It goes without saying that something really big is happening with Linux. It strikes me there are some parallels to what is happening. One is what happened with PC in the early 1980’s and the other is what happened with tcp/ip during the middle 1990’s.
Don Estridge at IBM made a very important decision during the very early stages of the development of the PC. The PC would be built using industry standard components and the key interfaces would be open. Being open meant that it would be easy for OEM companies to design and build printers, display units, and other peripheral devices that would be compatible with the PC. This enabled an entire new industry. Venture capital flowed. People became empowered to do their own computing. The rest is history.
By the middle 1990’s there were probably fifty networking protocols. They all had their strengths and weaknesses. Many people believed that IBM’s Systems Network Architecture (SNA) was the best and most major corporations had adopted it. But the fact that it was better didn’t matter because the world had voted and the winner was tcp/ip. tcp/ip had many shortcomings but among its great virtues is that it is easy and open. Because the protocol specifications were widely available and not proprietary people found it easy to deploy and interconnect previously incompatible systems. It grew like Kudzu. Venture capital flowed. People were now empowered and connected. The rest is history.
In 1999 Linux was considered great by its followers and a novelty by those on the sidelines. By the end of 1999 people began to take it seriously. Because the source code was readily available and the Internet enabled global collaboration it was easy to add new features and to support new systems. Venture capital flowed. People were now empowered, connected, and had many more choices. It is too early to say “the rest is history”. But what is possible?
The easy aspect of Linux may turn out to be profound. There are multiple dimensions to it. One is the spread to new devices. The widely available information about Linux and the power of networked collaboration is causing support of Linux on diverse platforms to spread rapidly. In the last few months it has become a reality on things ranging from tiny PDAs to IBM’s largest System 390. In the past the inhibitor for a new platform to flourish with a new operating system has been the availability of drivers. Drivers are those magical relatively small bundles of programming that enable your favorite mouse or modem or printer to work on the new platform. Nice that Linux may work beautifully on my ThinkPad 570 but if there is no driver for the 570’s internal Lucent 56K modem then I can’t use the ThinkPad. Along come companies like collab.net which solve the problem. The computer or modem manufacturer can go to collab.net with their specifications and collab.net can go to thousands of registered Linux developers and produce a bid to write the driver. Even major companies have limited resources and difficulty meeting all the priority demands. In theory the Linux community at large is unlimited.
The second dimension to easy is second order leverage. The limiting factor in today’s information technology industry is skills. Venture capital can fund new companies and ideas but it can perform no magic in creating skills. There are three sources for the skills needed; grow them, import them, or divert them. Growing them is hopefully underway in the educational systems of the world. Importing them from where there is surplus to where there is shortage is happening but is often impacted by political concerns. What about diverting them from less important areas of work? Like porting. How much of the information technology industry and user base is devoted to porting software from platform A to platform B? I have no data on this but I suspect that every information technology company and most information technology customer has a non-trivial amount of resource involved in porting. How about if the operating system began to homogenize. It won’t happen overnight but it may surprise us all how fast it happens. The leverage is there.
So where is Linux headed? Today’s most powerful operating systems have clear advantages over Linux in the areas of scaleability, security, manageability, and general reliability. As companies like IBM contribute key technologies (such as the journaling file system which IBM contributed in early February 2000) to the Linux community things will get even better. Not overnight but over time it seems likely that Linux will inherit the necessary attributes to become mission critical. In the interim Linux will expand as a development environment and clever techniques by those vendors who embrace it will evolve to enable Linux developed applications to run in the robust environments of more mature operating systems. Eventually things will converge. People will be empowered, be connected, and have a broad set of information technology choices to support them. The rest will be history.

• The Cathedral and the Bazaar (Eric Raymond on Linux)