Against whom is Linux competing?

It will take a major Linux vendor pairing up with a first-tier OEM and a major enterprise or government for Linux to make inroads on the desktop.

In the closing of his article What Linux needs to succeed, Paul Murphy writes of Linux devotees:

Stop trying to make Linux look like Windows, don't put those [Windows] people in charge, and don't let anyone pretend that Linux is some kind of cheaper Windows replacement. Linux is what it is: Unix, and it takes different reflexes, different ideas about networks, about the role of the computer, about data storage, and about application management to make it work.

While I agree with a number of his conclusions (such as the relative strengths of Solaris, BSD, and Linux distantly followed by Windows), I think he misses the point.

Paul attributes the success of Windows as momentum from the "after effects of the initial PC adoption in rebellion against mainframe control practices" but Paul forgets that very little of what people do today with their Wintel workstation was ever done on the company mainframe. Before the IBM PC, most of the things MS Office does today were done on a typewriter or with a pencil and ledger sheets, and the results were presented with hand-drawn artwork published by a professional typesetter. Of course, very little done on today's personal computer was even thought possible on a mainframe in 1980, when the IBM System/370 was still king.

John Carroll challenges Paul's advice when he writes, in his article A recipe for the failure of Linux, "Linux would attract more Windows customers by figuring out what those customers like about Windows, and riding that wave into Windows' users homes."

John likens the problem to that faced by US automakers who can't seem to figure out why Americans prefer Japanese cars. (Ironically, my Honda was made in Ohio, the Chrysler Sebring -- which I almost bought -- was made in Mexico! Go figure.) Paraphrasing, he says, it's best to figure out what the customer wants and give it to them. Well good advice -- but WHICH customer?

John suggests that Apple got it right when they moved their MacOS product to BSD because they now offer what is arguably the best user experience around in a UNIX-based package, with all of the inherent advantages of UNIX underpinnings. Well, that's all fine and good but where exactly has this wonderful "new and improved" MacOS / UNIX marriage (known as Mac OS X) gotten Apple? Has it cut into Microsoft's market share? No. Why? Because technical superiority is not always enough. (Just ask the inventors of BetaMAX.)

Both Paul and John seem to think that Linux has been a failure because it hasn't made serious inroads into the Microsoft desktop juggernaut. However, I am sure that Sun, IBM, and others would disagree regarding the relative success of Linux in the marketplace. Any assumption that UNIX no longer plays a major role is also a false assumption as virtually all Internet infrastructure is built upon UNIX infrastructure. How is it then that, in many an enterprise, Microsoft Windows is the predominant server? Not UNIX? Not Linux? Not MacOSX? (After all, they are all based upon UNIX.)

While UNIX was doing what it does best throughout the last 35+ years, it continued to be geared to the needs of the network and the needs of technical people whose job it is to get the most out of expensive hardware and bandwidth -- and doing so with costly technical expertise available only from the nation's universities offering programs in computer science. In truth, other than universities, no one in the enterprise cared very much about UNIX until they needed to figure out how to use leverage the Internet to their advantage.

Throughout the 1990s, Microsoft continued to saturate the market with a one-stop personal productivity solution for consumers. Apple continued to offer a superior solution but consumers were not interested in paying premium prices for what were quickly becoming commodity products. The personal computer made its way into the enterprise for the very same reasons it made its way onto America's dining room tables. It was easy to use and inexpensive to buy. But Microsoft faced a dilemma ... it was quickly turning its only product into a commodity with low profit margins:

  • How could Microsoft take Windows to the next level? By adding value to its products through a network of Windows computers all linked to Windows servers.
  • How does Microsoft make this solution more attractive than a similar UNIX solution? Two ways: By offering services not available via UNIX -- proprietary solutions. By making them less expensive to maintain by offering inexpensive training for non-technical people.
Today, high-performance UNIX hardware continues to be largely proprietary in nature and the skill set to maintain UNIX environment requires a degree in computer science and/or years of experience. Yet, anyone with a few hundred dollars and a high-school diploma can be trained in a matter of weeks to administer a Windows server based upon an off-the-shelf x86-based workstation. Is it a high-performance solution? No. Is it highly secure? Not out of the box (but it can be made to be.) Is it cost-effective? Absolutely! Virtually any enterprise is large enough and has enough money and expertise to support a Windows network -- even one connected to the internet. This is just not the case with UNIX.


Enter Linux ...

I said above that I thought that Paul was missing the point when he suggested that people stop comparing UNIX/Linux to Windows. Well I do but Paul is right on one score -- UNIX and Linux are well-suited to do the same things. Which means that they are also well-suited to do anything that a Windows environment can do. And both will do it more efficiently. So?

