Editor's Note: The following article is the text of the August 5th issue of SuitWatch, Doc's bi-weekly newsletter.
To those of us who attend both the O'Reilly Open Source Convention and LinuxWorld Expo, the two events together provide a midsummer course correction in our evolving understanding of both subjects. This year the two events nearly are consecutive: OSCon was last week in Portland and LinuxWorld is this week in San Francisco. I'm at LinuxWorld right now.
What we're seeing at both is a clarification of changes that have been going on for some time. They're forms of maturation, and they involve a dividing out of roles played in the market.
I see two fundamental divisions. The first is between noncommercial open-source infrastructure and commercial products and services that rely on it. The second is between traditional open-source development communities and the growing population of practitioners for which the main benefit of open-source is free (as in beer) building materials, rather than the deeper concerns (for example, freedom) of the original development communities. These are not opposed divisions but, rather, symbiotic roles in a maturing and proliferating marketplace in which large new species, all dependent on open source, are coming to dominate the commercial space.
In his opening keynote at OSCon, Tim O'Reilly said the new "killer apps" of the Internet age were not applications at all but massive back-end Web service providers whose branded offerings--books on Amazon, search on Google and Yahoo, auctions on eBay--masked their more significant roles as enablers of thousands or perhaps millions of businesses. Significantly, these new Web-based platforms are themselves built in various ways on open-source infrastructure. In other words, open source commoditized infrastructure that in turn supports business that can build on infrastructure that's both standard and cheap.
Tim talked about Clayton Christensen's Law of Conservation of Attractive Profits. "When an industry becomes commoditized, value migrates to adjacent levels", Tim said. So "massive Internet-scale applications" such as Google and eBay acquire "lock-in by network effects". Naturally, questions arise.
What does it mean when an application (such as Google) runs on a hundred thousand servers? What would be the point of "open sourcing" such a thing? What does it mean for Amazon to have ten million user reviews of the products it sells or supports other people selling, through Web services? What does it mean when a single supplier of source data supplies all the raw data in a category--as Navteq does with geographical and mapping data--becomes "Intel inside" that whole category?
If we want to move open source forward, we have to think, "How does its principles, its lessons, translate to this new world in which these massive services are being built on top of Linux?" One of the things I think we really have to get our brains around, when you look at these massive Internet-scale applications, the whole model of what constitutes open source just doesn't work anymore. You can give away all the source for Google, and yeah, it would be really nice to have all their algorithms. But you'd still have to have a hundred thousand servers in order to actually deliver the Google experience. You'd have to have a group of people spidering the Net. You'd have to be gathering all that data. It's a very different problem set than "Give me a piece of source code, give me a compiler and I can have my application." It just doesn't work that way anymore for a very large class of very interesting applications. What does openness mean in that new context?
Here's one thing it means: "People not seen as open source players are leveraging [the ability to contribute]". The people Tim is talking about here are users and business partners of these massive back-end business platform providers.
Users are involved, big time. Users contribute to these companies. They may not contribute as much, or even the same kind of thing, as programmers contribute to open-source development, but there is a form of contribution and participation going on, and we'd be wrong to dismiss it.
At OSCon the star of the Open Source Business panel was an audience member, Phil Moore, a high-level IT executive with Morgan Stanley for the last thirteen years. Phil got up, grabbed the microphone and explained how, as a customer, he was implementing open source aggressively at the infrastructural level, that others were doing the same and that vendors simply needed to face the facts about customer independence. "You can have the desktop", he said to the Microsoft guy. "I don't care."
Yet the users matter, though not necessarily in obvious ways.
J.C. Herz, who has studied the game market for many years, often has talked and written about the high level of symbiosis between gamers and the companies whose games they use. This kind of cooperation tends to be outside the interest scope of traditional open-source developers, even if they also are gamers). But, it's important to the markets growing on top of open-source products, such as Linux, Apache and MySQL.
When I talked to Dan Frye of IBM about the growing population of IT folks that used and implemented open source but were not part of the traditional community, he said this was a group that lived "in a heterogeneous world" and whose concerns were "almost entirely practical" rather than moral or political. Translation: for them, Linux and open source are not movements but building materials and methods. And what they're building, in most cases, is infrastructure.
So infrastructure is where it's at for open source. That's the niche. Beyond that, it's what a teacher said to me once about raising kids: "As they get older and more independent, your influence does nothing but decrease. By the end, all you can teach is values, and you teach those by example."
I described the relationship between teacher and student in two talks at OSCon, in 2002 and 2003. The first was on Infrastructure, and the second was on DIY-IT. In a slide from 2002 titled "The Way It Oughta Work", I showed how infrastructure supports commerce and how commerce in turn contributes back to infrastructure. In another slide, titled "How Hollywood Sees It", I showed how the support ecosystem can be corrupted by commercial interests that attempt to control infrastructure from above, with patents, copyrights and intellectual property law. In essence, going one layer below infrastructure to governance.
Which brings me to the governance issue. It's a biggie.
On Tuesday at LinuxWorld, I interviewed Larry Rosen, the author, attorney and licensing guru, on The Linux Show. The occasion was the launch of Larry's new book, Open Source Licensing: Software Freedom and Intellectual Property Law (Prentice Hall, 2004). Of all the issues that might concern the Open Source community, including the highly distracting SCO FUD campaign, the one Larry said needed the deepest attention was patents. "The community needs a strategy" for dealing with patents, he said. Perhaps not immediately, but eventually.
Two days earlier, Open Source Risk Management, "a vendor-neutral provider of Open Source risk mitigation and coordinated legal defense services", announced that it had found 283 issued patents that have not been court-validated in the plain vanilla Linux kernel. Microsoft holds 27 of them.
In other conversations at the LinuxWorld, Bruce Perens, an OSRM board member, and John "maddog" Hall of Linux International both also told me (and each other, in the same conversation) that patents are a huge lurking issue and that we--the Open Source community that provides the infrastructure--need to deal with them.
There are some good students out there. Who would have thought, two years ago, that Novell would be a prime community citizen? Or thought the same about IBM five years ago?
On Tuesday, IBM announced that it was donating Cloudscape, a compact Java embedded relational database, to the Apache Foundation, to be maintained and offered to the world under the wide-open Apache license. IBM valued the code at about $70 million.
On Wednesday, Computer Associates offered a $1 million "challenge" to spur the creation of an open-source database migration toolkit. The target base is Ingres, which CA is in the process of open sourcing, as part of a larger open-source strategy announced on May 24 of this year. It will be interesting to see if Ingres grows past its reported 15,000 corporate customers.
The word for this is adaptation. For companies that hope to live in a world where open source provides the infrastructure, there isn't much choice.
Doc Searls is Senior Editor of Linux Journal, for which he writes the Linux for Suits column. He also presides over Doc Searls' IT Garage, which is published by SSC, the publisher of Linux Journal.