, not actually a new OS, but allowing developers to use a graphical interface. Fed up with the meager sales of products for OS / 2 developers reacted very positively. At the same time, Microsoft was obliged to continue demonstrating the 32-bit OS / 2 to the press, and there was all that was lacking in the 16-bit version.
The pseudo-mutant platform Microsoft Windows 3.0 was launched with a lot of noise and began to sell well. It became clear that during 1989 at least some of the developers who were studying OS / 2 Presentation Manager actually turned towards Windows. Windows gave them a lot of opportunities and turned into a chance to get the market, even if it was modest and short-lived. But at least something was better than nothing. Soon, more and more applications began to appear under Windows, after which they began to flow, as major application vendors plugged holes in their portfolio. In just one month, Microsoft sold more copies of Windows than IBM copies of OS / 2 in two years.
During 1988 and 1989, IBM tried several times to convince Microsoft to shut down Windows, but Bill Gates refused. After that, the head of the PC division at IBM, Jim Cannavino (Jim Cannavino), made an unintentional but catastrophic mistake. Aware of Microsoft’s growing enthusiasm for Windows, he couldn’t recognize that they were right and give the system any chance of breaking the agreement with Microsoft. As a result, Windows received a cool endorsement - the opportunity to be born and the space for growth. Despite the complete incompatibility of the Windows API, it was positioned as a preliminary version of OS / 2. IBM had many opportunities to get rid of Microsoft and protect OS / 2, but this error was the most destructive.
This was in September 1989. IBM’s most terrible mistake was maintaining partnerships with Microsoft and refusing to suppress the development of Windows. The relationship between them quickly deteriorated, as Windows was sold in huge quantities.
The ecosystem benefits Ashton Tate and Lotus enjoyed quickly began to evaporate. Windows provided drivers for devices. New skills boiled down to the ability to handle Windows. Knowing the intricate sequences of nonstandard hotkeys was no longer an advantage. WordPerfect developers desperately tried to hold positions by transferring hot keys from DOS to the first version under Windows, but did not succeed.
Marsh on industrial systems
In the business field, Windows PCs have begun to use clumsy end-user applications with forms. In May 1991, Microsoft launched Visual Basic, a GUI-based scripting language that is primitive but capable of calling routines from compiled libraries. Windows also gave a huge advantage to the Microsoft application development divisions. In the DOS world, the market share of Microsoft Multiplan and Word was small. But the positions of their Windows heirs, Excel and Word 2.0, were much stronger. In addition, Microsoft began selling a bundle of its office applications with an Access database and an e-mail program that is noticeably cheaper than its competitors. The entire “Office for Wndows” cost less than one dBase or Lotus 1-2-3.
Two years IBM and Microsoft waged a PR war of attrition. IBM could not kill Windows and had to get rid of Microsoft.The inevitable divorce occurred in April 1991. Under the final agreement, IBM independently took up the development of a 32-bit version of OS / 2 and obtained a license to include the Windows code. Microsoft immediately stopped developing 16-bit OS / 2, although for several years it continued to sell the Microsoft OS / 2 LAN Manager, all reducing the font size of the word OS / 2 on the box.
Paul Carroll, in his book Big Blues: The Unmaking of IBM, notes that at the time of public divorce, Microsoft sold 13 million Windows 3.0 during the year, while IBM sold only 600,000 OS / 2 copies in three years - of which only 300,000 were actual sales, much less was actually used. In the wake of Windows sales and demand for Microsoft Windows-based applications, the company grew from 1,800 employees to 10,000.
But in spite of all this, power was not on her side. As Gates himself later wrote
When they broke off all ties and decided to continue working without us, we thought: “Now we are on our own and this is definitely very, very scary.
And the worst thing was that in the 32-bit version of OS / 2 there was everything that 16-bit lacked. It was a very impressive system, it was just not completed yet.
