.comment: In Praise of IBM - page 2
The Real World Series
Additionally, IBM does not arrive exhausted at its own product rollout. Everything they do they see as value added to an existing system -- an existing system free of onerous royalty payments, of dictates from the owner, and in many respects free of cost. This isn't to say that the company merely downloaded Linux and put it on PCs which it then began to ship. It could have done this, entirely legally. But instead, IBM is wading in to the development community, contributing code, documentation, and perhaps most important, putting its name behind Linux. It becomes far more difficult to write off Linux as a hobbyist operating system when IBM isn't just shipping it but also buying ads touting it.
People I know within IBM are excited about Linux -- really excited. The company has always had a wonderful population of real gearheads. Back in the days of OS/2, there was (and may still be, actually) an IBM ftp site devoted to "EWS." This stood for employee-written software, and it comprised all kinds of apps cooked up, often as not for the sheer love of the art, by IBM employees. It was here that the cool little Tiny Editor came to life (a port of it to Linux would be extremely welcome, at least at my house), as well as astronomical programs, programming tools, all sorts of stuff. It was all available, free for the download. So it's no surprise that a lot of people in Armonk and at Watson, and in England, and elsewhere, find the whole idea of Linux and the way it's distributed perfectly natural.
And it makes good corporate sense. IBM sells hardware and services. This is no more difficult to do with Linux than it would be with another, proprietary operating system; indeed, in many respects it's easier for skilled people, and no one has ever accused the IBM corporation of a lack of people skilled in computing. It's true that other companies offer hardware and services, but the IBM name continues to carry some weight, and the very notion of an IBM S/390 running Linux is still something the boggles my mind and, I suspect, even Linus's. But why not? Weigh the advantages and the disadvantages. The former list will be a lot longer than the latter.
The company is putting Linux on all its servers. And if you want something else, the company is prepared to argue with you about it -- or, to put it less confrontationally, to explain why Linux is better. If you look at IBM's Linux website, you'll find articles on the advantages of Linux over Windows or Solaris. (There's no mention of AIX in all of this, but AIX is largely a creature of internal IBM politics and is likely to stick around pretty much forever, and to those who love it, my good wishes. And of course there is AIX 5L, which employs the Linux API.) The company has taken out ads in places like OpinionJournal.com, which is a site visited frequently by corporate people positioned to make decisions.
Having been well and truly screwed by the Redmondians on several occasions, IBM is still hedging its bets a little -- it lists Caldera, Red Hat, SuSE, and TurboLinux as "key alliances." Actually, it goes beyond that -- there are segments of of the market or parts of the world where each of these companies is the leader. Of particular interest is Caldera, which has merged with SCO and therefore competes with AIX in the proprietary Unix space. Where this might lead is unknown, but it is worth noting. SuSE distributes IBM products in some regions. Turbo is doing some interesting things with its proprietary system control application, and it is the leading Linux distributor in the Far East. Red Hat, for all that I and others complain about it, does produce a decent distribution and it certainly is the biggest American distributor. Beyond the idea of not getting locked into just one distribution -- and dealing with distributions is a good idea, because it allows for handing off of some services, chief among them small business support -- IBM is recognizing the fragmentation of the Linux market. Some distributions are better suited to particular tasks than others are.
The Other Shoe
The question left dangling is where the desktop fits into all of this.
IBM has from time to time flirted with the single-machine market: One machine, or perhaps a desktop unit and a notebook, or even a little network of two or three nodes. And IBM, like others, has discovered that it's difficult to make money in that space. Support issues eat up unit profits pretty quickly. Windows is good enough for the gamers-and-Quicken market, or so goes the reasoning. That reasoning is pretty difficult to argue with. IBM's push of OS/2 for the desktop (Warp3) was disastrous, for a number of reasons. Will the company decide now that the low marginal cost of Linux justifies promoting it for more than server use?
I don't know, and neither do the people I talk to within IBM, though many of them hope so. The company has been involved in the Gnome Foundation, and has expressed interest in KDE as well. While there's no question that participation in these desktop projects is pocket change to an IBM, even a smart hedge move just in case Linux on the desktop takes off, the company's involvement is a good thing. But there are compelling reasons for IBM to ship Linux on desktop machines. Again, the costs of implementation would be low, and much of the support could be handed off to the distributors themselves -- no need for a special IBM distribution. The real issue has to do with the maturity of Linux productivity applications. Nobody is going to buy a PC with Linux so as to get EMACS. The office suite situation is still in flux (to say nothing of the idea of shipping StarOffice while having a Web page that slams Sun's operating system), and Microsoft Office filters, though much improved in the surrent Star/OpenOffice incarnation, are still subject to the whims of Microsoft, who could change file formats tomorrow and render those nifty filters increasingly useless. Changing file formats has proved an effective cash cow for Microsoft anyway, forcing users to "upgrade."
It would be a sheer delight to see IBM embrace the idea of desktop Linux, and even better to see the company throw some serious weight towards desktop applications development. My pet project here would be for IBM to buy and free the source code for the DeScribe word processor. Shawn Gordon of theKompany.com and I both discussed DeScribe for Linux with James P. Lennane, DeScribe's owner, earlier this year. Shawn seems confident that a port could be made without undue difficulty, and building it to Qt would make possible multi-platform versions, a good challenge to the Redmondian hegemon. Plugging in the best filters available in open source would be a cinch. (There have been questions raised about porting SmartSuite to Linux. I'm given to understand that this would not be all that easy.)
But even if IBM eschews promotion of desktop systems, the company's involvement in Linux still might push other vendors to consider it as a default operating system, with others optional at extra cost. The companies that have built and shipped Linux machines have fallen on hard times, not due to Linux but to the times, and those who have offered it have backed off to a greater or lesser extent. The margins on PC sales are already very low -- Ted Waite's cow now advises him to sell a fast machine with an LCD panel and a CD burner, and come to your house and set it up, and do all this for under a grand. PC vendors are hurting. Throwing Linux into the mix could prove interesting indeed. Here's hoping that IBM leads the way.