IBM

9056 IBM12 - PC DISASTERS

This chapter is longer than the relatively small contribution the PC has made to IBM's revenues over the years might suggest. But the PC has been the focus of people's attention; and too many false lessons have been drawn from its history. My main reason for expanding this part of the IBM story, though, is that the PC was the cuckoo in IBM's nest. The developments it spawned, across the whole range of its management activities, seduced IBM management; and ultimately to almost destroyed the company.

However, the Personal Computer (PC) was, at the time of its launch in the early 1980s, hailed as a triumphal vindication of IBM's management style. Indeed this was true, at least in the short term, but not for the reasons normally put forward. For the emergence of the PC was almost pure serendipity, a succession of happy accidents. The IBM management involved, though, must be given the credit (at least, again, in the short term) for the fact that they let this happen, and built on these fortunate occurrences. Over the next few years IBM did consolidate its position in a new market, very quickly gaining a dominating 80% share (as high as it had previously held in the mainframe market); and in one that was particularly important for the future; both for good and, especially, for bad!

Emergent Strategy…the most powerful strategy is normally the one that emerges from the market, not from within the organisation. If you succeed, even accidentally, in meeting the needs of the customer you have found a winner. The secret is in recognising that fact. 'Emergent Strategy' is one aspect of theory which is taught by some academics (most notably Henry Mintzberg); albeit usually after all else has failed!

In the mid 1080s, at a time when almost everyone else was still enthusing about IBM's success with the PC, I warned that "....the initial, unplanned, growth of the PC created at least two hostages to fortune. The first of these was the 'open architecture', which was forced by the urgent timescales but was also a great contributor to its initial success (allowing other suppliers to provide the hardware and software IBM could not deliver). It did mean, though, that IBM could not have its usual firm grip on the market. Its competitors could, and soon did, easily create 'clones', which were functionally identical; and IBM just as soon lost volume sales to such cheaper copies produced in the Far East." That process has gone much further since!

IBM PC Compatible - the Treacherous Standard

Thus, it is undeniable that one of the main reasons for IBM's dramatic success with the PC, in the short term, was its introduction of an 'open systems' approach. Many other suppliers were encouraged, by the opportunities available and by IBM itself, to offer the bits that IBM was failing to provide. As a result, IBM was able to very rapidly offer a very wide range of hardware and, especially, of software.

The problems emerged later on two fronts. The first was an obvious extension of IBM's experience in other fields; the emergence of IBM compatible suppliers. These were not content to supply the bits that IBM didn't offer, they wanted to replace the main units themselves - indeed, they eventually wanted to supply their own 'IBM compatible' PC (complete with everything - processor, disk drives, software etc - that IBM was trying to sell itself). IBM's only protection, the only bit it really owned, the BIOS which sat between the software and the hardware, was soon reverse-engineered; that is legally copied, in that the copy - whilst providing the same functions - did not use the same code (which was the part protected by copyright!). With the exception of Compaq, almost all of these were selling on price alone (as you would classically expect them to, as minor brands in the market). The real tragedy came later, when IBM chose to get down in the gutter and try and fight them on their own ground, on price (a mistake which no first year marketing student should make!).

Control Passes to Others…the second problem was that others were less charitable, and more astute, than IBM; and substituted their own monopoly 'standards' for those which IBM should have owned. The most immediately obvious of these were Intel (which now has been so successful, not least in advertising its standard, that PCs are now sold as containing one of its chips rather than as 'IBM compatible'!) and Microsoft (which is now valued, even after the dot.com crash, far in excess of the capital value as IBM, because of the virtual monopoly position IBM gave it). But others followed. Hewlett Packard now offers the printers of choice (and certainly the standard of choice for its competitors) and Cisco dominates the network hardware. Microsoft also has both Excel the spreadsheet (having also displaced Lotus, now - paradoxically - owned by IBM itself) and Word the word processor (having displaced Wordperfect) and once more controls the standards for these. IBM, unfortunately does not hold a dominant position in any of these fields (not even, now, in the provision of the overall PC!).

Core Competences…there are certain specific 'assets' - tangible or intangible - on which the future of an organisation depends. They represent its unique abilities (its 'core competences'); which it must protect at all costs - if it is to survive. Once more, this (core competences theory) is a theme which has emerged from corporate strategy theory in recent year; though the corollary, that this should not be put at risk by following the fashionable move to 'buy-in'  has been less forcefully promoted.

Dealers… I also highlighted another problem which again later came to haunt IBM "The second potential problem was that of losing control of its end customers. A key to the success of the PC was the introduction of dealers (Third Parties) to sell it. IBM enthusiastically embraced these, though in the process it failed to recognise that it still needed to control the end‑users; but by using the techniques of the mass marketer..." IBM's failures in this context are discussed in more detail later in this chapter.

PC Group Sub-Culture

Yet again as I forewarned in the 1980s, in the process of creating its PC group ".... IBM built a new group of personnel (largely recruited from outside IBM), with a very different culture, which in addition to the successes made a number of atypical (for IBM) blunders; including that of producing follow‑on products that did not as fully live up to their technical promise and, most importantly, that of stimulating a price war that was ultimately to be very damaging to the rest of IBM."

I have said a great deal, elsewhere, about the overall IBM culture. What I have not mentioned was the existence of sub-cultures which could be almost as strong. These tended to emerge as a reaction against that which those involved saw as the dominance of the larger systems groups - thus, GSD culture had some special characteristics, though it was never far from the norm, but OPD (Office Systems Division - which, quite correctly, thought it was neglected by IBM) had a very different culture; at times it almost revelled in 'cowboy' approaches which would have been anathema to the rest of the IBM (and maybe that was why they adopted them).

The PC Group, in turn, also developed its own sub-culture. In part this was for the traditional reason; it saw itself as in competition with, and dominated by, DPD. In part this was forced on it by the management directives to the IBUs in general - to ignore the IBM bureaucracy and 'do their own thing'. As with the previous IBUs, it became almost a test of virility in PC Group to pick a fight with the 'bureaucracy'; after all, its members had been told by top management they had the right, and duty, to do it their way. As with the other IBUs, however, this meant that they often found their way blocked. It was not, after the initial honeymoon period anyway, possible to beat them. Previously, my own Biomedical IBU in the UK was one of the few which was successful; precisely because I knew the system, and worked with it from the very beginning. I never picked a fight with the rest of IBM. I made it my partner, and sold the bureaucracy very hard (and successfully) on the idea of how exciting it was to work with my group. PC Group, on the other hand, generally followed the combative route; which made it that much less effective, and left a degree of bitterness which long clouded relationships with the rest of IBM.

Cowboys…much of the culture came, though, from the rather strange make-up of the people in the group. PC Group had been told, as had the other IBUs, that it could have the pick of IBM's best personnel. It didn't work out that way. For a mainstream IBMer the thought of joining this tacky upstart, when everyone knew that what really counted was the large mainframes, was literally unthinkable. As a result only a handful of PC aficionados, myself included, joined it from the larger divisions of IBM. Of course, with almost nothing to lose, members of OPD flocked to it; bringing with them their own peculiar sub-culture. To this mixture were added the many outsiders who had to be brought in, and who found the OPD version easier to understand than the sedate version to which the rest of IBM was committed. The resulting culture almost became an 'Anti-IBM' one!

The result was often like a shoot-out in the Wild West, with everyone hankering to draw first. Certainly, shooting from the hip was the preferred management style - not least because it established a key difference from the rest of IBM (and, to be fair, it was essential - in the early days - if the deadlines were to be met). This was often unproductive, especially where the managers involved were not the best in IBM (and those few of us who were, often were seen as moles from DPD). More important, it built a tradition of conflict with the rest of IBM which was totally counter-productive as, in later years, the success of both was going to depend upon their ability to work together.

Culture War…strong sub-cultures can, at times, be built in order to defend a group against its dominant superiors. Sometimes defence can turn to attack, and the sub-group can start to wage war on the rest of the organization. Social Science theory, at least, is a useful starting point for understanding the clashes between such sub-groups, especially those embedded in bureaucracies.

The History of the Personal Computer

As has often been the case in the past, IBM probably did not invent the personal computer (depending upon how you define the process and the product); but in the same way as Japanese companies have developed and expanded other markets, IBM made the PC its own - for a while - at least.

Legend has it that the personal computer revolution started humbly in a garage in Silicon Valley. Are there perhaps biblical undertones in such reverence for the new technology born in a 'stable'? As with most legends there is a grain of truth, but there is also a great deal of hype. The Santa Clara valley, the Silicon Valley of the myth, had long been the home of leading edge electronics developments, ever since the 1930's when Fred Terman, as Professor of Radio Engineering at Stanford University, started to sponsor high‑tech companies in the area. Indeed one of his biggest successes, William Hewlett and David Packard, also started their own company (Hewlett Packard; HP to the industry) in a garage there, but they did this while still Fellows at Stanford (so even their story was not quite rags to riches, more academic gowns to riches - as were many more which followed).