The Empire Strikes Back
A new, 32-bit version of OS / 2 fully utilized the capabilities of the Intel 386 processor: a flat memory model and the ability to virtualize DOS sessions while maintaining the connections between their processes. The 32-bit version also supported unreliable DOS memory extenders, giving DOS applications fast 32-bit memory access so that they themselves did not notice this and running DOS applications running in fall-protected virtual sessions even better than earlier. IBM and Microsoft were equally proud of this. IBM took Microsoft Flight Simulator, one of the most demanding graphics applications for DOS - and ran one copy after another, until a whole dozen started to work at the same time.
Suddenly, OS / 2 began to look much better than its competitors, solving many of the problems of DOS business users with whom they had been tormented for many years in a row. Even Windows programs worked in multiple virtualized sessions faster than in DOS. OS / 2 combined the best features of all worlds: old DOS applications with their trained users and third-party add-ons, a new wave of Windows applications and native OS / 2 applications.
However, the system was delayed. IBM decided to add OS / 2 shine with the radically new Workplace Shell. It was not just a user interface, but a collection of object-oriented class libraries that could be overridden or extended by developers. Workplace Shell has added many Mac features and new features to OS / 2. Even nowadays, she looks very good. Oddly enough, OS / 2 has become increasingly popular among enthusiasts, thanks in large part to the stunning demonstrations of David Barnes, one of the great demo gods in our industry. Here is Barnes in all its splendor in 1993 .
Microsoft needed a beautiful story and they found it. They had a new, multi-user and portable operating system on different processors at an early stage of development. It turned out to be fateful. OS / 2 3.0 or OS / 2 NT has become one of the largest vapourware in the history of the industry. In 1990, it was only able to boot the core, only three years later someone was able to buy it, and only six years later it began to have a noticeable impact on the market. But she radically changed the public opinion.
RISC business and Unix killer
At that time, the ability to work on different processor architectures seemed very important. The common wisdom said that the microprocessor architecture from Intel is exhausted and the future for RISC. The RISC processors from MIPS and Sun (SPARC) dominated the workstation market, followed by DEC (Alpha) and IBM (Power). It was difficult to understand which of them would win in the end.
The new dummy platform also promised something that Unix manufacturers had to do, but could not: a single platform with rich APIs. In the world of Unix reigned the tyranny of small differences. The developers fought for the smallest details that had no meaning for most users. They made complicated plans and plots, merged and diverged, threw out most of the attractive changes (like the NeWS Display Postscript UIfrom Sun) for the sake of fruitless attempts to find a compromise.
Unix was supposed to be united, but IBM joined the all-except-Sun alliance called OSF, and Microsoft’s enterprising marketers were able to take advantage of the resulting vacuum. NT was a dummy, but Microsoft promised help in porting it to MIPS, Alpha, and even IBM / Motorola PowerPC. The company promised that 16-bit OS / 2 and POSIX Unix libraries will work in NT.Microsoft also promised a uniform Win32 API that allows it to run programs on both NT and Windows DOS-based machines. In 1993, the participants of Unix-wars realized the threat and signed the single " Spec 1170 ", but this didn’t bother anyone.
A non-existent platform allowed Microsoft to retain customers. Microsoft offered customers to wait for a 32-bit version of Windows, Chicago, which was supposed to be released in 1993. It was still based on DOS, but offered the user many of the advantages of OS / 2, such as long file names - and had to work on 4 megabytes. They also had to prepare for the Unix killer, Windows NT. NT was supposed to not only kill Unix, but eventually turn into a comprehensive, all-powerful, object-oriented Cairo system .
Promises went beyond the clouds. IBM realized that it had the last chance to return the market and at least part of the multi-billion dollar cost of developing OS / 2. In the fall of 1994, a fairly optimized version of OS / 2 3.0, Warp, was launched, followed by IBM-wide mass television advertising. It was supplied in two versions - one included Windows, the other used Windows already installed on the hard disk. IBM much earlier than Microsoft realized the potential of the Internet - it was the first OS that came with a browser. Microsoft at that time was busy with an ambitious project of a carefully fenced garden, Compuserve Killer, which was supposed to enter Chicago, and did not even include the TCP / IP stack in Windows.
