The Long Tail and The Code Keiretsu
By webmink on Feb 06, 2006
I spent the weekend pursuing butterflies and, as always, the membrane between lives got breached thinking about them. Monarch butterflies migrate south for the winter and settle on the California coast in a small number of locations, providing safety in numbers and making the most of the coastal habitat.
It's a bit of a stretch from there to here, but I will anyway. As I spent time last week at Sun's Analyst Summit, I frequently found myself making a link between Japanese culture and open source in the most hacknied terms as part of a discussion of the value that can be found in open source communities. It's a little embarrassing to admit as both terms I combined have a fine history of hypesterism, but I'll set the snake oil aside for a moment and tell you my thought-journey.
First, the naming of parts. The phrase "long tail" will be pretty familiar, either as a connoisseur of James Gosling's photographs or perhaps as a student of the Wired article by Chris Anderson. The thought describes the way that a business such as Amazon.Com can make as much - or even more - money by selling one or two of a large number of low-volume items as it can by selling many of one or two 'popular' items. The consolidation this produces in the market overturns the traditional economics of local retailers. The name "long tail" comes from graphing products in order of decreasing popularity (the logo on Chris's blog illustrates nicely).
The other word is "keiretsu" and may be less familiar. It's a Japanese term describing a set of companies with interlocking business relationships and shareholdings, and it's well described by Wikipedia. It's a less formal structure than a business group and less focussed than a cartel. It's more an extension of the Japanese family concept into business. The members have a 'family duty' in common and where possible trade within the family.
So how do these ideas relate? By combining them, we have a potential model for certain kinds of open source community such as GNOME, Apache and Eclipse. An open source community is bound together around a source code commons, with many different developers using the code to build software works that meet their own motivational needs - revenue products, social works or personal interests. Apart from that (important) union, they are otherwise unrelated. They contribute back code to the commons as they go along, not solely or even mainly out of duty but because to do otherwise increases their support burden as the community moves ahead.
This community, gathered around a source commons, might be termed a code keiretsu - many parts acting independently but to each other's benefit.
This "code keiretsu" is unlikely to include direct competitors, but will include a diverse array of interests serving many points in a target deployment system. In the case of the evolving OpenSolaris community, there are already members serving diverse goals. Addressing a large potential market of data centre users there's Sun creating Solaris Express (the early version of the next release of the Solaris operating system) from the OpenSolaris commons. Then, aiming at a completely different group there is the team building the Polaris port to PowerPC, with an emphasis on embedded use. In another, different market, Nexenta is engineering a version of the GNU operating system using the OpenSolaris kernel instead of the usual Linux kernel (and presenting many of the more general-thinking population with the challenge of what to call Linux when just Linux is replaced).
All three of these uses (and there are plenty of others, now and to come) can be represented as different opportunities in a 'long tail'.
So to the phrase. I found it helpful last week to explain open source communities like the ones I mention above as "code keiretsus around a long tail of applications". As the long tail develops, more and more developers join the keiretsu, each targeting a different area of the long tail, each benefiting from the code commons by being fast to their market yet with commodity-level costs. I believe we will see the software market increasingly dominated by these long-tail code keiretsus.