Metaphors make the operating system

Linux per se is not a specific set of ones and zeroes, but a self-organizing Net subculture. — Neal Stephenson

Neal Stephenson is a science fiction author perhaps best known for his magnum opus Cryptonomicon. I recently read and enjoyed a couple of his novels, so I was interested to discover his 1999 essay In the Beginning was the Command Line, which discusses the state of the operating system market four years ago from both a realistic and metaphorical viewpoint. If you haven’t read it, it makes tasty food for thought.

Stephenson clearly has some hands-on experience with a variety of platforms. Some of what he talks about is obsolete (BeOS is gone, beunited.org notwithstanding) but much of what he says about Windows, MacOS, and Linux is remarkably true even today. For instance:

It is because Microsoft’s excellent management has figured out that they can make more money for their stockholders by releasing stuff with obvious, known imperfections than they can by making it beautiful or bug-free.

I read that quote in a cautionary light the week after Microsoft released its first major server operating system revision in three years. However, either Stephenson is wrong or Microsoft has changed its attitude toward beauty since he wrote his essay. I think “look and feel” has been a big factor in Microsoft’s success. Its desktop, icons, and typefaces have a comfy feel. And the “user experience” was the company’s main target for improving its desktop operating system when it developed Windows XP. I believe a big reason Lotus’s SmartSuite applications got creamed by Microsoft’s Office software was that Lotus’s look was clunky and two-dimensional. That sent a subconscious message to users that the underlying capabilities of the software were subpar too. (Well, Lotus struggled because of that and Microsoft’s marketing machine and questionable strongarm business relationship tactics.) Open source application developers have to make their programs at least as accessible as commercial software.

Hostility towards Microsoft is not difficult to find on the Net, and it blends two strains: resentful people who feel Microsoft is too powerful, and disdainful people who think it’s tacky. This is all strongly reminiscent of the heyday of Communism and Socialism, when the bourgeoisie were hated from both ends: by the proles, because they had all the money, and by the intelligentsia, because of their tendency to spend it on lawn ornaments. Microsoft is the very embodiment of modern high-tech prosperity–it is, in a word, bourgeois–and so it attracts all of the same gripes.

People who are inclined to feel poor and oppressed construe everything Microsoft does as some sinister Orwellian plot. People who like to think of themselves as intelligent and informed technology users are driven crazy by the clunkiness of Windows.

We still see this today, and it puzzles me a little. I can understand having a philosophical aversion to something — Microsoft, veal, country music — but you have to temper it with an objective viewpoint. Microsoft <≶ evil. Yes, it has played unfairly in the business arena, for which it should be punished. Yes, its products are less secure and less stable than they should be. But it weakens opponents’ contentions when they argue from an emotional perspective instead of addressing the tangible problems.

Because I discount the sinister plot school of thought, does that make me the intelligentsia?

There are only two ways to sell a product: price and features. When OSes are free, OS companies cannot compete on price, and so they compete on features. This means that they are always trying to outdo each other writing code that, until recently, was not considered to be part of an OS at all: stuff like GUIs.

This is a key lesson for Linux proponents to learn. Linux is a powerful operating system — no one argues that. But Linux started out as a difficult operating system for the average person to learn. There are far more average users than expert users out there. To grow the community, Linux must be as simple as possible to work with. A larger group of users benefits both average and expert users, so even experts should support “frills” like GUI utilities and package installation tools.

Note that I’m not advocating stripping out all of Linux’s powerful features and dumbing it down. The power stays available to those who can wield it, while those who can’t can rely on GUIs and helper applications. After all:

Most people who shop for OSes (if they bother to shop at all) are comparing not the underlying functions but the superficial look and feel. The average buyer of an OS is not really paying for, and is not especially interested in, the low-level code that allocates memory or writes bytes onto the disk. What we’re really buying is a system of metaphors. And–much more important–what we’re buying into is the underlying assumption that metaphors are a good way to deal with the world.

Much of Stephenson’s essay talks about metaphors for different operating systems, and how their creators get consumers to buy into those metaphors, then goes on to talk about how the metaphors affect the companies and their customers:

There are only two reasons to invest in Apple and Microsoft. (1) each of these companies is in what we would call a co-dependency relationship with their customers. The customers Want To Believe, and Apple and Microsoft know how to give them what they want. (2) each company works very hard to add new features to their OSes, which works to secure customer loyalty, at least for a little while.

Linux has no one creating metaphors to help sell it. Instead, it creates its own metaphor — brilliant Finnish student creates enterprise OS with help from worldwide community of self-motivated developers. It’s smart software! Root for the underdog! It’s a more sincere than metaphor than those created by Apple and Microsoft, but the marketing money behind the two commercial giants is beating sincerity hands-down today.

I commend Stephenson’s essay to you because it provides some interesting perspectives on operating systems and how we think of them. There’s more in it than I’ve been able to touch on here. I’ll leave you with one final tantalizing quote:

The only reason Torvalds had cheap hardware was Microsoft.

History repeats itself

When I read the news from Novell’s BrainShare conference I’m mentally transported back in time 10 years and in space 2,000 miles east from Salt Lake City to bucolic Westborough, Massachusetts, once home to Banyan Systems. Novell is taking NetWare squarely down the path Banyan took VINES. Can Novell avoid Banyan’s fate?

Ten years ago VINES was the best networking software for large organizations. (Full disclosure — I used to edit ABUI Network News, the magazine of the independent Association of Banyan Users International, but I was never employed by Banyan itself.) VINES offered centralized administration for large numbers of servers, with robust services for security, messaging, and backup, among others. Even more essential, it had StreetTalk, a highly scalable directory service that no one else’s technology could touch.

