Tim O'Reilly interview - O'Reilly on Linux (part 1)

Posted on 14 Jul 2003 at 12:30

This sounds like a victory for open source, but it could easily be a defeat. Despite employing a cadre of open-source developers, the companies building these next generation applications don't necessarily see themselves as a part of the open source community, and in fact are often fiercely proprietary. What's more, existing open-source licences are rooted in a model where software is distributed to individual computers, while these new applications are simply 'performed' on remote systems. And even if their source code were given out, you wouldn't be able to reproduce the application without the massive database and the massive customer community.

And finally, it's interesting to note that the interfaces to these programs are not strictly software, but something that I've been calling 'infoware' - the Amazon interface, for instance, includes not just software but a constantly changing array of content, much of it contributed by parties outside of Amazon. It's quite remarkable when you come to realize that more people have contributed to the construction of Amazon than have contributed to the construction of Linux. I believe that more than 1 million people have contributed reviews, listmania lists, ratings, and other elements of the Amazon user interface. But of course, all this belongs entirely to Amazon.

And because the open-source community hasn't understood the paradigm shift, it hasn't made it a priority to reach out to the developers of this kind of application. I've been speaking on this subject since 1997 - see tim.oreilly.com/opensource for some of the talks and interviews I've given on the subject, including 'Hardware, Software and Infoware' (my 1997 talk at the Wurzburg Linux Kongress, and 'The Network Really is the Computer', my 2000 Java One keynote, as well as a number of more recent pieces.)

Fortunately, I've made some progress working this issue myself. I was able to persuade Amazon that prosecuting their 1-click patent was bad policy, that it is important for them to be friendlier towards the kind of give and take that built the open platform that they rely on. I was also able to persuade both Amazon and Google to open up APIs to their data, which allows developers to re-use their data in new, creative applications. And I'm starting to work on the issue of who owns the data, and opening that up a little more as well. For example, I've started talking with Amazon about supporting mechanisms that would allow people to write reviews on their own blogs or web sites that would be syndicated to Amazon, rather than simply owned by Amazon and run only on their site.)

And at the Open Source Convention in July, I'll be leading a panel entitled 'Do We Need a Bill of Rights for Web Services?', in which I explore what kinds of analogies to open source we might need in an era of Internet-scale databases and web services based on them. (You may know that CDDB, the song/artist database now used by most desktop music programs and CD burning software, was originally a free database built collaboratively by Internet users, and was then 'taken private' in much the same kind of land grab that started the free software wars against the Lisp developers that Richard Stallman was involved in.

Unlike Richard, I don't think that proprietary is necessarily bad. But I do think that we need to find the right balance between free and proprietary, and how best to build interfaces that users are encouraged to contribute towards. And to do so, we need to think deeply about the shape of the future. And in the future I see, software per se has become much more of a commodity (much like hardware is also a commodity now), with value capture increasingly driven 'up the stack' towards information providers. So the frontier of open source is actually open data, not open code.