Wang Laboratories Inc looms as an awful warning to the people charged with finding a successor to John Akers at the helm of IBM Corp. After four years of grinding attrition and blood-letting, the company is in Chapter 11 bankruptcy protection, shrunk to a husk by the forces that are now blowing like a whirlwind through IBM. If the new team at IBM gets it wrong, on the Wang model, in about three years’ time, IBM’s 300,000-strong workforce will have shrunk to 80,000 and IBM will be a $16,000m-a-year company. Almost all the talk has been of one man – The Saviour who will come in with fresh, uncluttered thinking and wave a magic wand over the company to put right everything that is wrong. There must be people left within IBM many of the brightest and best have already gone in one or other of the successive ways of voluntary redundancies, but this time around, its doubtful that an insider could get his way: the entrenched IBM bureaucracy would defeat him. There is so much that has always been done that way that insiders will inevitably miss many of the things that an outsider would instantly recognise as ridiculous. We related how the response to the question how do you pick up the text of announcements on IBMlink in a continuous transmission rather than a screen at a time was What mainframe are you using, but that is only part of the problem – that while IBMers sell the stuff, most of them never actually use personal computers – or even AS/400s.

Superseded

Another aspect of the IBMlink system highlights the fundamental difficulty. It is a very trivial example – until you realise that it must be replicated over and over again in all IBM’s operations. Last week, IBM launched five new PS/ValuePoint 6384 models, and the text describing the machines took up almost 20K-bytes on IBMlink – for a product that will have a currency of perhaps six months before it is superseded. Someone had to brief the person that wrote that text, which bears no relationship whatsoever to the text in the press releases or the glossy flyers that will be prepared to promote the machines. Someone had to write that 20Kb of text (an eight-page issue of Computergram is about 60Kb of text), endless other people had to check the text, sign it off, pass it up for approval, before it could go up on the system. That might have been acceptable when the product being described was a mainframe carrying a 60% gross margin that would remain a live item for two or three years; it is totally absurd for a personal computer that costs $2,000, yields gross margins of about 1% and will be stone dead in a year. But if there was not an army of IBMers prepared to defend that way of doing that to the death, that 20Kb of clumsy verbiage on a commodity personal computer would have been reduced to 20 lines five year ago.

Lady Thatcher

There is also a general assumption – not incidentally one shared by that wise old bird for the chip industry, Wilf Corrigan, that one man is what IBM is seeking (we’d say one man or woman but we don’t believe that Lady Thatcher would take the job). In truth, the new man has to bring in his own people in most of the key positions, so that the first two tiers of management are sprinkled with outsiders, many of whom have had to survive trying to compete with IBM in the glory days. It has been suggested that someone from the semiconductor industry would be a good choice for the top job: we concur, but if he’s really good, he would be wasted in the top job – he should be running IBM’s chip business. Until the latest round of closures, maybe even after it, IBM was the biggest chipmaker in the world – and jealously kept all its chips to itself. Either IBM is going to have to write off half its chip fabrication capacity, which would be a criminal waste, or it gets out and makes the life of every other chip manufacturer a misery because it makes every type of part better and more cheaply, and gets it to market more quickly. Yet far from being a force in the semiconductor market, IBM’s efforts to sell its chips have been pathetic. Why not? There’

s nobody in IBM that has ever had to sell a chip in his life, and someone with long experience of competing successfully with the Japanese manufacturers is urgently needed to run IBM’s semiconductor operations if one of the company’s few enduring assets is not to be shamefully squandered. Should IBM continue to own its chip business? Shouldn’t it spin it off as a stand-alone company? One of the most charming sophistries that IBM top brass has been allowed to get away with is that its new semi-autonomous business units have to have three years of audited results before it can sell shares in them to the public. What rot! That is only the case if IBM wants to maintain majority control. It didn’t prevent the company divesting Lexmark International Inc and retaining 10%, but there are plenty of other approaches that could be taken, the simplest being to take control of a shell company whose shares are quoted – Computer Automation Inc for example, and sell the assets of the business to be divested to the shell in exchange for new shares in it.

Oblivion

The enlarged Computer Automation would then be free to offer a large tranche of those shares to the public. In the present climate, IBM’s semi-autonomous businesses do have to be run by people directly answerable to outside shareholders: in the present climate, if IBM simply structured them as wholly-owned subsdiaries, no-one would trust it to run them at maximum efficiency. IBM’s other recognisable assets that are not obviously wasting to oblivion are the disk drive business which needs an Alan Shugart or a Finis Conner to run it – and the AS/400, which has been so incompetently run that it is teetering on the brink of a slide that will see it following the Wang VS into the trash can. The failure to win over full half the System/36 base has left the line dangerously – and avoidably – exposed, and far more imagination and iconoclasm is needed to rescue it than anyone within IBM has shown. Major new markets outside IBM have to be found for variants of the AS/400, the most promising approach being to produce a variant of the machine configured as a sealed box database server sufficiently attractive that even those that vote the Unix ticket or are wedded to nothing more than personal computers and NetWare, will want it. And the software must be priced in such a way that it looks a bargain when the user signs the first cheque, not merely after laborious – and probably spurious five-year cost-of-ownwership mathematics have been absorbed. Either the Nomination and Executive Compensation Committee conducting the search for a new chief executive takes all that on board, or their place in history will be that of the people that turned IBM into a second Wang.