Ballmer muses on opening up NT source code

Microsoft President Steve Ballmer suggested during a speech to CIOs and information industry executives yesterday that Microsoft might release some NT source code. His comments came when a questioner managed to slip one in about Linux. This is by no means the first time Ballmer has mused out loud about open source, without making any specific commitments, but he had quite a lot to say. Clearly he's been turning the matter over a lot, if he can spiel at this length in response to a simple question. So we'll look at the full text of his response, lest any mythology develops: QUESTIONER: "Could you please comment on Linux and the development model, and how it affects Microsoft strategy?" BALLMER: "Well, the Linux development model is interesting. I think we all got a chance to learn and study. Today, you know, as we assess Linux we see a reasonable amount of interest, more interest than I guess I would wish. I mean, we're competitive guys. More interest than I'd wish, but what it's caused us to do is really focus in and ask what is it about the Linux model that really rivets people. Initially some people thought it was the price, and I frankly don't think that's the case. In almost every application that we talk to people about, people want to have a good price, but the most important thing is that you get a platform that does the job, that is reliable, that does what is needed to get done. "Some people think it's the development model. Does the development model let Linux run more quickly, because you have a bunch of people throwing code at the thing? And that's been a little bit overstated, but only slightly. "I think there are some things that you get with that development model. You certainly get more people trying to add value at one time. That may mean some good work happens. But because of the openness [yes, he really did say "openness"] of the Windows system, there are plenty of opportunities for people to add to the Windows operating system today through their own dynamic link libraries and device drivers, et cetera. The place where I think we have something we must learn from Linux and must respond to is in the area of, excuse the word, open source, because it means different things to different people, but certainly the notion that there are parts of our source code that if we published, on some basis, whether we put it up on our Web site or licensed it to the kinds of customers in this room, would help you be more effective in your job, it would help you get applications built more quickly, it might help you support the deployment and management of applications that have been deployed in your environment. We see that fairly clearly as a requirement being expressed essentially out of people's interest in Linux. What is it that customers want? "And we're trying to figure out exactly what it means, and exactly what kind of licensing or no licensing, and which pieces of the source code. I don't think everybody really wants to sort of dig through the code that puts up menus, but there are parts of the system where if you had the source code I think people would feel they could be more effective. I'll pick on the database connectivity pieces for a second, because I think a lot of people who work in that area have found them complicated to understand. And so we do have the team off thinking through what kind of a strategy is appropriate to make our source code or parts of our source code more available to customers so they can be more effective at what they do. "I won't call that a full embrace of the open source model. On the other hand, we're trying to understand what it is that really brings the benefits. There are a lot of downsides with the open source model as well. There's a bit of chaos. You have four standards distribution. Things can seem like they're moving fast, but they don't necessarily respond to any given customer needs. People who bring it in almost all wind up having to have system hackers who want to go hack the systems. Many of you would prefer to have your people more productively deployed on the business needs. There's no way to guarantee response on fixes. "So there are some downsides to the Linux model, as well. We're trying to learn from the things that represent, I think, clear upsides and where change in our approach will benefit you." So there we have it. Microsoft recognises that it is challenged to open its NT source code. Whether this means that Microsoft would be prepared to publish some of the code as a way of stopping Linux remains to be seen. It is perfectly clear that Microsoft's desktop OS monopoly cannot now be broken by a rival product without some draconian legal handicapping, but the server space is quite a different picture. If Microsoft could really switch much of the Linux energy into picking over the Windows carcass, it might preserve and strengthen its position in the server market. Ballmer's Linux comment came after a sales pitch to the Tibco strategic directions conference yesterday. Tibco is the outfit that digitised Wall Street in the 1980s and made rather a lot of money in the process. Miscellaneous useless fact: TIB stands for The Information Bus. The only thing of interest in his main presentation was a reinvention of question parsing in SQL Server so that plain English (well, American actually) questions were turned into SQL (the language, not the product, that is - Microsoft's usurpation of the SQL name is a particularly dirty example of trademark plundering). Ballmer said that this so-called "English query" was "a good example of one of the technologies that's come over from our Microsoft Research Division and moved into product". The problem is that this is not research: it's very old hat, and has been done very many times by countless organisations. At least we know what Microsoft means by "research". The demo Ballmer presented actually worked, but he had to admit that the scenario "won't really happen". So there we have it: Microsoft "inventing" a business application that nobody wants, on his own admission. Ballmer was gracious enough to say that "There are still some applications ... where we haven't seen customers embrace NT. We know we still have a lot of work to do on reliability, on availability, on scalability ..." Nevertheless, Windows 2000 "remains on target to be shipped by the end of this year", but the weasel words were "we make no explicit commitments". ®