Linux for the End User—Phase 1
Several months ago, I bought some clothes with my LinuxFund credit card. The guy behind the counter looked at the card and said, “Linux! That's the future, right?” He then leaned over the counter and asked “What do you think is going to happen to the PC?”
The biggest single change in that loose amalgam that is the “Linux community” in the last year is that it now includes people, many thousands of people, who have heard about Linux in the press and liked what they've heard, but don't know anyone who uses it. The old model for the spread of Linux—you tell two friends and they tell two friends—is over, because the press attention has spread news of Linux far outside the community of actual users, to the point where a random cashier in an unrelated business feels confident of its revolutionary impact, even without any direct experience with the OS itself. For these interested-but-unconnected people to become Linux users, they will have to work harder than any of the Old Guard did, because they don't know anyone with enough experience to help them. They will have to find their way to Linux more or less from scratch.
I was reminded of my encounter with the Linux-loving cashier a few weeks ago when Almaz, my wife, decided that she was writing too much to keep working on her laptop and wanted a desktop. All she needed to do on the desktop, she said, was use Netscape, WordPerfect and a printer. And so an experiment was born, an experiment to see whether, armed with only a credit card and a willing Windows user, one could buy a complete Linux desktop system with a printer which that user could set up and get started with on their own. Almaz, who uses a computer every day but has no burning desire to understand what goes on under the hood, seemed to be the ideal test candidate.
This experiment is really two questions rolled into one. The first obvious question is, “Have the recent improvements in usability created a Linux which a casual user would feel comfortable with?” The second and less obvious question is, “Are there companies thinking about how to package and deliver such systems to interested home users?” Since I needed to procure the system before testing it, I set out to answer the second question first.
To begin, where to find this mythical, pre-configured desktop system? Although I have bought several systems with Linux pre-installed from several different vendors, they were all web servers. In this case, looking for a desktop system, I was almost as much of a newbie as my wife was. I set out down three different paths—using a search engine (Yahoo!, in this case), going to the major Windows PC vendors to see if any of them would ship a PC with Linux, and going straight for Linux URLs.
Reasoning that average users would have little idea of where to turn for pre-installed systems, I went straight for Yahoo!, usually a beacon of clarity in the fog of the Web. Yahoo! did not disappoint in this regard, with a manageable 22 categories which included the word “Linux”. But of these, only three categories were links to hardware vendors. Furthermore, two of those three were specifically business-to-business, leaving me only the slightly off-putting “/Hardware/Custom_Built_PCs/Linux/” as a category.
The problem with this category is that there's no such thing as a “custom-built” computer anymore—in the age of Dell, all computers are custom-built. Customers of commodity PCs expect to be able to adjust the configuration to their taste, and they don't think of this as buying a “custom-built PC”; they just think of this as buying a PC, period. A big part of serving interested but inexperienced users is creating a sense that they are not going to be bombarded with too many choices, but no one selling Linux systems has figured that out yet.
Yahoo! included a scant nine sites which advertised Linux hardware, and these were not always listed in the most user-friendly terms. Would you send a want-to-be Linux user to a site advertising itself as “a Linux-centric systems integrator offering support, hardware, applications, development and services”? Nevertheless, these sites went on the master list. Time for the next strategy: going straight for the traditional PC retailers.
The second strategy I pursued was going after existing PC vendors. I had heard that Dell, Compaq, HP and IBM had all “embraced” Linux, though I wasn't really sure what that meant and visiting their web sites provided few additional clues. While a search for “Linux” on all four sites turned up thousands of pages of documents, FAQs and press releases talking up their commitment to Linux, I was unable to find out, using only their web sites, whether they would sell me a Linux desktop system or not. In all four cases I had to resort to the phone, and in three of the four cases, Compaq, HP and IBM, I was told they sell Linux only on servers.
Dell, as usual, was a bit further along than its competition, and sells high-end Linux PCs, but the base price is over $2000, too high for a casual user's desktop. Furthermore, this option isn't even listed in the Home and Home Office section, only in the Business Users section, which would be daunting for newbies.
Nevertheless, if desktop Linux becomes a reality, Dell will be the company to beat. In my conversations with them, they always put me in touch with knowledgeable staff and they never sugar-coated anything, telling me, for example, that printer setup was likely to be “quirky”. I opted not to buy what was really a business-class workstation for Almaz, but when and if Linux for the desktop becomes a reality, Dell is better positioned to take advantage of it than any other Windows PC vendor. Everything they do is geared toward making the customer feel comfortable about making a decision.
