A report on a series of panels held at the end of O'Reilly's Perl Conference.
I have just returned from the Open Source Developer Day, August 21, held by O'Reilly & Associates at the end of their Perl Conference. The stated purpose of this conference was “How to Set Up an Open Source Business in the Real World”. It had a few hundred attendees. This column is as much an editorial describing my opinions of how Open Source businesses should be run as it is a blow-by-blow description of the OSDD event.
OSDD was an interesting event for me. Since the day's schedule was set up with people explaining how to run a business using the Open Source model, I originally thought it would be an open forum rather than a series of talks. I was wrong. I also expected the audience to be mostly business people. I was wrong again. About 95% of the attendees were already using Linux and very few had on suits.
I am glad I went, as I learned things. However, that learning came from individual conversations and watching the ambiance rather than from actual talk content. Others, with whom I spoke, seemed to feel the same way.
I think Tim O'Reilly had a good idea but the wrong audience. The speakers were talking to the converted. We don't need to tell Linux believers that Open Source software like BIND, Sendmail and Apache virtually run the Internet, as Tim O'Reilly did. We don't need to tell this audience that “Open Source software creates the broadest, most robust software platforms” as Michael Tiemann of Cygnus did, or even that Open Source software creates a culture of open discussion as John Osterhout did. We needed an unconvinced audience who would benefit by hearing all these things along with Bob Young's (Red Hat) “talk about benefits, not features” and James Barry's (IBM) is it a problem or an opportunity story. Good try, Tim—next year, maybe we can get you the right audience.
I wish all software was Open Source. It has the immediate advantage of allowing you to choose your own support rather than having to depend on the software vendor. This protects you if a vendor vanishes from the market, and it also forces the vendor into a position of providing good support or losing business to another vendor.
We have seen the most popular Linux distribution change from Yggdrasil to Slackware to Red Hat. This was certainly less painful than the transition of software from IBM to Microsoft. It has also meant that other distributions such as Caldera and S.u.S.E. can stay in the market, and even gives them the chance to become the new market leader.
That said, I don't want to go to IBM or Oracle or any other huge company and say “Open Source is the answer.” I believe it is, but we don't have the ammunition to make that statement today. Besides, we can't afford to be exclusive. If we were, we wouldn't have Informix SE, Applixware, StarOffice or many other software applications in our camp today. Yes, I would like to see these companies go to Open Source, but I would rather see them do it on their own schedule and because of market conditions rather than from being sold on the concept by fast talking.
At a business models panel, IBM talked about its open involvement in Apache, and John Osterhout talked about Tcl and his company Scriptics, which will keep the Tcl core free but charge for various enhancements. After they finished, we again experienced how closed Open can be. Richard Stallman went to the audience microphone and embraced IBM's involvement in Apache and called Osterhout's company a parasite. Why bother? As Tim O'Reilly said in an effort to terminate this speech, the market will determine who is right.
Open Source should help prevent monopolies. I say should because I see a potential problem. When Eid Eid was Chief Technical Officer for Corel Corporation, he told me that as soon as Corel purchased WordPerfect, Microsoft stopped releasing information to them about operating system internals and future plans/changes. Microsoft did this because Corel had become a competitor.
While Open Source might have helped, it still wouldn't prevent a distribution vendor from adding a feature they shared with their partners but not with other vendors. Once the distribution was released, everyone else could get the information, but they would have to play catch-up.
Contracts where only one company (generally a distribution vendor) can sell an application fragment the Linux market. If a certain application runs only on distribution A and another application runs only on distribution B, then the user is forced to choose between the two applications. We should demand compatibility between Linux distributions in order for the Linux market to expand and not become a monopoly.
One other potential monopoly scenario is the act of buying out the competition. Look at Microsoft's history to see examples of how this works. Microsoft bought the right to ship a product from another vendor (the C compiler from Lattice) until their homegrown product was ready to sell, invested in a competitor (SCO) just in case, bought a big chunk of a new technology (Web TV), and ported their applications to another operating system (Macintosh OS for now; expect Linux in the future).
While Open Source doesn't eliminate monopolies, it certainly makes them harder to create.
Over the past few months, three different attempts at a Linux standard have been made. The good news is that Dan Quinlan of Linux Filesystem Standard fame has taken over as chair of what is now a combination of the first two standards attempts.
This effort has support from a reasonable cross section of the Linux community and, with Dan at the helm, I expect support to grow. You can read more about it at http://www.linuxbase.org/.
For my two cents, to make the effort work there has to be a high level of involvement from the application vendors in order to ensure that applications will run on all major Linux distributions. I think a face-to-face meeting of all participants is needed to get this effort rolling.