After 1945 a number of other high‑tech electronics firms were started in the area, often as break‑aways from earlier start‑ups. At the same time a number of larger companies brought plants, and in particular development laboratories, to the area. Foremost amongst these was IBM, with no less than four development labs in the area (Menloe Park, Santa Teresa and two in San Jose). The area, as a result, was an ideal nursery for new electronic developments (with the complete infra‑structure of suppliers within a few miles radius), and a significant number of new computer companies were indeed founded there; including Rolm, IBM's later acquisition.

Apple…it was in this environment, then, that Steve Jobs and Steve Wozniak made the first Apple computer, indeed once more in their own garage; using the proceeds from the sale of Job's car, a humble Volkswagen, for their capital. The original Byte shop ordered the first 50 units, and they were on their way to becoming billionaires. These are some of the facts behind the legend.

The key facts that are omitted from the legend are that Apple was still a very small scale operation until Arras C Mirrkula, already a millionaire, put in $91,000 himself, persuaded Bank of America to chip in $250,000 and relieved other venture capitalists of a further $600,000. Thus Apple was bankrolled to the tune of a further $1 million beyond the cash from the sale of the Volkswagen before the main take‑off occurred. The message for the budding inventor is first find your tame venture capitalist, and only then start in your garage!

The Apple II computer then produced was, as the PC too was later to be, largely assembled from "off the shelf" components. Wozniak's brilliant design used these in new ways, however, and the floppy disk controller, in particular, only required 8 IC's (integrated circuits) instead of the 30+ used previously; and allowed disk storage of data, a basic requirement of business computers, to be a practical addition to personal computers. By the end of the 1970's, then, Apple had emerged as the leading force in personal computing. It was in the right place at the right time with the right product, just when program developers wanted such a vehicle. Indeed its main strength by then was the literally thousands of programs which had been developed to run on it.

Despite the on-going interest in the media, which loved every one of the company's imaginative developments (not least because they made IBM look clumsy), Apple was never - after its heyday (before IBM entered the market) - much more than a side-show (even its much vaunted desk-top publishing capabilities were only relevant to a minority audience). Though many of its ideas were interesting (and often exciting) it never took enough market share to even excite the interest of the clone-makers who were IBM's downfall. But it did popularise one new feature.

With its initially less than successful launch of the non-IBM-compatible Macintosh PC - following the disastrous launch of its Lisa competitor to the IBM PC - it popularised the GUI (Graphical User Interface) approach. The ideas for this had in any case been derived from the (also unsuccessful in terms of market success) work of the Xerox PARC (Palo Alto Research Group) team. Thus, Apple set the scene for future users to spend their days lovingly fondling their mouse and gazing at picture of desktops and  dustbins; rather than gazing at incomprehensible instructions written by programmers. Even the IBM compatible user was later, through Microsoft's windows which copied the principle yet again, to fall in love with GUIs!

PR as History…be very careful when you listen to the stories which are told about how success was achieved. Most of these stories are invented by the company PR professionals rather than their historians!

IBM's PC History…at the beginning of this section I said that IBM probably didn't discover the personal computer. This is not to say that IBM didn't investigate the market; just that initially it was relatively unsuccessful. Ever since the formation of GSD in the 1970s there had been a drive to produce ever smaller machines, to tap ever wider markets; which was then the mission of GSD. Thus, the first "personal computer" launched by IBM was its 5100. Announced in 1975, after the first Altair 8800 (but before this stimulated Apple and other developers to produce their own machines) it did not have disks (like the Altair it was based on a tape cartridge for data storage), but it did have almost all the other features of the later PC's. It had only 64k of RAM memory (the same as Apple) but it had up to 400k of ROM (read only memory) with a very sophisticated operating system and the ability to run programs in APL language as well as Basic. It was even portable, though at a weight of around 70 lbs. this did stretch the imagination (and arms) somewhat.

It was primarily intended for the scientific market, hence the provision of APL (a mathematically based language primarily used by scientists). It was a dramatic flop, since the marketing force in GSD did not give it the resourcing needed ( and in any case GSD no longer had a customer base in the scientific area).

The development team at Rochester rethought, and came out with the 5110; aimed fairly and squarely at GSD's traditional commercial market. It was a 5100 with diskette drives capable of storing up to 4 megabytes of data on floppy disks. At around £8,000 it was more expensive than modern PC's; but only just so (where the average PC, with its peripherals and software, probably cost around £5,000 in the years immediately after its launch). Once more, though, it was a relative failure; but not because of the technology. As the Announcement Manager (New Product Marketing Manager in non‑IBM parlance) I, together with my colleagues, was well aware that the 5110 was unlikely to be a dramatic success, since the marketing resource would be limited to what could be obtained riding on the back of the mainline GSD products, such as the then very successful S/34. We recognised that no salesman worth his salt would, or should, consider selling a 5110 when he could spend the same time (in a resource constrained GSD) selling up to ten times the value of larger computer. The significant factor, of which we were then not aware, was that this machine took only 90 days from conception to production; though I should have guessed as much, where the backplane of the pre-production machine I had in my own lab was clearly handwired! It achieved this miraculously short timescale under the management of Bill Synes, who - as a member of Bill Lowe's taskforce - later did much the same for the PC.

The Better Mousetrap…maybe in Ralph Waldo Emerson's time it was sufficient to invent a better product, and the world would beat a path to your door, but now the world has to be sold its benefits. This is a lesson many involved in R&D have failed to learn.

I learned this lesson long before I landed at IBM. Early in my career I had to kill an oven-spray (a technologically advanced product which protected an oven from dirt ever sticking to it), when I found that all the advertising had to start "This is not an oven-cleaner…"!

It is arguable, therefore, that the greatest breakthrough the PC eventually achieved was that of opening up the Third Parties distribution channel. Equally, as we will see later, it was also a Trojan horse; and was one of the major contributors to IBM's later downfall, But, until this outside marketing resource was made available, IBM simply could not release the internal sales resources necessary to reach the PC market.

The earlier developments contributed nothing to the PC Group; though the research and awareness that they promoted certainly did (as early as 1976 the UK, at my own prompting, was putting pressure on the US labs for a genuine micro to compete with Commodore and Apple, then just starting out). As explained above in the event the PC did not use conventional IBM sales channels. It did not even borrow any of the technology. Indeed GSD's Rochester development group was still struggling with their own separate line of development, to produce the follow‑on to 5110; the S/23 Datamaster, to be sold as before  through GSD. Indeed this was launched at the same time as the PC; as IBM inevitably hedged its bets! Needless to say, although it was an excellent product technically, it too disappeared into obscurity.

IBM was, therefore, definitely in the business of trying to enter the PC market. By 1980, a number IBM'ers (myself included), had been pressing it  - for almost half a decade - to get into what was obviously (it seemed to me, and the events to come justified my view) a potentially enormous market. In a sense, therefore, I should accept some of the blame for the final outcome - though my excuse would be that I would have done it very differently (and I did later spend much time trying to persuade IBM management to change their ways).  Indeed, by that year, apart from the new GSD S/23 entrant there was already yet another entrant, from IBM Instruments Inc. It was almost inevitable, therefore, that an IBU would also be set up to explore the opportunity; this was after all the heyday of IBU's!

The first effective product champion, in 1980, was William C Lowe, the manager of the Entry Systems Unit, in Boca Raton, Florida, where the PC eventually came to be manufactured. It was he who conceived the project, then code-named 'Chess', and assembled the thirteen engineers who later entered legend as the 'Dirty Dozen'; and finally obtained the go-ahead from the CMC towards the end of 1980. The ease with which this permission was partly to do with John Opel's drive at that time for 'intrapreneurial' Independent Business Units, but mainly because it was seen as peripheral to IBM's business as usual. The PC was only expected to have a lifetime demand of 250,000 units, and only needed $14 million in start up capital. If only the CMC had realised what a Pandora's box they were opening!

PC Development

William Lowe's first decision, at the end of that year, was to give Philip (Don) Estridge the "mission" to set up an IBU at Boca Raton; initially supported by just those 13 engineers (albeit with proven track records). According to Buck Rodgers they were given the brief "We want an IBM Personal Computer. We're already late, so you'll have to hurry. Do whatever you feel is necessary to get it done". In a matter of only a few months he and his small team then put together the whole PC operation. Indeed, the first demonstration model (put together to show the July 1980 CMC)  was assembled from the standard components - which were to become the PC's special feature - in just 30 days! Somewhat mysteriously Peter Drucker claims that four task forces were set up as early as 1977, but there is no evidence that any action was taken before 1980 that directly led to the PC itself.

