There is an old saying that "better the Devil you know". It simply means that even though you are aware that something may be imperfect, it is best to make do with it than to experiment with something else which promises to be better, but isn’t worth taking a risk on.
In the IT industry this used to manifest itself as "you never got fired for choosing IBM" and now it has become "we can put up with the unreliability and security problems of Windows". Things don’t change that much, do they ? In retrospect the IBM stance lead to serious problems and Microsoft are obviously trying to change direction towards the Internet in order to divert their own problems.
IBM got into trouble because the IT world changed and by-passed them while they were defending their established empire. In particular prices were driven down and the life span of products reduced by technology advances, leading to new mass marketing strategies, so expertly exploited by Microsoft. IBM also failed to cope with the increased range of computer applications and continued to focus on their established strengths in transaction processing. In so doing they failed to recognise the growth of more flexible, lower cost communications networking. The obvious higher robustness of SNA came with a high price tag and was hopeless for use in simpler applications like office local area networks, let alone futuristic systems such as the Internet. So while TCP/IP was not as reliable for transactions as SNA, it proved a much better bet overall and IBM lost to Cisco, despite their earlier dominance. There is a moral here!
Office applications not so important as transaction systems (we can’t do without data processing, but the value of office automation is debatable) but the situation is far worse because Microsoft has a monopoly; in contrast while IBM dominated transaction processing they had plenty of competition. IBM were forced to restructure, successfully as it happens, thanks to excellent management, by the growth of alternative technologies, first Digital, but then HP and Sun; as an aside only IBM are left in the mainframe arena, just as the Internet is driving the model back to centralised computing and mainframes!
With IBM now a force in the Unix, Intel and OSS camps, as well as mainframes, they are a very different company today. At the moment however Microsoft are in a similar position to that that nearly caused IBM’s downfall, although they seem more aware than IBM were of impending problems and are making efforts in the server, Internet and domestic (games) markets to diversify their source of income. So far this has had little impact compared to their dominance of the PC market and there is no guarantee that the new markets will generate profits to go with the turnover because they are unlikely to reach another monopoly situation.
It is thus a good time to look to other factors which can hold market share, if not grow it. Brand loyalty is a significant factor. One of the key reasons behind IBM’s restructuring success was the simple fact that it was a very complex exercise to replace any transaction system with another one. The impact on disturbance of the business was far more significant than any technology consideration. In practice the Unix and now Windows systems came in with new applications in the beginning, rather than replacing existing systems. The same thing is going to help Microsoft. It amazes me that after the trap created by Microsoft Office, IT departments allowed Microsoft to penetrate the server market with products that created a similar tie-in to the office systems. The NT servers of course have nothing like the dominance that their partners have in desktop systems. It would be interesting to know just what percentage Microsoft holds of the total Server market, since one suspects that published figures count one PC server and one mainframe as the same, and clearly Sun, Compaq and HP are far more dominant and competitive with IBM. In the Internet related servers, Linux has already made big dents into NT markets and continues to grow. Nevertheless Microsoft with NT and the servers such as SQL Server, IIS, Exchange, etc. and their proprietary programming tools have a big enough presence to have a huge following of dedicated users, just as do IBM, HP and Sun, not to forget Oracle. In conclusion some things will change but only slowly because "better the Devil you know"!