Proprietary Software's Premature Obituary

Another prediction of Linux World Domination(tm) has appeared on Geek dot com. Not atypically for this sort of thing, it's coupled with another prediction of the demise of Microsoft and proprietary software in general. Is it likely that we have actually heard a "death knell" for the proprietary approach? To provide a partial answer to that question, we can look at the successes and failures Linux has had on the road to world hegemony.

First, it's helpful to remember that we are dealing with at least three separate markets when it comes to Linux vs the proprietary competition. Those would be the server, embedded and desktop markets. Taking them in reverse order, please allow me to pontificate on the relative strengths and weaknesses of Linux in these market segments.

Desktop

Desktop dominance has hovered like a shimmering mirage in front of Linux enthusiasts, and some Linux companies for years. The massive failure of Vista gave people lots of hope that the day of desktop dominance had finally arrived. But the fact is that Linux gained a minuscule amount of share in this market over the years since Vista's release. Apple benefited more, but Macintosh is still below 10% share. Why is that?

Dirty Tricks

Microsoft has been infamous for using illegal methods in pursuit of its businesses. I followed the antitrust trial back in the 90s, and read about all the dirty tricks MS pulled to fend off the twin platform threats of Netscape and Java. The company seemed willing to go to any lengths to crush competitors. But Microsoft is unable to be quite so bare-knuckled these days. The antitrust trial illuminated a lot of Microsoft misbehavior. And, although the US DOJ under George W. Bush backed off on tough sanctions that might have been effective in modifying Microsoft's behavior, the European Union later stepped up and actually enforced their antitrust laws. This provided more evidence of Microsoft wrongdoing, and came with sanctions and restrictions that actually had some effect on the company. Bill Gates' reduced role in the company may have led to less over the top behavior by Microsoft too. But despite having to tone down some of the excesses of the past, the company still ferociously defends markets in which it is entrenched, and remains a potent threat to any competitor trying to play in one of them. However dirty tricks are less likely to be the only factor driving Microsoft's success today. What else could account for that?

Hardware

Though the hardware driver picture has improved considerably on Linux, Microsoft's market dominance still means that hardware vendors are more likely to deliver ready to run drivers for their products on Windows first. Your latest PC may have trouble in that department. My not-so-new MSI Wind U100 netbook runs Ubuntu 9.04 Notebook Remix. It's gorgeous, but the wireless won't do WPA2 authentication. This means that until Ubuntu fixes the bug, this cool little netbook will work fine in the coffee shop or the airport, but not at home or in the office. You can obtain, patch, build and install a kernel module to fix the problem yourself, but if you think about it, you'll see that's irrelevant to this discussion.

Applications

Perhaps the weight of Microsoft's installed base, what the DOJ called the "applications barrier to entry," is the reason Linux can't seem to gain a lot of traction on the desktop. It's certainly true that Microsoft enjoys enormous leverage with software vendors, due to the massive market their platform provides. This leverage makes it more difficult to defect from Windows on your home PC. Your proprietary apps may work on Linux through some combination of wine and virtualization, but try to get support for those solutions, in the general case. What I mean by "support" is not just help when it breaks, but smooth and easy installation and initial configuration. And though high quality native equivalents to important commercial applications exist, few can boast the installed base, and the concomitant support from vendors and community resources that popular commercial apps enjoy. On business desktops, it may be more feasible to deploy Linux with a limited set of applications, either virtual Windows ones or native. But the IT staff is still faced with a relatively more difficult job supporting those apps given less vendor support.

New User Friendliness

Then there's the difference in ease of learning you often see between proprietary and F/OSS. A specific example may show better what I mean. Take Photoshop and Gimp. Go to http://www.adobe.com/support/photoshop/ and compare the new user documentation to http://gimp.org/docs/. Pretend you have never used either app and try to figure out how to get started. Hint, you'll find the info at Gimp's site, but you'll have to dig deeper, and you won't get the same quality for the new user. And this difference extends to third party support for the applications as well. Do a book search on Amazon, first for "photoshop" then "gimp," and count the number of results. Go to Lynda.com and check out the Photoshop video tutorials. Try to find anything 1/2 as good for Gimp there or on YouTube.