The truth is that nobody else in IBM wanted it. Even though there was a real buzz about the new project, Richard Young (then President of IBM's Office Products Division, OPD, which might have seemed the most obvious home for it) said he had no intention of forcing it on his salesforce. Worse, although he supported the idea of it being sold through OPD's 'Product Centers', he opposed it being sold elsewhere (since that might mean the dealers, such as Computerland and Sear Business Centers would also want to sell IBM's typewriters!). Fortunately (or taking the longer term view, regrettably) he was overruled by Opel.

The birth of the PC - by then code-named Acorn - has sometimes been presented as a triumphal vindication of the IBM system; a careful implementation of Armonk strategy. The timescale, however, precluded such niceties. Not least the move to 'open architectures' might have been inevitable given their near impossible deadlines, but nobody in IBM apparently asked the question "If we can do it in 30 days, how long will it take our competitors to copy it!" To achieve what they did, in months rather than years, they had to bypass almost all of the traditional IBM systems. As earlier, the factor that allowed this was probably the, temporary, hiatus in the Armonk bureaucracy. In this "window" the PC slipped out in a form for which it might otherwise have been difficult to obtain approval.

Skunkworks…although much of the incremental development that takes place comes from formal R&D (though the hype does not stress the mainly incremental nature of this work), many of the genuinely new to the world products have emerged from places they never should have been found. The skill, as I said earlier, is recognising the value of these. Apart from the work of Tom Peters, which is much undervalued in the academic world, this element is largely unexplored.

The Buy-In

The form of the PC itself, and of its marketing operation, was in part determined by the short timescales, and limited resources, available and in part by the trends in technology, and in the marketplace, that were already clearly evident. It was also believed that the limited workload imposed on a (personal) desktop machine would not require the extensive fail-safe testing demanded of the rest of IBM's computers. So 'industry-standard' testing of off-the-shelf components would suffice. This still left some critical decisions to the team, and to their great credit it is these decisions that made the PC so successful; albeit in the short term. Of course, some were to have almost fatal impacts on the whole of IBM later - but it would be wrong to blame the development team for that. Their job was to get the product out of the door and in that they succeeded magnificently.

The existing trends meant that the PC would almost inevitably use Basic as its main language, and it would at least offer the existing main operating system, CP/M. Its physical layout would follow the individual box approach so successfully adopted by Apple; with a processing unit, a separate monitor and a separate printer. It would, in its most minimal form, take input from a cassette recorder, but would also offer five and a quarter inch floppy disks - as an option. The processor (or "system" unit) would be based on a "bus" or "motherboard" approach so that extra cards could be easily added to offer additional facilities (and to allow for unpredictable future developments); and indeed this was the most revolutionary departure, in hardware terms, for IBM. All of these technical requirements were firmly in place in competitive equipment, and in established market demand, when the PC was designed; and it would have been difficult for IBM to avoid them.

IN the event, IBM signed with Intel for its 8088 chip, a decision which would have been difficult to avoid, Tandon for the disk drives (unusually for IBM as a single-source supplier which was to pose problems later), SCI Systems for the circuit boards and Zenith for the power supplies. Epson was to supply the separate printers. Some bits, just a select few (such as the keyboard, though even this was non-standard), were from other parts of IBM!

Open Architecture…some of them would, however, not have been so inevitable if the "window" through the bureaucracy had not been open. In particular, the "open architecture", for which the team was loudly praised by the outside world, would probably have been anathema to more conventional IBM thinking. The use of existing, commercially available, operating systems together with a motherboard, that almost begged to be filled with competitive cards, would surely have rung alarm bells in an IBM busy fighting off plug‑compatible suppliers on other fronts. It was, indeed, ultimately to prove IBM's undoing!

The timescales and lack of resource also forced a number of decisions onto the team. It had to buy as many components as possible off the shelf. The omens were there, however. Thus, the printer (a standard Epson, rebadged) was almost immediately to  prove somewhat of an embarrassment when Epson started to sell their next generation, improved, printer for less! But it was a period when there was considerable emphasis in the IBU's on sub‑contracting as much as possible. This was an ideal philosophy for the PC: and indeed without such sub‑contracting the team would probably still be trying to develop the product. This was particularly true of the decision to sub‑contract the main operating system software - of which more in the next chapter.

Standard (Assembly) Development…in an age when electronics, especially digital electronics, technology is dominant, the work of developers has increasingly come to revolve around assembly of off-the -shelf components; backed up by the maintenance of good relationships with suppliers. The theory of R & D has yet to catch up with the computer age!

This basic product was still discernible at the time of the launch. There was still a socket to plug in a cassette recorder for data storage, and even more obviously the machine "booted up" (i.e. initially loaded) the Basic for cassette tape support (the advanced version, Basica, which was normal in later versions, had to be loaded from disk!).

Fateful Decisions

Even so, there remained some decisions open to the management and these initially made the PC dramatically successful; though arguably any PC with IBM's full blessing would have been unlikely to be an outright failure. On the other hand, though, the later PC Junior, was just that!

The first decision, and most obvious in the initial publicity, was the adoption of a 16 bit processor chip. It was not a full 16 bit chip in all respects - unlike the 8086 chip it only had an 8 bit address bus - a decision which reportedly was not due to technical considerations but to a political requirement that the CMC did not see it as a threat to other IBM ranges! Even so, the Intel 8088 was markedly more powerful than its predecessors, such as that used in the Apple II. In particular it could address much larger amounts of main memory; the operating system allowing more than half a megabyte of main memory to be used, as against  the previous limit of 64K ( or 65,536 bytes of storage, to be precise!). With the benefit of hindsight this was clearly the correct decision, but it was not an obvious one at the time; since the chip was that much more expensive and all the existing programs were written to run within 64K. Indeed the Basic language offered with the machine was still limited to addressing only 64K! Equally important, 'expansion slots' were provided; so the meager 64K originally shipped in the machine could be expanded - as it soon was (indeed the fiddly first task of any proud new owner was to add the extra chips needed for this!). In the event this extra memory was used to great effect by the program developers. The now familiar sheet metal box was included, however, simply because there wasn't time to make the tools needed to injection mould a plastic enclosure!

The next main decision was to commission a new operating system; PC DOS. This again was a brave decision when it would initially have no application software to run on it, and was incompatible with IBM's other offerings. In practice there was a degree of luck in that the expected developers, Digital Research who had developed the industry standard CP/M, unexpectedly (according to reports in the press) played hard to get and the contract went to Microsoft (and thus made their fortunes). The new operating system offered program developers a better tool; and they accordingly seized on it.

Third Party Distribution Channels

As I indicated earlier in this chapter, the final decision was perhaps the most important. It was a decision taken very early, even before any of the technical details were finalized. Again, though, at the time there probably was no real alternative to selling through Third Parties; and in particular through PC Dealers. On the other hand, it was not an inevitable decision. PC group could have been forced into the normal IBM channels, as the S/23 team were. It could also have followed the more controlled route of "agents"; as IBM attempted later, though with a marked lack of success, on its smaller mainframes.

Perhaps even more important, Sears Roebuck and Computerland executives were involved with the IBM team from the start. The IBMers - especially H.L. ('sparky') Sparks, who was in charge of sales and marketing - relied on them for much of their knowledge of the marketplace. In turn, almost by default, they were to become the main outlets for the new product. Sears Roebuck would set up a handful of centers and, most important, the more than 190 stores of Computerland already existed. From IBM's point of view, this meant that even at announcement there would be immediate widespread distribution across the US. In the event, Sears Roebuck failed to live up to expectations, when the new PC turned out to be selling to the office market rather than the home - where it had originally been targeted. For Computerland, however, it was almost a license to print money; and its then owner, William Millard (who wisely soon sold out and bought a South Sea island kingdom) became one of the first billionaires to be created by the IBM PC!

The Launch…as almost everyone now knows, the US launch of the basic PC, in August 1981, was an outstanding success. On the day of announcement there were already orders for 30,000 PC's; from IBM's US employees alone! This meant that around 20% of its own workforce bought a machine, and in the process began to develop programs and act as trend setters in their local communities. Not that much leading was necessary, for sales far exceeded target and were limited by production capacity for the first 18 months of the product's life; and this was the main reason for the delay in announcing the PC in World Trade, where it was not until 1983 that it finally reached users.