Ultimately, if one does not wish to compare Linux to Windows, then one is left with two small markets for Linux. The technical consumer (some might say geek) -- who downloads Linux for free, and the large enterprise who wants 24/7 support for their software investment. Two groups which are pretty much mutually exclusive.

At first glance, Linux can perform all of the functions of a Windows desktop and a UNIX server and cost the same or less than the equivalent Windows box (and a lot less than an equivalent UNIX box.) So why is it not making significant inroads onto the Windows desktop? Well, this question takes me to my question to John. Who is the customer?

Microsoft has clearly delineated two groups of customers. Consumers and the enterprise. Linux has not. How does Microsoft do it?

Consumers are sold through Microsoft's OEM's. This market is left pretty much to Dell and HP for the mass market and a lot of smaller OEM's for the specialty market. In the end, Microsoft doesn't care whether you are a stay-at-home mom or a C++ programmer. The key is that no matter what their level of expertise, or the size of their pocketbook, the consumer can buy a personal computer with Windows already installed which will do everything they can imagine. Why not a Macintosh instead? Because of the price-point. Because there are fewer points of sale. Because there are fewer available applications. Enough said about Apple.

The Enterprise, as Microsoft well knows, is made up of the very consumers who have a PC at home on their dining room table.

The small enterprise without any technical expertise can put up a robust, reasonably secure Windows network for a few hundred dollars. They can do this through their favorite OEM and Microsoft need do nothing. Why? because the small enterprise does not have the expertise to support either UNIX or Linux. (It doesn't matter if Linux is less expensive to buy if you cannot support it on an ongoing basis.)

The large enterprise is a different beast altogether -- and is asking different questions. These are the folks which Paul is really talking about in his article. They know that there are places where Windows servers are well-suited to the job and there are places where UNIX/Linux has a clear advantage. Microsoft is in a position to be very aggressive in this market because robust hardware and software is always expensive and costs are measured in TCO over an extended period of time. Up-front costs mean nothing compared to productivity. Unless the large enterprise is a start-up, they have been doing IT for years and have plenty of UNIX/Linux expertise -- and can afford to pay for it if the efficiencies of UNIX/Linux can be translated into improved productivity. Ironically, the mainstream UNIX vendors are being squeezed from both sides in this environment. By Microsoft offering proprietary functionality and low training costs and by various Linux vendors offering low up-front costs but little or not training path for non-technical staff.

So, how does Linux make inroads against Microsoft? Well, they cannot in the proprietary services arena for the same reason UNIX cannot. That leaves the desktop itself. But how? Find out what the customer wants. For the desktop, the overriding concerns are cost and ease of use. Let's look at these one at a time:

Cost. This is the easy one. If you are a robust technical user, you can get Linux free but you will spend hours setting it up and configuring it -- most likely on a machine which came with a copy of Windows included. If the machine did not come with Windows installed it was probably because the user put it together themselves -- or they paid an OEM a premium to leave Windows off the machine. Sure there are a few third-tier vendors selling pre-installed Linux desktops at Wal-Mart at entry-level prices but if one wants a tier-one hardware vendor offering robust technical support or out-of-the-ordinary peripherals they are out of luck.

Ease of use. This one is a little tougher. Everyone reading this article knows that Linux can be made as easy to use as Windows or Macintosh. The problem is that this is now a commodity marketplace and, while Linux vendors such as Linspire have been making great inroads in this area, profitability is the challenge. So is customer confidence. Linux will not make much headway in the consumer spac until a consumer can call up Dell and say, "I want Linspire on my Dimension desktop. Oh yeah, I also want all of these peripherals."

Hardware Compatibility. This is a HUGE issue for anyone wanting to put Linux on anything other than a plain vanilla personal computer -- mainly because Linux vendors can only afford to invest so much effort into testing before they have to increase the price of the OS. This problem goes away once a major Linux vendor enters into an agreement with a first-tier vendor such as Dell or HP but whether that will happen anytime soon remains to be seen.

Unlike Microsoft, which long ago established highly profitable agreements with its OEMs, Linux vendors are trying to figure out to whom they are trying to sell their products. To the large enterprise to replace aging UNIX systems or to consumers which represent very low margins but with the potential to be profitable at high volumes? Short-term profitability favors the enterprise as the target customer.

It is going to take a major Linux vendor pairing up with a major first-tier OEM and a major enterprise or government to make the inroads on the desktop which will make Microsoft finally take notice.