For a short time, OS / 2 captured the imagination of a small part of customers, but was never able to change the situation on the market - largely because of all the major PC makers, only IBM decided to sell computers with OS / 2 pre-installed. IBM support did not know how to work with unprepared users. New users of OS / 2 met the time-consuming installation process, poor support for hardware, a small number of native applications. OS / 2 supported Windows applications so well that in the absence of dangerous competitors among native OS / 2 applications, all serious developers focused on the much more profitable Windows market.
As a result, Chicago came out only in 1995 under the name Windows 95 and did not work on 4 megabytes. NT was too big and slow for the market, it lacked office applications - the promise of portability Win 32 was never fulfilled. NT came out a year later and users demanded the Windows 95 shell. By that time, the system architecture had changed so much that most of the promised features were never implemented. Microsoft imposed restrictions on the number of network connections for workstations (which could be used as departmental servers for this) and moved the graphics drivers to the core, sacrificing reliability for the sake of performance. About Cairo quietly forgotten.
Last spurt to glory
In the late summer of 1996, IBM admitted defeat and transferred most of the programmers to other tasks. She released another version of OS / 2 in 1996, incorporating voice recognition and a more familiar interface. Support for this system has continued for many years. But the war was over - Microsoft won. OS / 2 IBM CEO Lou Gerstner finally convinced the death of native OS / 2 applications in the senselessness of further development. This was partly due to IBM itself. The famous strategy “Workplace OS” implied the launch of personalized versions of the OS on top of the microkernel - but failed to attract customers and was never completed. More than 2000 developers were engaged in a portable version of OS / 2, mainly designed for IBM workstations with RISC processors, but never brought it to a stable state.Apple's joint object-oriented OS projectTaligent ended in nothing. IBM could not buy Apple. Microsoft Win32's tenacious strategy has conquered the market.
WPS - WTF?!?
Workplace Shell first appeared in OS / 2 version 2.0, released in April 1992 — simultaneously with Windows 3.1. There is no task pane, no program directory, no system menu, no shutdown menu ... all applications are located in the “System” directory. Minimized applications go to the folder. OS / 2 had a radically new interface that lacked many of the elements of modern UI. Where did he come from?
He was born from IBM’s plans to create an office of the future - literally object-oriented functionality was invented in research laboratories - this is in the print ads of 1990 that preceded Workplace Shell.
AboveOfficeVision employed 1,500 programmers, but the project was never completed. To keep the promise made to customers about developing a collaborative environment, IBM bought Lotus in 1996.
Such office automation tools were difficult to sell, but IBM emphasized their flexibility. They were so flexible that you could change the font and background image of each window and each window element individually. Workplace Shell had more useful features. Like Mac, shortcuts (links) to files tracked their movement. Template directories could contain applications and quickly be cloned. Closing the directory minimized all applications and documents that were associated with it. There were other oddities. By default, the icons could not be moved with the mouse, although this quickly changed in the settings.
The latest version of OS / 2 included much more familiar elements. The system menu gave quick access to each file, and Mac icon designer Susan Kare, who also made icons for Windows 3.0, worked on the appearance of the icons.
OS / 2 Warp 3.0 in the fall of 1994. Under pressure from customers, a hurriedly written taskbar was added to OS / 2.
Last battle: OS / 2 Warp 4.0 in 1996.
For some reason, IBM made deafening sounds for each mouse action (for example, dragging).
Why did IBM lose?
Around the rise of Microsoft to the top of the PC industry, passions are still raging. If in 1987 you had said that Intel and Microsoft would determine the future of the industry for 10 years, they would simply laugh at you. Intel could not create a competitive RISC chip, and Microsoft never wrote any operating systems. But the questions "Why did Microsoft win?" And "Why did IBM lose?" Have completely different answers.