The machine at launch had just 16K of memory installed, and one diskette drive, but still cost $1,595. Even so, and despite the cassette bias to the basic machine, it is clear that the team had already taken another (hidden) key decision. This was that the main product would be primarily diskette based and for business use. Prior to that time many personal computers were in reality used as home computers, and indeed initially many of the IBM PC's in the US were bought for home use. I have pleasant memories of visiting the home of an IBMer in the Santa Clara valley, at that time, and proudly being shown the family's PC which was almost exclusively used to control the "inventory" in their cellar; they were wine buffs as well as PC enthusiasts and saw nothing odd in combining their two hobbies. Despite these idiosyncrasies, however, the PC was eventually placed firmly in the mainstream of business use.

The PC was one of the most successful launches of a new product ever. Its impact on the media was even greater. Orders outstripped production by a massive amount, and production only started to catch with demand in 1983 when production had been increased six-fold. The PC became, especially as conveyed by the Charlie Chaplin commercials, the new face of IBM. Time[i] enthused "The tramp campaign has been so successful that it has created a new image for IBM…has given IBM a human face". It was, however. a hostage to fortune which came to haunt IBM later in the decade when its overall performance (especially in terms of customer service) came to be judged by that of the PC (which was in the hands of its dealers); even though the PC still represented only a minor part of its business. But, at the time, IBM had never seen such adulation and success; and perhaps this (coupled with the flattery offered in Peters & Waterman's book, 'In Search of Excellence') went to its head!

Problems of Success

As a taste of things to come, the problems even started with the initial batch of 1,700 machines shipped to Computerland and Sears Roebuck. They had a potentially fatal electrical fault - which necessitated PC Group personnel flying round the country to rectify in the hours before they were to be shown to the public! But, in general, the early days of the PC group were spent in dealing with the problems of success. The resulting business process, which was eventually extended to World Trade, was essentially a screening process. At one end the 'manufacturers', the technical experts in the software and hardware groups, screened suppliers of products. At the other the 'sales management', DAM's (Dealer Account Managers), screened the applicants for dealerships. At both these extremes there was a long queue of excellent applicants waiting at the door and, as the volume of business was more or less directly proportional to the number of suppliers and dealers accredited, the most necessary IBM skills were those involved in rapid processing of large quantities of application forms. For a while it was a marketing operation totally unlike anything found elsewhere in IBM. The simultaneous glut of supply and demand was a bonanza for IBM, but it also hid some possible problems which were to emerge in less hectic times.

Not least, IBM effectively lost control of the dealers - never to regain it! In particular, the volume discount structures, which pushed dealers into buying more than they could themselves sell, so they sold-on the surplus through a burgeoning 'gray market' amongst non-authorized  (and price-cutting) dealers, destroyed the standards IBM wanted to impose. Worse still, the IBM sales teams, themselves incentivised on the basis of volume sales, would not clamp-down on even the most glaring examples of bad behavior amongst their (highest volume) 'customers'! It was, indeed, difficult (for anti-trust reasons) for IBM to discipline its wayward dealers; but many observers, myself included, thought IBM often hid behind these legal technicalities, happy to sell ever more boxes!

Emergent Strategy…whilst it is sound policy to respond to the dictates of the market, and the Japanese are past masters of such flexibility, it is worthwhile stepping back occasionally to see where these developments are heading - they may be going in directions which will pose problems for the longer-term. Emergent Strategy of this type has been highlighted by Henry Minzberg, and can be a good approach for many organizations which otherwise do not plan for the future, but in any case it is a valuable exercise to test the drift of strategy against a range of scenarios of the future! 

The hard disk version, the PC XT, was announced in 1983 and was as great a success as the basic PC. Indeed the PC XT rapidly became the workhorse of the business world; though, even then normally configured with the maximum 512K of memory, it could hardly have been further from the home computer many people in IBM had expected!

Problems with PC Junior

So, for the first two years at least, IBM apparently could not put a foot wrong. Of course infallibility always begs retribution and the image of the group began to slip with the announcement of the PC Junior. Some commentators now suggest that this was the turning point in terms of IBM's subsequent less successful history; but that is, I suspect, fanciful, since at the time its demise was soon lost in the welter of hype about the rest of the PC range. At the time, though, Time magazine said "The withdrawal of the computer giant from the home marks a rare and embarrassing chapter of failure in IBM's long record of industry dominance."

With a design team led by Bill Synes, but with few other members of the original PC team, the PC Junior was conceived as a genuine home computer, which could also be used for work brought home from the (compatible) office PC; though there was an ongoing debate about how compatible it should be with the ordinary PC (since it was feared it would undermine the business for this).

To complicate matters, while this was happening, Entry Systems Division (ESD) as it was now called though still under Don Estridge, was made the core of a new division; which added on new factories at Austin in Texas, Greenock in Scotland and Wangratta in Australia. But production of the PC Junior itself was (in the fashion of the times) sub-contracted out to Teledyne in Tennessee; and the resulting problems with that company's design and engineering capabilities posed major problems.

It was targeted to be a high volume seller (1.2 million units), in Europe as well as the US. Despite a massive advertising campaign, with $100 million reportedly spent on a wide range of media advertisements - including 160 newspapers and television spots during the world series - coupled with a direct mail campaign to no less than 45 million households, it never met this objective. Though heavily promoted and eventually discounted, it sold just half a million units over its short life. Due to this sales shortfall it was never actually announced in Europe (though it was a mark of IBM's optimism about the product that large stocks were delivered to European warehouses and had to be subsequently destroyed!). It was eventually, in 1985, withdrawn from the US market; an unusually  humiliating experience for IBM.

It was not immediately clear why it failed almost as spectacularly as the ordinary PC succeeded; for the specification was not too dissimilar and the new Dealer Advisory Council (DAC) - whose 'sharp-end' views Estridge perhaps took too seriously - supported it. Even the delay caused by the failure of the outside contractor, Teledyne (which in this case made the whole machine), to meet the quality standards should not have been fatal. On the other hand it did have a relatively poor keyboard (almost a toy, which was annoyingly rubbery rather than positive, but which was eventually replaced), and it was more limited than the PC. It seems that the market‑place demands only the best (the general tendency even then to configure the PC with 512K of memory, when this was not used by most programs probably illustrated the consumers' drive to have nothing but the very "best"). The PC Junior also fell into the black hole that seemed then to lie between home and business use. IBM failed in its move down into this void, and at that time almost everyone else failed in trying to upgrade their home computers into this area. It may also be true, as the commentators have suggested, that PC Junior was deliberately crippled (as traditionally had been most low end IBM products) so that it would not compete with the more profitable business PCs; though Paul Carroll says that this was an internal decision within the group rather than by Armonk. Perhaps the biggest problems of all, however, were caused by the unrealistically high expectations which the media, spurred on by judicious leaks, had about the machine. It could never have lived up to these, and when it failed it was seen as a disaster for IBM; which it probably wasn't. Indeed, the real price to IBM (and especially to its PC Group) was the destruction of its reputation for invincibility; the emperor had been seen without his clothes on, and the audience would watch much more closely in future!

Disastrous Expectations…there is a temptation to over-hype developments, and thus to set up unrealistic expectations. When the offering, even if it is perfectly acceptable, does not meet these expectations it may be seen as a disastrous failure; with potentially catastrophic impacts on the organizations image - not least in terms of its trustworthiness. The story of the boy who cried wolf has a very similar moral.
 

DPD PCs…1983 also saw the US launch of the PC 3270, which was the first "Big Blue" development of the PC. This was the machine inspired by research which showed that for every 1 mip (million instructions per second, a measure of computing power) purchased by 'end‑users' a further 2 mips would be needed centrally (usually on "Big Blue's" large mainframes) to support the additional communications and database requirements. The concept was clear in the design of the PC 3270, particularly as it could maintain communications simultaneously with four separate mainframes; in the process absorbing mips at a rate that 'Big Blue' salesmen dreamt about! This capability is something that ordinary PCs, and many workstations, still could not offer a decade later in the 1990s - evidencing IBM's great expertise in this area. Despite the commentators' assertions, it was even then IBM's marketing expertise which held the company back rather than its research and development capabilities.

Indeed, almost as if to prove the point, a further "Big Blue" development in 1983 was the XT/370. IN this case the ubiquitous PC was enhanced with a complete "S/360", complete with most of the mainframe VM operating system, all on one card! It was a technological tour‑de‑force, but was not an indication of things to come. It was instead intended to be a development tool for programmers in mainframe installations, and targets were accordingly modest.

