LINUX AT THE CHASM Proposal for a Business Track paper at 1999 Atlanta Linux Showcase I Innovators, Visionaries, and Pragmatists The Technology Adoption Cycle is has been known and respected for a long time; it was Geoffrey Moore in his Crossing the Chasm book who applied it rigorously to the high-technology marketplace. To review it briefly: a technology is first taken up by a group called Innovators, who value technology for its own sake and who have no budgets. This is the stage at which Linux first entered corporate America, as a quiet in-house solution to a low-budget problem, specifically Web connectivity. Linux is currently in the second stage of Adoption Cycle, that of the Early Adopters. This segment is driven by Visionaries, who are constantly scanning new technologies in order to pick one with which to gain a significant competitive advantage. Because they exist inside large corporations, Visionaries do have budgets, and will use them to induce vendors of new technologies to make extensive customizations to serve the vision and gain the advantage. Signs that Linux is in the Early Adopter phase are the number of projects inside large corporations; these signs can be difficult to read, however, because advantage-seeking visionaries do not particularly want to advertise what they are doing. Linux advocates are convinced that there is much more going on than meets the eye, but the number of companies willing to commit publicly is still quite small. Beyond the Early Adopter phase lies the famous Chasm, the vast gap that represents the difference between the Innovators and Visionaries on the one hand, and Early Majority on the other. This Early Majority is made up of Pragmatists who demand reference sites among their peers, a multiplicity of suppliers, and an established base of support organizations. References to NASA use of Linux leaves them cold: NASA doesn't sell anything. Above all, unlike the Visionaries, they are looking for continuity with their current operations and technology, not for an abrupt break. The traditional way for an application to cross the Chasm is to target a niche and then gain dominance in it. At this stage everyone recognizes that Linux is simply not going to become a desktop replacement for Windows in any general way. The success of Linux among the Visionaries should give us ideas as to what niches Linux might target. Web Connectivity This is the classic Linux case. The dominance of the Linux Web server is well-known; not only do corporations use it, but it is popular with ISP's. The reliability in this area has led to the further use of Linux as Print and File Server This success, particularly helped by the SAMBA connection has helped further the Linux reputation for reliability. The next step in this progression will be Application Server. Embedded Systems Like the use of Linux inside corporations, this is not a well-documented area, but all evidence shows that it is growing and that Linux is well-suited for this niche. Point-of-Sale is a Linux strong suit. Several retailers and at least one hotel chain use it. The most obvious follow-up here would be to sell Linux POS systems to other large retailers and hotel chains. But in all of these examples, Linux has been an internal project, and there is a lack of large integrators who are pushing Linux as a solution for their customers. II Can the Invisible Hand Pull It Out? It is true that there are some large integrators involved with Linux, but so far none has consented to be identified [except for one, and that only at the end of June]. Further, the involvement is generally limited to providing support for customers' internal Linux projects rather than actively offering a Linux solution. One important closet case, however, is supposed to have been the motivator behind SAP's issuing of a Linux version of its product. On the other hand, the main motor of Linux promotion and development is group of talented individualists who fall into the Innovator class. As such they are largely bereft of budget and influence at high levels of the corporation. The traditional view of Linux is that it will go mainstream because thousands of small integrators will introduce it to their small business customers. But we find there is little overlap between the integrator and Linux communities. One on hand we have Linux experts who have little or no business experience, and are thus at a disadvantage in setting up as integrators, and on the other we have established integrators who, unless they are already at work in the UNIX field, have little chance of acquiring Linux skills without considerable dislocation in their own businesses. Finally, if the target is small businesses, many Innovators will find that these businesses are loath to disturb what works, or to spend money where there is no urgent need; they are reluctant to give up their Windows 95 or 3.x as long as it is doing the job. As recently as a year ago, Linux enthusiasts were expecting some sort of Microsoft disaster to open the breach through which the tides of Linux would rush. The release and failure of Windows NT 5 was predicted to be the "greatest trainwreck in computer history," and the forcing of Windows desktop users onto the large, ungainly, and difficult-to-manage NT was foreseen as another discontinuity that would flip desktop users onto a waiting Linux system. Finally, the Department of Justice was seen as savior that would put a stake through Microsoft's heart. So far not even the first shoe has dropped. Windows NT 5 is still waiting to be born (a cut-back paid Beta 3 of what has come to be called Windows 2000 is now in circulation). And just as steam-engine technology was rapidly improved in the 1930's by the arrival of the Diesel locomotive, Windows NT seems to be overcoming some of its problems. Even Linus has to admit that Linux does not perform as well at some tasks as NT. Nor are desktop users are not going to be forced over to NT when Windows 2000 arrives: Windows 98 SE and one more bug-fix/update after that are scheduled over the next two years. Rather than being goaded by the constant churn of Microsoft updates, users have enjoyed a period of relative quiet, or at least as much quiet as their unstable operating systems will allow them. None of this further habituation in Windows habits can be good for Linux adoption on a broad scale. If Linux were the product of a single company, a coherent strategy could be worked out for its advancement, particularly with regard to the press and public perception. There would be a disciplined concentration on niches, a scouting of opportunities, and a lining-up of enabling technologies and partners preparatory to a push into the mainstream market. Unfortunately the press has not allowed Linux this quiet time. It is too colorful, too exciting, too different to leave alone, and public expectations are being built to a dangerously high point. Without a solid base of integrators able to implement solutions in large corporations, and without the bench depth in applications, Linux is in danger of being mistaken for vaporware. Recently Linus admitted that Linux was developer-driven and not market-driven, and that maybe it was time to pay attention to the customers. We can hope that the Invisible Hand of the marketplace will cause developers, vendors, and customers to converge; what high-level strategy there currently is, is that exercised by the vendors of Linux distributions. It is they who seek out the alliances of complementary vendors, who sign up the integrators, and who make the large corporate sales. The attack on the niches is being carried out by other vendors. None of these parties just named is trying for an all-out assault into the general marketplace, no matter how much the press may talk about it. But there are some important pieces missing which the Linux community has yet to supply that would bring the time of Early Majority closer. I am not going to mention all of them, but instead to bring up once again one that is very important for the goal of persuading the Pragmatist, or Early Majority, that current operations will be enhanced, not disrupted by the new technology. This missing piece is a set of filters that will reliably and accurately transfer data among file formats. This is an obvious candidate for widespread cooperative effort, and should be placed under the GPL for the use of every Linux application, but to date I have not been able to find a project working on this goal. I have found a proprietary vendor who is not about to let go of his filters, but no public project. Surely there is a gap here for a proprietary product that could be licensed among applications. Over the next 18-24 months we can all watch the progress of Linux niches. As an operating system, the progress will be closely related to hardware: embedded devices, machines with multiple processors, and Internet appliances. It is the arrival and spread of applications across customers (rather than in-house implementations) that will mark the first toeholds on the other side of the Chasm. If current projects for ease-of-use development are successful before then, and if there is a reliable and simple solution across applications for easy transfer of data among file formats, and if we wake up to the potential of binary compatibility, Linux has some hope of success in passing into general use. We can only hope the world does not tire of the hullabaloo before then.