Unfortunately, Novell was eating Banyan’s lunch in most situations when the two went head-to-head. To combat Novell, the marketing geniuses at Banyan decided to advertise directly to the corporate decision-makers instead of IT managers through strategic ads in publications like Business Week and Fortune. They also unbundled many VINES services from the operating system, which would allow them to run on other platforms as a product called Enterprise Network Services. I was fortunate to edit an excellent book by Rob Sparre on how to implement ENS for SCO Unix.

By the time Banyan resorted to those tactics, it was too late. NetWare 4.0, released in March 1993, included NetWare Directory Services, which competed directly with StreetTalk. Although the initial product had several flaws, it provided a reasonable alternative for organizations looking for a directory service. Microsoft at that point was only talking about Active Directory, but Microsoft’s pre-announcements routinely paralyze organizations who feel they need to see what the dominant software company will come up with before they can commit to existing products. Despite its advanced technology and its operating system’s low operating costs Banyan was never able to gain ground against its more established competitors. In 1999, Banyan discontinued its products, leaving Novell and Microsoft to fight for the lucrative low- and medium-end networking market.

With NDS, you’d have thought Novell would have a major leg up against Microsoft, but the company created a major distraction for itself when it bought Utah neighbor WordPerfect Corp. in March 1994 and tried to diversify into the applications space. While Novell was dividing its energy, Microsoft successfully expanded its operating system expertise from the desktop to the server, putting Novell in its current perilous position.

That brings us to where we are today. How is Novell going about reversing its fortunes? Well, it’s taking out ads in publications like Business Week and Fortune to target corporate decision-makers instead of IT managers. And it’s unbundling some of its services to make them run on other platforms — Linux, to be precise.

Hmmm.

Luckily, that’s not all Novell is doing. It purchased Cambridge Technology Partners two years ago to add a strong service component to its business. Service, when done well, is a proven money-maker. Banyan had only a tiny service business, because it sold its product through resellers, who wanted to retain that income stream themselves.

One major difference between 2003 and 1993 is the emergence of the open source community. By opening its software, Novell can use work by programmers not on its payroll — a significant advantage. It remains to be seen, however, if Novell can interest the open source community in its products.

Can Novell survive? My heart says yes, but my head isn’t so sure. Novell needs to prove it has a lower cost of ownership than the alternatives, but that won’t be enough. Novell has to hope for an emotional backlash against Microsoft from Microsoft’s current user community. Without that, it’s just too easy for organizations to follow the natural upgrade path without investigating possibly better alternatives.

Spring into Linux training

On April 24, as the flowers appear on the earth and the voice of the turtle is heard in our land, Microsoft releases Windows 2003 Server. That marks a good time for Microsoft shops to begin some training — in Linux administration.

Many Windows shops are still using Windows NT Server 4.0. Microsoft has already announced that that product is on its last legs; Microsoft will stop providing hot fixes at the end of this year for all but security issues, which it will continue to provide until the end of 2004. Given that deadline, these organizations must decide how to upgrade their server infrastructure. If they’ve bought into Microsoft’s Software Assurance program, they’ve already paid for their server operating system upgrade, and will likely continue on to Windows 2003 Server. But many companies rebelled against Microsoft’s attempt to extract a steady stream upgrade dollars from its customers. These organizations can benefit from Linux — if they have the expertise to take advantage of it.

One of the main barriers to corporate Linux adoption is simple unfamiliarity. A cure for that is employee training, which benefits not only the organization but the employees. Training is key to helping organizations keep competent employees, and helping workers stay satisfied with their skill levels and hence their jobs.

Nowadays, however, with companies cutting back on all budget categories, getting your company to pay for formal classes may be difficult. Bringing an instructor in-house is costly, and sending employees out to classes means not only losing their skills for a day (or longer) but also paying for travel and expenses.

Fortunately, with open source software, hands-on training is as inexpensive as you could wish for. Dig out an old computer that’s too slow for Windows 2003 Server, download your favorite distro or buy it from your favorite ISV, and set your admins to work on it.

If yours is like most organizations, it’s easier to talk about training than it is to find time to do it. At the same time, training is essential. Good system administrators who feel they’re falling behind in their skills are going to demand to keep up, either in your organization or someone else’s.

Take advantage of that desire to learn. Mark out an hour or two a week for your staff to perform tasks under Linux that they already know how to perform on other operating systems. If you can’t spare even that amount of time, just make the machines available; motivated administrators will use the equipment on their own time. If you have a Solaris, AIX, or HP-UX expert on your staff already, you have a jump start; make him a mentor for the rest of the group.

Make sure your staff is thoroughly familiar with the most commonly used Linux components and applications — Samba, lpr and lpd, Apache, an e-mail server like mailq or sendmail, cpio, and ipchains, iptables, and Snort. Set discrete tasks and goals — for example, ask them to provide equivalent file and print services to what the organization is currently using, have them pilot a migration from Exchange Server to open source e-mail, or have them create and monitor a secure subnet.

The experience your staff gains this way should benefit your organization, if not immediately, then in the medium term, as it broadens the feasible options for new server operating systems.

Some employees will want tangible substatiation of their skills in the form of certifications. Possession of a certificate demonstrates that your skills have been tested against a defined set of tasks by an objective third party. Groups like Linux Professional Institute, CompTIA, and Red Hat offer certificate programs in Linux administration. Human resources departments seem to find certificates worthwhile. I’m less impressed. Classroom and book learning can’t hurt, but they’re no match for hands-on real-life experience.