Portable PC…by 1984 the PC group was reaching maturity. But, again in the full glare of publicity, its first announcement of that year, the Portable PC, was also a failure. More important, it had been beaten to the draw by a start‑up company, Compaq, which announced its own better portable earlier - taking advantage of the drastic shortage of desk-top IBM PCs (for which use, in the absence of any alternative, it was bought). IBM's later offering was heavier (albeit only marginally so - but weight even then was an important measure in this market), and once more was crippled with a poorer built in screen and no hard disc version available. The Compaq was sold at virtually the same price as the IBM (and finally at significantly more, as IBM cut price to sell its stocks). It should be noted though that IBM's portable was still a good product, which I can vouch for - as my first book was written on one. It was only marginally less attractive than the Compaq, yet the market once more voted firmly for the slightly better machine. Again the market's  embarrassing habit of choosing the best product, even if it was so only by a relatively narrow margin, and giving it by far the greatest share of sales was evidenced; as was its willingness to adopt an unwelcome lésé majesté approach to IBM! If the audience had been looking to see if the emperor had any clothes on they were now expecting him to be naked (and IBM thereafter had to earn its reputation on each new machine, just like any normal company!).

The Best Possible…one of the most important lessons, which conflicted with IBM's idea of being the lowest cost producer (fortunately - for IBM never was!), was that consumers want the best - even if it is only marginally better (which was unfortunate for IBM which no longer was producing the best in the PC market!). Although the theory recognises quality, it is usually in the context of a balance with price; and it does not allow for the importance of even small differences.

The IBM PC AT

IBM recovered its nerve, and its market leadership, in August 1984 (just 3 years after the launch of the PC G) with the launch of the PC AT. This was the next generation replacement for the XT; though, by announcing the AT well in advance, IBM had foolishly thrown away the market for the XT long before the AT (eighteen months behind schedule, the last time IBM got away with such a delay!) was ready to replace it. However, when it did arrive, the AT was immediately received to rave reviews as setting the new advanced standard for PCs. It had a much more powerful 80286 processor chip, which could be used to address in excess of 2 megabytes of memory and support as many as three terminals. In practice neither of these facilities was extensively used in its first 2 years of life; it became instead the up‑market replacement for the stand‑alone XT (which however still managed to keep most of its business rather than falling away as planned).

Problems with the PC AT…unfortunately IBM once more ran into significant production problems, this time with the hard disks for the PC AT. For whatever reasons, and were many rumours, the quality was not up to specification and IBM could only ship small quantities for a year; when the AT was already 18 months behind schedule. This meant that there was a whole year for the competition, in particular Compaq, to step in with compatible 'ATs'. As a result, IBM for the first time lost the overwhelming psychological lead that it previously had (though it should be noted that it still maintained, and indeed increased, its market share). In addition, the AT never fully realised its expected potential, due to the hidden snags in its, Intel 80286 chip, architecture. Thus, it was near impossible to switch between 'protected' mode where the real power of the 80286 was available (particularly in terms of addressing large amounts of memory and multiple workstations) and the 'emulate' mode where the AT looked like its predecessor (the XT with an 8088 chip) and could utilise the vast library of software available on that machine. The result was that the AT became, in effect, merely a faster XT; which was a particularly inviting target for the "clone" suppliers to attack. The flawed design of this chip also came to haunt IBM as it struggled to produce its own software, OS/2. This was meant to replace the interloper PC DOS, but IBM promised to deliver it on the AT machine (powered by the 286 chip); and this fatally crippled this software development.

The 'Clamshell' Failure…at the beginning of 1986 IBM experienced yet another failure, when the PC Convertible (the 'clamshell' lap‑top designed to simultaneously replace both the Portable and the PC Junior!) was announced. Once more IBM announced a 'crippled' machine, to avoid competing with other products - and indeed the time taken to cripple it, from the much more powerful specification expected (and which was available in the lab) actually caused a crucial delay in the launch.  Toshiba soon afterwards launched a much more attractive portable (its T3100), and made this 'lightweight' section of the market (later to be called laptops) its own - IBM had lost face against another competitor, this time from Japan.

Cripples…in today's fast-moving world it is doubly foolish to offer anything less than top performance to customers. They will not forgive any organisation which offers them the second-rate, and especially not one which manipulates them in the process. Competitors will gladly exploit the resulting weakness. 'Crippling' an offering is sufficiently rare as to not feature at all in theory.

At the  same time there were upgrades to the XT and AT (both designed to make it more difficult for dealers to pack the boxes with non‑IBM cards). But they still did not meet the expectations of the market for the much trailed arrival of the PS/ 2; the next generation expected, and demanded, less than two years after the last - where previously 'generations' of mainframes had been a least a decade apart! Perhaps more important, they marked IBM's abandonment of its original 8088 based market, where it stopped all development (and effectively handed this to the clones!).

The Myth

By this time the PC (or at least its dramatic success) had become important in IBM's mythology, if not in its cash flows, and DPD (Data Processing Division -  "Big Blue") was becoming more interested in it. The projections then showed between a third and a half of IBM's long term revenue deriving from Third Party (largely PC based) sales. This has since proved to be true - albeit not to IBM's advantage as it then thought it would be! But the PC was a single user system, with very limited mainframe communications facilities. As we have seen, "Big Blue" answered this in part by rushing into its own PC 3270, using an architecture that was unfortunately out of step with the main PC line (and for once in its life even "Big Blue" found itself with a relative failure, having to follow the line of one of its smaller brethren!). Almost all this business, however, remained outside the control of "Big Blue".  Although the target for the PC 3270 was for it to sell more than a third of total PC volume, using large volume sales to big customers, this never happened and "Big Blue" typically achieved only around 10%.

LANs…one important outcome of the PC revolution was the drive for Local Area Networks (LAN's). Previously communications, encapsulated in IBM's SNA (Systems Network Architecture) had been largely hierarchically driven, with communications controlled by the central mainframes; and even the move to DDP (Distributed Data Processing on the minis) had not significantly changed this. Suddenly, though, there were many thousands of personal computers on the periphery with a need to talk to each other without the overhead of going through the mainframes. The result was a shift in emphasis to LAN's, where local communications are on a peer to peer basis between PC's (using specialised communications controllers) with the mainframes only involved when a PC in one of these small networks wants to talk to another network.

In the rush to meet this latest market once more PC group and "Big Blue" were out of step. Thus the PC LAN was a limited capacity "collision detect" system, very much like its competitive predecessors, where the "Big Blue" version was a full high capacity "token ring". To make matters even worse, at first the two LAN's couldn't even talk to each other; though a solution to this was provided in 1986. The 'token ring' approach symbolised many of IBM's problems of the time. It was a very sophisticated, and very powerful, approach to communication. It ran at up to four times the speed of the much simpler - Ethernet - competitive offerings (based on the collision detect approach); which were unfortunately, once more, much cheaper as a result. In addition, IBM offered no meaningful explanation  as to why its own offering was better; which it was, but in ways which were of no use to the customer! The real secret was that the power of the token ring approach was to be essential when 'multi-media' systems came into use; and started pumping pictures and sound (and even video) around the system. IBM had these systems already running its laboratories, but (for whatever reason) they never emerged to justify the need for the token ring. Nearly a decade later they started to emerge, from competitive suppliers, but developments in fibre optics had finally killed - at least in terms of widespread usage - the poor token ring to which IBM never really gave a chance. In 1994 IBM bowed to the inevitable, and it too started to move its support from token ring to Ethernet!

Anarchy

In the mid 1980s, therefore, IBM thus found itself with a new, and very rapidly growing, phenomenon on its hands; and one that showed every symptom of being out of control. In some respects this was inevitable. The original team of 13 had - of course - grown rapidly, but it had also taken over other parts of IBM, most notably manufacturing plants, which were driven by IBM's more normal culture; and, as early as 1983 the number of its employees had passed 10,000. This was no longer a challenge to be solved by the 'intrapreneurship', which to this day still applies inside much of Microsoft. The reaction, albeit delayed rather longer than might have been expected, was predictable. PC group was integrated into the, "Big Blue", business as usual, and its IBU "licence" was withdrawn! Over a couple of years a large proportion of the management team (from Don Estridge, who was replaced by the original product champion Bill Lowe, downwards) was substituted by "Big Blue" appointees. By the end of 1985, at the time of the massive "regionalisation" reorganisation, PC group was firmly entrenched as a division (albeit a disproportionately important one) of "Big Blue". Its new management - following in the image of John Akers who had just taken over as IBM's CEO - was gradually instilling the traditional IBM values and disciplines; so that, for example it no longer made the mistake of single sourcing on manufacturing or development. They had, whilst reassuring its then 2,000 authorized retailers that they were still an integral part of IBM's product marketing strategy, even announced that they were to tighten up the quality of the dealer network; though, as was described earlier, they never made any headway on this front. In the US to try to enforce this they even used the equivalent of a headcount freeze, IBM's traditional internal answer; they froze the dealer network, not allowing any new additions, and still failed to control it!