IBM has never before faced with so enterprising and ruthless competitors like Bill Gates. Gates fought desperately and IBM could not get rid of it. But regardless of the presence of OS / 2, IBM could replay it or even trite it with the masses if it had developed a coherent strategy that gives customers good reasons to spend money on IBM.
IBM could have realized that in 1988, OS / 2 was too revolutionary for customers. She could take control of DOS and offer clients a smoother transition to the world of graphical interfaces. In fact, IBM tried to do just that with a project called PM for DOS - which was never completed. It was too late. IBM could buy Apple with the most attractive user interface and put it on a solid technological foundation. As an even more revolutionary solution, IBM could recognize the real value of software and move the competition to a higher level — by creating a free platform like Linux.
Many alternative history options can be offered - but most of them will remain a game of imagination. After Microsoft gained strength and captured the minds of the developers of the OS / 2 single platform, it would be very difficult to get them back. More interesting is another question - why did Microsoft win?
First of all, Microsoft ruthlessly followed the requirements of the market and cared only about the number of copies of Windows sold, it was not concerned about the attitude of others and the theoretical correctness of the actions taken. The market did not like Microsoft, but Microsoft brought money to the market. Software makers, hardware, peripherals - they all earned more with Microsoft. The programs infuriated you every day, but the “ecosystem” as a whole looked much healthier. IBM had no idea how to build new ecosystems. The high authorities believed that if they made the system and it would be good (and it eventually became so), then there will be buyers - simply because of the magical power of the three letters IBM. The IBM sales team has mastered many dark methods, including the use of other IBM products and services,but they did not help much in open knife fights with Bill Gates.
Microsoft was completely focused on one task. She did not make compromises in order to appease the units occupied by other products. Gates was not going to, as it seemed to many then, to limit himself to writing applications and tools for foreign platforms. He wanted his platform. Microsoft did a great job on Windows. Needless to say, Microsoft’s huge advantage was the extremely rigid terms of licensing agreements. PC makers had to pay Microsoft for DOS regardless of whether they installed it on computers or not - so why should they install alternatives? In the end, Microsoft lost the antitrust lawsuit and was forced to give in to the requirements of Novell, Be and Sun. But do not forget that two are needed for tango. Partly to blame and short-sighted PC makers.
OEMs did not protest against Microsoft’s similar policies and later it hit them themselves. After the death of competitors, Microsoft raised prices - this is one of the advantages of a monopoly position. As the price of computers fell, the larger part of the price was leaving Microsoft. Sometime high OEM profits plummeted. None of them could stand out from the crowd, Microsoft did not allow it - she insisted on a uniform "Windows experience". Sometimes it is more profitable for a business to sacrifice current profits in order to maintain competition and preserve future profits. Many are actively engaged in this - but not the PC industry.
In the OS proceedings following the end of the war, many OS / 2 fans blamed the computer press for total hostility to the system. In many ways, they were right - journalists who used and praised OS / 2 can be counted on the fingers of one hand. One of them was John Lettice, the founder of The Register - the other your humble servant. The beautiful OS / 2 2.0 souvenir from the illustration above is still in our office. But the press simply reflected the then generally accepted opinions and economic decisions. She also did not want a world in which IBM defines open system standards.
Everyone was so busy fighting the outgoing war that they did not notice the approach of a new one. In 1995, when the OS war was still raging on Usenet (although it was clear to everyone who won), Guy Kewney summed up his annoyance:
My friends tell me that Microsoft will save us from IBM. But who will save us from Microsoft?
More on OS / 2 - the memories of a person who at that time worked at IBM (if they are interesting, they can also be translated)
IBM insider: How I caught my wife while bug-hunting on OS / 2
Where were the bullet holes on OS / 2's corpse? Its head ... or foot?
An alternative overview of the history of OS / 2 on Habré with detailed descriptions of technical issues
Source: https://habr.com/ru/post/187242/
All Articles