This focus on newbies goes to the heart of the Windows platform advantage on the desktop. New users of Linux tend to be significantly disadvantaged compared to their counterparts in Windows as far as learning new applications goes. My feeling is that's so because the folks that develop the software, as opposed to the people who integrate the software into a distribution, tend to lack a new user's perspective. They produce software that is easy to use once you get to know it well. Since Linux desktop domination requires coaxing lots of Windows users onto an unfamiliar platform, this deficit in hand-holding newbies bites hard. Linux distributions struggle to provide the new user with a consistent and usable environment for desktop computing. New users of Linux are likely to perceive the difficulty in picking up a new (insert user's critically important app here) as the whole story on Linux as a platform. On the other hand, it's not like Windows as a platform offers a whole lot of help in the integration department either. But the apps tend to be designed by teams that include people who want to suck new users into using them. With apologies to the minority of application projects that have worked hard to design in discoverability, and who have provided outstanding, lucid and accessible documentation with the naive user in mind, all that is just not a priority in most open source application development.

Which is it then?

You can argue about how much of Window's advantage is due to inertia and market size, and how much is due to apps that are relatively easy to learn, but there's an experiment underway that can help answer that question. MacOS X is a Unix based OS famed for usability. Applications on MacOS X are often easier to use and learn compared to their Windows equivalents On the other hand, Macintosh suffers from a similar disadvantage in critical mass that Linux does with respect to hardware and software. Macintosh enjoys about a 10% share of the desktop market, whereas Linux is around 1%. Since the two platforms face similar (though not identical) challenges trying to overcome Microsoft's market domination, we can factor out those disadvantages when comparing the two OS in terms of market share. To a first approximation, a substantial portion of that tenfold advantage in Mac desktops over Linux must reflect the advantage usability confers.

Embedded

In the embedded market, Linux has a big advantage in cost. If you are talking about a mass market item, like a cellphone, that cost advantage is a huge factor. Also, embedded applications tend to be under tighter control than typical desktop apps. An alarm system is a special purpose device - achieving usability is straightforward. (Even so, it's remarkable how many embedded applications suck bigtime.) The basic cellphone applications, making and receiving calls, accessing voicemail, managing contacts, and so forth, are well defined and relatively uncomplicated. As you move up to more general purpose computing on a mobile phone, complexity increases, and many of the usability factors that are important in the desktop space come into play. But there's less historical MS hegemony here. Finally people seem to be more willing to accept learning a different way of navigating a new phone's interface compared to learning a new computer OS. So Linux has a clearer field in the embedded space.

Server

Finally, in the server market, Linux has done and will continue to do very well. GNU/Linux started life as a Unix clone, and Unix was and remains a server OS (MacOS X notwithstanding.) Basic server applications on Linux are more mature, and requirements for application usability are different. Engineers volunteering time for server development on the GNU/Linux platform look more like their users than their desktop cousins. Indeed, the users are often those very engineers. Linux attracts quite a good deal of free R&D for new server roles. Today it's virtualization, yesterday it was clustering. Linux led both those technologies, and maintains strong positions in both today. Tomorrow the popularity of Linux in academia, and the support of big companies like IBM and HP guarantee new trends in servers will be quickly taken up, if not actually pioneered on Linux. And businesses care about some of the the advantages F/OSS confers on computer users. Avoiding vendor lock-in and providing fundamental transparency in the infrastructure are attractive to business, along with low cost and high software quality. I'm aware there are advantages to Free Software beyond the ones I just listed, but those are the ones that tend to appeal to businesses interested in using Linux in the server room.

No Panaceas

Linux is not one thing. Diversity abounds. People contribute to it, and use it, for a huge variety of reasons. So it's natural that Linux succeeds differently in different contexts. It would be unreasonable to expect anything else. Panaceas don't actually exist in the real world. Linux is not a panacea, it's a real, live, vital OS, with it's unique set of strengths and flaws. It will continue to do well in many areas, and may do better on the desktop some day. But Linux has farther to go there than in the markets which leverage its natural strengths.

Dead Software?

If you stop thinking about Linux and F/OSS as singular answers to the problems posed by computer systems and applications software, you may be able to see that software sold for profit actually has strengths and weaknesses too. However much you may wish it weren't so, from the user standpoint, software written for money rather than love is often a better solution. Alternatives to many of the flagship proprietary applications need to provide more than high quality code. They have to be designed, packaged, documented and supported to compete with their closed source competitors. One problem is that software engineers who code for love may often not be the best people to deliver on all of those requirements. Until that, and many other things change, proprietary software will stay off life support.