The dramatically rapid growth of the PC was an outstanding success, but it left IBM with a number of problems; largely because the pressure of events demanded such fast responses that long term planning was not always possible. As I have already mentioned, the first of these was the "Open Architecture"; with consequential vulnerability to, and reliance on, outside suppliers. This concept had a great many advantages, ranging from faster market growth (which was always IBM's goal) to competitive position (where IBM as the lowest cost producer, as it thought - before the smaller Asian producers undercut it - was supposed to be well placed). One suspects, though, that had IBM planned the strategy in its usual meticulous manner there would have been a significantly greater degree of IBM control written in. The clear evidence, from the  PS/2 range - and even from the RISC based PC RT - for example, is that IBM was gradually attempting to reassert some greater degree of control. At the time, I always added the footnote "always assuming it can get its own act together!", which was, in retrospect, an especially important qualification of my (then) optimism; since that was exactly what IBM never did!

PC Price Wars…a bigger problem was that of the price war. As the market became less buoyant (largely due to insecurity posed by the constant rumours, which started at the beginning of 1985, of the imminent availability of PS/2), the dealers, which IBM had still failed to control in any meaningful way, started a price war that helped nobody, least of all IBM. In part this was due to the shake‑out of those dealers who couldn't make the grade, and - with cash-flow problems - started to liquidate their stocks. Partly it was due to the more aggressive box shifting attitude of the PC group management; who did not recognise that end of life promotional discounts would destabilise the market even more. In the main though it was probably due to IBM's discounting structure which had price breaks that encouraged dealers to sell the last few machines at a heavy discount; so that they could reach the next price break and obtain a better price overall on their purchases.

Many commentators have claimed that the decline of IBM was brought about by external forces which it could not handle. I will repeat, once more, that this is not true. IBM had survived many worse changes in its environment. The problem was that IBM itself created the changes in the environment which destroyed its position, changes which it could not then handle! The dealer price war was one particularly important example of this - especially in terms of its spread to the rest of IBM, and its eventual undermining of its margin - to unprofitable levels - across the whole of IBM. The war was started by the incompetent handling of the PS/2 launch, and was exacerbated by IBM's lack of control over its dealers; indeed, it was fuelled by strong encouragement from IBM's PC Group management who wanted the extra short-term sales this would bring. IBM scored, yet again, an own goal - and one which was especially important to its long-term future!

At the time it might have looked on paper as if it should have maximised IBM business; loading in stock to create stock pressure on retailers is a standard consumer goods marketing technique. In fact this was true of the system units but not of anything else. At that time, the end buyers did want to see the IBM badge on the front; though increasingly retailers were persuading them that Compaq, which was not so heavily discounted in the price war (and hence more profitable to the retailer), was a better box! The retailers, on the other hand, wanted to see everyone else's cards inside it, and peripherals attached, since these competitors offered better margins; and the customers were often unaware of the differences. The result was that, even then, IBM was suffering - by shipping a number of relatively "empty" boxes; a fact which mightily offended  it.

Price Wars…there always is a temptation, especially if you are the lowest cost producer, of starting a price war to take advantage of an expanding market. There is only one piece of advice, don't! Don't ever do it, under any circumstances! Almost all the markets that have experienced price wars have seen dramatically reduced profit levels across the board; and often shake-outs of significant numbers of companies, frequently including the one starting the war! Theory, especially economic theory, sees price competition as being the rule. Fortunately it isn't!

Quality of Support

Perhaps the worst effect of the price war was that it finally shifted the emphasis from quality of support to price. IBM's extensive market research at the time clearly showed that the buyers wanted high quality support above all else, and were far less interested in marginal cost savings. This was an eminently sensible attitude, where a poorly designed and supported system could cost a company many thousands of pounds in extra costs; and the saving even on the highest discounts ran only into hundreds of pounds. This 'support' was justifiably IBM's strength in the other markets it operated in; and the PC market, with its relatively unsophisticated users, was in greater not lesser need of it.

Unfortunately after visiting a number of PC dealers who stressed the importance of discounted prices, and could barely suppress a smile when 'support' was mentioned, the user was - to say the least - confused. It was not surprising that PC dealer salesmen were rapidly ousting used car salesmen as the symbol of poor, blatantly dishonest, salesmanship. It could not have been further from the traditional IBM style of salesmanship! It was bad for the image of the PC industry, and it costs customers millions of dollars in wasted assets.

This was not necessary. It was a wound self inflicted by the dealers - with, as we have seen, a great deal of help from IBM. Those "respectable" dealers that I talked to at the time stressed that they were desperately aware that their customers needed the support, and the lack of this was holding sales back. Despite this they continued to discount on larger orders, they claimed by necessity, and the essential support could not be offered. The less reputable 'box shifters' staggered from discount to ever larger discount, and closer to bankruptcy; but they had nothing else to offer. It was definitely not the IBM style! Perhaps the greatest short-term damage that was done to IBM was that it legitimised 'cheap' machines. This in turn let in large numbers of "PC Clones", virtually identical copies of IBM hardware, to the major markets. At the time these were sometimes of questionable quality and copyright; but, with practice, they came to be of just as high quality as the machines of IBM - and often had better features. Of course, they were also much cheaper! IBM rapidly lost a significant share of the PC business to these clones, typically produced in Taiwan and Korea (not Japan as IBM originally feared) by very low priced labour, selling then at perhaps half of IBM's price. If, as the price war apparently demonstrated, price was all that mattered then such clones were just as legitimate as the IBM originals. John Akers at the time even went as far as to foolishly legitimize the position, by publicly stating he thought the PC was probably a commodity,

At the time, once again, I said "Low price is a simplistic story (for there a many more factors involved), but IBM will not find it easy to counter. On the other hand there is every indication that IBM is trying very hard to bring its unruly infants back into line with the more normal IBM philosophies. All those in the industry, and its customers, must hope it succeeds." Of course we now know that it didn't succeed; and the rest is, as they say, history!

Price Wars…to reiterate the point, nobody wins a price war. Only fools start them. Unfortunately, there are plenty of fools in senior management positions!

Marginal Problems…at least PC group had, at long last, got its production act together. By then it had some of the most automated plants in the world supplying it with the highest quality (zero defect) products, and the previous supply problems were a thing of the past. It did though still have a profit (margin) problem; which was there even before the savage price cuts forced by the competition from the 'clones'. It had a great success in volume terms, but its own internal PC margins were undoubtedly less than the rest of the business. They almost certainly were, even then, as low as half those in the other groups (perhaps as low as 30% gross); though the marketing and admin costs may also have been much lower (perhaps below 10%). It was for this reason that, in the first instance, the World Trade subsidiaries were set up as separate companies; largely to avoid the 10% royalty on product cost that normally was paid back to the US.

Volume and Margin…it is not sufficient for a company to achieve volume sales, it must also obtain a workable margin on these. This should be obvious, but it is too easily forgotten when the pressure is on. Management theory is very clear on this point; but it rarely explains just how easy it is to be distracted by events.

It would appear that these relatively very low (in IBM terms) margins may have caused IBM to fundamentally rethink its strategy. In mid 1986 even John Akers started to talk about the possibility of IBM abandoning parts of the business "where it had become a commodity market" (i.e. very low priced), and IBM froze its advertising budgets as part of a drive to cut costs. Of course, IBM never did leave the market. It probably would have been better had it done so, for the evidence is that it has never made any profit out of the business, and the effect on the rest of IBM (in terms of image and the extension of the price-cutting war) was disastrous. This was one of the first of John Aker's acts of indecision which were to cost IBM so dear.

Almost as disastrous for the PCs competitive position, and for IBM's image, was that it started to lose the product development race. The nature of this perversely switched to the power, evidenced by the generation - 286, 386 etc- of the chip inside the machine. This was, of course, one of the gifts which IBM had made to a supplier which then became in effect one of its major competitors!

Intel

Unlike some of the other beneficiaries of IBM's PC developments, however, the origins of Intel lay indirectly some two decades earlier. Robert Norton Noyce (then working at Fairchild) invented integrated circuits. He subsequently formed his own company, Intel, in 1968. The basis of the company fortunes rested then, and now, on the concept of putting a computer, or at least its CPU (Central Processor Unit) onto a single chip - originally as a way of generalising the circuitry of devices such as calculators (since the chip could be changed to do a different job, in a different device, simply by reprogramming it rather than the long process of making new circuits) and then - as everyone now knows - to provide the power for the new PCs. It was Intel's invention which really made the PC age possible. Even so, it was IBM's entry into the PC market which really made took the Intel into the big league.