So it's a strike-out with the traditional PC vendors. Time for the last strategy: going straight for Linux URLs.
In terms of newbie reach, what's the most valuable piece of Linux real estate on the Web? (Hint: it isn't Slashdot.) It's Linux.com, the Linux portal owned by VA Linux Systems, because end users typically think of dot-com sites as the Web's anchor tenants. Linux.com is a beautifully simple piece of web design, but as with so many sites, it suffers from insideritis. Clicking on “Desktops”, for example, gets you a page with a bewildering array of conversations about KDE vs. GNOME, but nothing at all about how you could buy a Linux desktop. Clicking on “Get Linux” gets you a page of every Linux distribution out there, but again, no information about actually getting a computer running one of these distributions. Searching for the word “PC” brings up page after page of BBS flames about Windows 98 vs. Linux, but yet again, nothing for the user who wants to buy a Linux PC.
Linux.com has no obvious, Yahoo!-like list of Linux system vendors. I have no way of knowing if this is a simple omission, or if VA Linux refuses to list competitors on a site it owns, but it seems like a huge oversight to host the single most valuable Linux URL from the outside world's point of view without directing interested users to places where they can actually buy a computer running Linux.
The problem here is that ordinary users don't buy operating systems. They buy computers. If there is ever to be a real spread of Linux to casual users, it will mean a move from a focus on this or that Linux distribution to a focus on pre-installed systems. Vendors will have to drop the idea that someday the average user will be comfortable installing an operating system—they don't even like to install ordinary software, so no matter how easy an OS install can be, it will never be as easy as buying it pre-installed. As the size of the Linux user base grows, the number of people who are comfortable installing from scratch will grow in absolute numbers, but will shrink dramatically in total market share. Whoever focuses on selling Linux computers rather than Linux distributions will win in the long haul.
Abandoning Linux.com, I went on to Linux.org (though I am convinced that most end users would not take this step on their own), and finally hit pay dirt.
Linux.org has the most complete list of vendors of Linux systems I have found anywhere, with almost 80 vendors listed in the U.S. alone. However, the list is hard to use, as it's in no particular order and has spotty information about what the vendors do. Furthermore, it looks like a prototypical web site circa 1996, with lots of white space between entries for “readability”, meaning that you can't see more than a handful of vendors at any one time, wrecking the ability to do any comparison shopping.
Nevertheless, this (plus a few sites from the Yahoo! list) is clearly the best source of vendors I'm going to find. Phase 1 is now done: I have gone from a standing stop to a long list of Linux system vendors to check out. Elapsed time: 48 hours. Frustration quotient: medium to high. Amount of work required: much too high. Phase 2 will be evaluating the sites themselves, and Phase 3 will be road-testing the ease of setup and use for whatever system we buy.
As a side effect of looking though these sites, it became clear to me that as the user base spreads, pre-installation will trump ease of use for determining which flavor of Linux users choose, and by this measure, Red Hat is way out in front. No matter how easy the Caldera or Corel distributions are (the two easiest installs in my experience), the real trump card in terms of growth in user base will be deals with hardware vendors. Surprisingly, no home page for any Linux distribution I looked at (Red Hat, Corel, Caldera, Debian, Mandrake) had what I thought would be an obvious link: a big red button that said “Click here to buy a computer running FooBar Linux from our partners”. The owners of different distributions are still assuming that most users will install their own Linux and, though that may be true for the moment, it will not be for long.
Conclusion: On a scale from “Rough” to “Smooth”, Phase 1 surely counts as rough. For someone interested in using a Linux box as a casual user, but with no idea where to turn, there is no obvious answer about where to look first. This bootstrap problem, where you have to know where to look before you do the looking, is the Achilles heel of easy-to-use Linux right now. While it is unclear whether Linux will ever make any real inroads into the MS desktop monopoly, if the software were ready tomorrow, interested users still couldn't find it. As we have learned from Amazon, eBay and Yahoo!, this means that if this space ever does take off, the first vendor to associate themselves firmly in the Web's collective mind as the official vendor of Linux desktops could walk away with a lion's share of the traffic. Next month, we'll evaluate Linux systems vendors.
email: clay@shirky.com
Clay Shirky (clay@shirky.com) is currently the Professor of New Media at Hunter College, where he teaches in both the undergraduate and graduate programs. He has worked as a writer, programmer, and consultant, for Business 2.0, FEED, Silicon Alley Reporter, word.com, Urban Desires and net_worker magazine.