Kapor's Thoughts on Desktop Linuxby Daniel H. Steinberg
Mitch Kapor is one of the pioneers of the PC industry. Tim O'Reilly introduced Kapor as the creator of the original killer app, Lotus 123. Kapor is also known for his involvement in the Electronic Frontier Foundation (EFF) and the Open Source Applications Foundation (OSAF) and for his work on the Chandler project helping to create an open source PIM.
Changing with the Times
Kapor asked, "what stands in the way of Linux as an end-user operating system?" He thinks that Linux will end up with a significant share of the desktop in the near future. This is partly because the business and political climate have changed in the last 20 years. While at Lotus in 1985, Kapor met Richard Stallman. Stallman was picketing the Lotus building because Lotus was aggressively suing people to protect their 'Look and Feel'. Kapor agreed that there should not be intellectual property protection for Look and Feel but he also thought that free software was one of those "hopelessly idealistic ideas that would never work in practice."
Six years later, in 1991, Linux was introduced. Kapor argues that Linux is so successful on the server that it is attracting parasitic companies whose business offerings consist of little more than Linux-centered litigation. As for desktop computing, Kapor asserts that Linux is gaining credibility in this era of Net-centric computing as this focus on the Net is in the DNA of Linux.
Kapor points to other business indicators where Linux is beginning to show success on the desktop. Microsoft understands the pressure to provide less expensive software. It doesn't make sense to consumers that falling costs make PC hardware available for as little as $200 while a license for the software required to run it sells for $500. In addition, there is a resistance to upgrades among the install base because people are reasonably happy with what they have. The business community also resents and is pushing back against onerous licensing terms.
Kapor concludes that the climate is more open to alternatives. He also cautions that it doesn't matter how good Linux is without success stories. Before people switch they want to know that someone else has switched successfully. There are massive deployments of Linux desktops. Animation studios have been switching their render farms to Linux. You see mass installations of Linux at different sites such as 14,000 machines in Munich and 80,000 deployed in the Extremadur region of Spain for an estimated savings of 18 million dollars. Thailand has deployed 160,000 of the up to 1 million computers they plan to deploy with a localized version of Linux. Having a mandated program of Linux on the desktop isn't enough, though. There is a fear that users can wipe it out and replace it with a bootleg version of Windows as is happening in China.
Desktop Linux Adoption Cycle
One theory of adoption is that there are killer apps. Kapor does not think Desktop Linux will be adopted because of killer apps. He sees the adoption starting out on the edge in the less than mission critical applications and then moving towards the center. He identified three phases in this adoption cycle.
Phase 1 adopters are technical users. Millions of people are now in this group. The problem is that there is a limited number of developers. This market will exhaust itself.
Phase 2 adopters are transactional workers. These are people who use a computer to perform a specific task such as a call center. In this situation you don't need a computer that can support a wide array of commercial applications. You tend to have focused and specialized applications used by these operators. The IT managers responsible for call centers are looking to replace desktop machines with Linux because they only need such a discrete set of applications. Kapor expects to see significant numbers in this type by the end of next year and identifies attracting this audience as the immediate challenge.
Phase 3 adopters are the knowledge workers and consumers. Kapor expects that it will be a while until a significant portion of this population is interested. His guess is not before 2007. For this phase you will need to have the applications end users expect. OpenOffice.org, for example, is not as polished as Microsoft Word. More importantly, it lacks complete file compatibility.
Next, Kapor turned his attention to evaluating factors that effect the end-user experience. The first is a lack of consistency in the end-user experience with applications. Linux does not have a good desktop developer platform to provide a uniform look and feel. To developers these oddities are small and easily overcome or ignored, but the inconsistent UI is a problem in attracting larger markets.
Installation and support for peripherals and key applications still need to be addressed for Linux to be more widely adopted. For the most part, Linux is easy to install on desktops but often more difficult on laptops. If things go wrong, Kapor notes, you have to be an expert to bail things out. When it comes to peripheral support, a device may not be supported or it may be supported but you need to find a driver or even have to recompile the kernel. The basic applications such as OpenOffice.org, PIM alternatives, and browsers are available for Linux, but there are issues. In Asia, for example, many sites only work with Internet Explorer.
The overall eco-system is also a factor in widespread Linux adoption. This includes the people and institutions that are committed to Linux, that work on it, and that support it. There is a thriving community of open source developers but there is a lack of commercial pickup. There isn't much of an independent software vendor community or OEMs or distribution channels. Also the OEM's who charge a "Windows tax" result in Linux users paying for an operating system they never intend to use.
That Which Must Be Done
Kapor laid out a technical agenda for improving Linux on the desktop. Most important is addressing the issue of Office file formats. Kapor suggested two strategies in this area. The "Join 'em" strategy is built around interoperating with Microsoft Office. Although OpenOffice.org has done a good job, there remain many issues at the margin. An alternative approach is the "Beat 'em" strategy. This would involve creating an alternative — possibly an XML based format — for the standard office applications.
Another part of the technical agenda is to strengthen the desktop foundations. X is very important in this regard. One of the missing pieces remains a solid hardware abstraction layer. Also there needs to be integration between various desktops. As an example, different Linux distributions are not consistent on which file systems are mounted and show up on your desktop. Much of this type of work either spans multiple open source projects or doesn't seem to belong to any of them. It is, however important to address these desktop integration and usability areas as they are a barrier to widespread adoption.
Kapor sees a bright future when it comes to Linux adoption. He noted that we are in an era where because of the momentum of Linux on the desktop we'll continue to see Microsoft price cuts to remain competitive. He added, "Predicting share numbers is tough but it wouldn't surprise me to see 10 percent of desktops running Linux in the not too distant future." One of the markets driving adoption of Linux are non-U.S. governments that are looking for low cost alternatives. He also predicted that in this country the public sector would lead the adoption. Although he anticipates selective adoption in enterprises, he still does not predict a huge amount of consumer momentum for a while.
The OSAF was founded to help accelerate of Linux on the desktop by identifying points of leverage and supporting particular efforts. One service is to fund extensive compatibility testing of Excel functions and more generally support the testing of Open XML file formats and Microsoft Office compatibility. Another service is to act as a fiscal agency for some projects as appropriate. OSAF is not an incubator but it can provide an improved infrastructure for open source projects so that the projects can concentrate on their software and not on compliance with various regulations. In addition, OpenSector.org has been introduced as a site for announcements that include studies and details of deployments of Linux in public agencies.
Daniel H. Steinberg is the editor for the new series of Mac Developer titles for the Pragmatic Programmers. He writes feature articles for Apple's ADC web site and is a regular contributor to Mac Devcenter. He has presented at Apple's Worldwide Developer Conference, MacWorld, MacHack and other Mac developer conferences.
Check out all of our ongoing O'Reilly Open Source Convention News for continuing links and commentary.
Return to ONLamp.com.