What was then seen as a miraculous device, the 8088 processor, now seems puny; as the range of chips has been incrementally upgraded through the 286, the 386, the 486, the Pentium (renamed because, a decade or more after the escalation started,  Intel found it couldn't copyright a number!) and beyond. Intel plans to boost the power of the chip with a new generation every two years - even by the time of the launch of the Pentium there were already 3 million transistors on a chip, which had the power to process 100 million instructions per second (100 MIPS - as much as a mainframe supercomputer of less than a decade before); now, with clock speeds running into several thousand million instructions per second on even cheap PCs, it seems as it the sky is the limit - though it is difficult to see what (other than video which may be the salvation of the chip manufacturers) needs anything like this!.

Considering the degree to which Intel is now a major competitor, it supplies most of the chips which go into PCs (most of which are no longer assembled by IBM) and heavily advertises the Intel chip (so that PCs are no longer seen as IBM-compatible, but as Intel based), IBM's attitude has been very strange - almost acting as its godfather. Unlike Microsoft, where similar initial support created a multi-billion dollar corporation - but relations then deteriorated (albeit taking a decade to reach all out warfare), Intel has never been really challenged by IBM. Indeed, when Intel ran into a financial crisis in 1983, when it could not pay for the rapid growth it was experiencing, IBM stepped into buy a 20% stake, and thus fund its further development (as well as to cover the losses Intel was then making). This IBM support was consolidated by a technology agreement between the two companies in 1986. IBM sold the stake, as soon as it could get its money back; a mistake, when - in the 1990s - that 20% stake would have been worth more than the whole of IBM!

This position was doubly surprising, since IBM then had by far the world's greatest expertise in chip technology. It was able make chips, in its laboratories, which were probably an order of magnitude more complex than those of Intel and had invented the RISC (Reduced Instruction Set Computer) technology which potentially made chips run many times faster still; technology which had cost IBM billions of dollars to develop. It was then itself probably the world's largest chip maker (albeit for its own machines), yet it still treated Intel (80% of whose production was supporting IBM's competitors) as a favoured godchild - even as the company beat IBM hands-down in the 'chip war'. According to Robert Heller, in 'The Fate of IBM', over the five years after IBM's RISC chips were finally launched on the world they sold only 300,000 chips to Intel's 20 million!

Paul Carroll believes it was Bill Lowe's indecision that gave Intel its dominant position - much as it did for Microsoft. The turning point certainly seems to have been Intel's launch of the 386 chip. Before that point IBM probably had the upper hand, after it (and especially where Compaq and others had beaten IBM to the launch of machines based on this chip) Intel was in control. As we have already seen, IBM had misjudged its position in relation to the market; and had lost its position as technological leader. It had also misjudged its technological capabilities. It had assumed that, with its cross-licence agreements on Intel's chips, it would be able to later produce a modified, and better, chip which would make those of its competitors which had already launched 'IBM-incompatible 'obsolete. Unfortunately, despite its great technical capabilities in the field, it never did this - and conceded the standards to Intel. Thus, the 386 - which became an excellent workhorse of the PC industry, and sired just as successful follow-ons (in the form of the 486 etc) - effectively sidelined IBM as the hardware standards setter in the PC field. Yet, despite his rages at almost everyone else in sight, John Akers still seemed to have this godfatherly approach to the corporation IBM was trying to beat!

Beware Those Accepting Gifts…it may seem a civilised gesture to help other organisations, and where they do not seem to be competitors it may even seem to be a potentially profitable one, but be very, very sure that you are not handing over your birthright. The Japanese in the 1950s bought a whole range of licenses very cheaply, because nobody in the West ever saw them as being important actors. Of course in the 1970s they used these licenses to destroy the originators!

By the 1990s, therefore, Intel had set the standard for PCs in a way that IBM had failed to do - enjoying the same sort of dominant position (something like two thirds of the chip market) as IBM had in mainframes earlier. IBM's position was, however, still equivocal. It entered into a new agreement with Intel in 1991, but by 1994 it was also working with Intel's clone-producing competitor, CYRIX. Above all, it had earlier been reduced to seeking - yet another - a partnership; this time with its rival Apple and Motorola (which supplied Apple's chip - the only one which was not 'Intel - compatible') to try and develop the RISC approach - seemingly to dent Intel's dominance. In 1994 this eventually bore fruit in the form of the PowerPC chip, which was - in another strange addition to the history of IBM's joint ventures - massively promoted by Apple as a vehicle for IBM users to switch to its PC offering! Surprisingly, though, IBM - in another equivocal gesture - reported that it would itself remain with Intel designs for the PC; reserving the PowerPC for its workstations and low-end mainframes!

Compaq

Returning to the mid 1980s, however, the initial challenger was Compaq; which was in many respects what IBM PC group should have been - had PC Group followed the IBM's three philosophies like the rest of the corporation - and, indeed, two of Compaq's key marketing executives  (Jim D'Arezzo and Sparky Sparks) had moved there from IBM's PC Group. After its initial, and unexpected, success with its portable(the market for which IBM abandoned to it) it had used its new found expertise to enter the main desk-top market; and was able to use the delay in IBM's PC AT to establish its own better (faster) version of the 286 machine alongside that of the giant - and became a serious (albeit then much smaller) competitor. The biggest mistake by IBM came in 1986 when the 386 was announced by Intel (a company in which IBM then had a major stake), at a time when IBM was busy developing yet another lemon; the XT 286, a cut-price AT in an XT box which couldn't even take the standard AT cards! Compaq immediately set about producing a machine (the Deskpro 386) based on this, while IBM - preoccupied with its own attempts to wrest the market back again (with the PS/2 and OS/2 - which paradoxically were themselves were largely doomed by IBM's blindness to the potential of the 386) - missed the opportunity. It had long followed a policy of being late into markets, so that its competitors would sort out the problems (and, in any case, it thought - mistakenly as it so happened - that Intel would have as many problems with its 386 as it did with the 286). IBM was wrong, Intel's deliveries went smoothly.

The IBM 386 machine eventually reached the market seven months later than Compaq. For the first time IBM paid the price for its traditionally conservative approach. It did not just miss the sales, it lost the market position (by the time it arrived Compaq was the 386 supplier of choice). Even the (loyal) IBM dealership, which I owned at the time, was selling four times as many Compaqs as IBM to the more sophisticated users in which we specialised. In addition, Compaq was not competing with its dealers, where IBM had its own competitive salespeople selling to the large customers which the dealers dearly wanted to enter. Compaq allowed them to do just that - to IBM's detriment. Above all IBM lost its image of technical leadership. It had been made to look a fool by a small start-up company. David had beaten Goliath; and the media loved it!

If your Role Model is Goliath, you had Better Win…everyone loves a winner against the odds, including customers and - especially - the media. It is one force which can overturn markets. So, in the first place try not to be the unlovable monster. Then, whatever your image, make certain you do not lose crucial contests - and particularly the highly visible ones - to your smaller rivals. Management theory has still to learn the psychological aspects of warfare.

PS/2 and the Micro Channel

It seems clear that, by the time of the move to the 386 generation, IBM had realised that it was vulnerable to attack from the clones - and had, due to the gifts it had made to Intel and Microsoft, no defence against these. Its reaction, therefore, was to try to reverse the damage it had caused itself by so enthusiastically embracing an 'open systems' approach. Accordingly, it attempted to set a new standard which it alone would control. This control of standards was an approach which had long been successful in maintaining its dominance in the mainframe market. It is not clear why, in this case, it did not do this via a different version of the 386 chip (which it reportedly had available at the design stage);  but in any case IBM was used to having the market follow it almost wherever it went. IBM's choice, therefore, was 'Micro Channel Architecture'. This was technically a step forward, since its performance was significantly better than the standard (then PC AT based) approach; and, best of all for IBM, it would not accept all the cards which other manufacturers had been putting inside the previous versions of the PC! On the other hand, to most PC users (who had only just begun to understand what a 386 chip was), it was just one more, rather esoteric, element inside the machine - which seemed to have no relevant benefits for them. In reality,  despite Paul Carroll's view that it was little more than a gimmick, this new 'bus' (connecting the various parts of the system together) was technically important. It was designed to allow much more sophisticated distributed processing within the PC. In part this would be for connection to sophisticated (token ring) networks, which would be distributing vision and sound as well as the existing data, and for the powerful (and specialised) processors which would be needed to handle this torrent of new information. Unfortunately, as we saw with the token ring, IBM never completed development of  this much more sophisticated usage for the PC.

What was worse, was that IBM did not even explain what it intended to do. It was used to its mainframe customers simply accepting a new approach simply because IBM said it would be good for them; and then, later, being astounded when it turned out to have revolutionary implications. IBM, accordingly, announced Micro Channel Architecture as an answer to problems caused by radio-frequency emissions! Unsurprisingly, the PC customers found this not just to be unconvincing, but to be a remarkably boring justification for the totally new form of computing, which IBM was claiming for its new range of 'systems'. I remember attending IBM's high key launch to dealers of the PS/2 - as I was then the CEO of one of IBM's larger PC dealerships - and leaving it, alongside other dealers who were enthused by its professionalism, thinking to myself 'what will they think tomorrow, when they try to work out what benefits it offers to their customers'. The contrast with Compaq's launch of the 386 chip, which IBM did not even feature in the launch of its PS/2 line,  could not have been more complete. Compaq met what customers wanted. IBM met what it thought customers ought to need - at some time in the future. It was not surprising that the PS/2 was yet another relative failure; and IBM's share of the PC market fell by more than twenty points over the next couple of years or so - to languish at around the same level as Compaq and Apple.

Although for once Europe did not follow suit, Bill Lowe compounded the problem in the US by immediately by killing off the AT, to make clear just how much IBM was staking its future on the PS/2 range. It was a gamble, though I doubt that he realised that, and one which IBM once more lost! The rest of the industry was now powerful enough, and the existing standards widely enough adopted, to ignore IBM's attempt to impose its will. Yet again, David (or this time a tribe of many David's) defeated Goliath - to the delight of everyone (especially as the PS/2 seemed to be a remarkably silly move).

Standards are Ultimately Set by Customers…manufacturers may initially suggest standards, but it is the customers who will eventually decide - by their buying patterns - whether they agree! Using a (proprietary) standard as a means of locking in customers is a very powerful marketing tool (creating a quasi-monopoly - as Microsoft has amply demonstrated since - but only if the customers believe it is worth sacrificing their freedom for it. Though it recognizes the value of patents, management theory has little, and economic theory even less, to say about the use of (proprietary) standards to create quasi-monopolies.

Customers, with commendable logic it turned out, wanted either the power of the new 386 PCs or the simplicity (and cheapness) of the old ATs - and IBM could offer neither! In one master stroke it had effectively cut itself off from the market. It is understandable, perhaps, that IBM should have pushed the Micro Channel Architecture; though it should have thought of a better reason, to offer to its customers, for doing so. It is a mystery why it cut off the other routes - and did not also offer the 386 machines everyone was clamouring for, and did not even keep its workhorse PC AT! Once again, Bill Lowe's judgement (presumably backed by Armonk) was fatally flawed; though in this case one could hardly criticise IBM for indecision.

Burnt Bridges…if you must take risks, try not to make them suicidal by cutting off all the escape routes yourself! Military theory might suggest that it is foolish to offer your front-line troops too obvious an escape route, since they might be tempted to make early use of it ; but the possible annihilation of your troops, offered no escape, is one outcome of business conflict which is not allowed for by any theory.

The PS/1

The announcement of the PS/1 in 1990 was intended to make this IBM's home computer. Perhaps needless to say, once more it did not set the world on fire; which was not surprising where, once again, it was crippled (not least by using the 286 chip when almost everyone else had moved on to the 386, or beyond, even for their lower price machines). It did not bomb in quite the way that the PCJunior did, but that might have been because - despite its crippling - it could be seen as a low price business machine (and was often sold - heavily discounted - by dealers as IBM's offering in this range).

The PC Laptop…for once IBM managed to get its development timescales in line with the outside world, for the first time since the development of the original PC, with the PC Laptop; which was announced in March 1991, reportedly after only 14 months of development (compared with IBM's more normal lag of two to three years). According to Robert Heller (in 'The Fate of IBM'), this was because IBM had adopted development processes which were Japanese in style:

"1. Relying on proven technology.

2. Designing and assembling the machines, save some of the chips, from components supplied by other firms.

3. Abandoning the traditional insistence that every detail of the specification be worked out and agreed before production began.

4. Bringing in some manufacturing and marketing people at the start of development.

5. Speeding up negotiations with outside suppliers.

6. Using a very small team - a mere nineteen people in Boca Raton, Florida, worked on the laptop [compared, according to the Wall Street Journal, with teams ten times as large in the rest of IBM]

7. Refusing to allow changes to specification once set.

8. Listening to customers when designing the product."

This approach certainly is much more productive, or at least allows much shorter development times, where the development is incremental; as the Japanese, whose development processes are typically incremental in character (and accordingly relatively low risk in nature) have so clearly demonstrated - as also have the PC clone suppliers. Heller believed that this would resolve the crucial dilemma of long R&D cycles which both he and Paul Carroll ('Big Blues') believed were at the heart of IBM's performance problems. What he did not address, though, was the problem of the radical breakthrough (one which genuinely leads to revolutionary consequences, as the 360 - and even the original PC - did), which Western R&D (including IBM) focuses on. This is much more risky than incremental development. The whole incremental progression from Intel's 8086 to its (80)486 and Pentium chips was, despite the great technical achievements in production technology they represented, predictable; and hardly put Intel's future at risk. The IBM 360, on the other hand, gambled everything. Paradoxically, it was IBM's failure on one hand to produce another 360, and its long R&D cycle times on the other, which were being criticised as if they were one problem - when in fact they were opposites!

The laptop was a technical success, not least because it showed IBM could meet the shorter development times needed. On the other hand, it was not a great commercial success. IBM's planners insisted it be brought in at a high price to 'skim' some initial profit (even though the Japanese had long since shown that it was better, in a very competitive market such as that of PCs, to enter with a low price to achieve 'penetration' and keep everyone else out). The subsequent price cuts which IBM had to make, to match competition were seen as a weakness - which was not helped by IBM's financial collapse in 1991. The death knell, however, was handed out by Compaq which brought out its first 'notebook' computer - and effectively moved the fight on by a generation! IBM has, though, recovered since with the launch of its successful series of 'Thinkpad' notebooks - which have been as technologically advanced as any of its competitors' models.

Penetration Not Skimming…new products were traditionally launched at high prices, to 'skim' off profit from their short-term monopoly position. Now, they are launched at low prices, to guarantee 'penetration'. This is supposed to create volume sales, with their accompanying economies of scale, most rapidly. It almost certainly protects the very valuable position as brand leader in the new market - and, as such, is the approach to be recommended regardless of any economies of scale (though the price must not, in this case, be unrealistically low). Management theory does talk about the decision between skimming and penetration; but, in many markets, there is no decision - you must penetrate.

OS/2

As far back as 1983, IBM had started to develop an interface which would allow PC users to have access to more than one program at once, by dividing the screen into various windows. Perhaps as a foretaste of what was to come, when it was launched in 1984 (as TopView) it was a resounding flop; though it does belie the idea that IBM was late on all its products (the evidence is that this one might even have been ahead of its time). Bill Gates had started his own development earlier when he had seen an early version of the Apple Macintosh - but did not announce it until 1983 and (beset by so many problems that it was about this offering that the derogatory term 'vaporware' was reportedly first used) it was not delivered until 1985. Even then it was something of a dog - and was not bought by many PC users. PC DOS remained the standard, however, and Bill Gates continued to rake in the money. The problem for IBM was the conflict which was building up - and would come to dominate the next half decade - as to whether Microsoft's 'Windows' or IBM's 'OS/2' should be the strategic product; where, at least in theory, the two were partners in development.

The concept behind IBM's own OS/2 was eventually conceived, after many false starts in 1984 - though, with a typical lack of awareness of the market, it did not have any graphical interface. It was still to be a joint development, despite the growing IBM antagonism to what it saw (correctly  as it eventually turned out) Microsoft's commitment to competitive work on its own 'Windows' product; the first version of which was delivered in 1985 - though it was desperately slow and deservedly unpopular. The real problem, which dragged out its own launch for what seemed like an eternity, seems to have been IBM's own ham-fisted development effort which eventually wound up with 1,700 programmers working under two management structures on four sites across two continents. I was involved, for a time, with the team working at IBM's Hursley laboratory in England (though to add to the confusion many of them were actually working in an office building twenty miles away!). In my experience this group did not seem to be aware of any dramatic problems, but - as the later record shows - it took the whole team nearly a decade to bring out a genuinely workable product; so presumably something must have gone wrong. This has to be a record even for IBM. The comparable teams on the 360 software and even the later S/38, which was based on the Future Series and had to be a whole level of complexity higher than the PC, took no more than two or three years to bring their final products to market. To take nearly a decade requires