"Open core, open complement": something's missing

A lot of people are struggling, just now, to understand how to make "the open source magic" into something repeatable (or anyway, as near to repeatable as other business models). 

  • Dirk Riehle, of SAP Research, has captured and codified a wide sweep of thinking on The Commercial Open Source Business Model
  • Matt Asay, of Alfresco and cnet, has given us an Open Source Continuum, a handy framework for open-source related commercializable license models
  • Eugenio Capra (Politecnico di Milano) and Anthony Wasserman (Carnegie Mellon West) have provided A Framework for Evaluating Managerial Styles in Open Source Projects
  • Guy Martin, in this blog, has drawn out the distinction between community "management" and "leadership" (with an appeal for both) 

It seems to me, though, that there's still something missing. I guess it's easiest to discuss in the context of Matt's Continuum. He talks about openness of the process and code relative to "core" and "complement", but there's really a third aspect as well: infrastructure.

As Matt points out, some open-source companies open up their core functions: MySQL and JBoss are two examples where the core functions of the product are open-sourced, and the money is made on on "complementary" aspects, which might be service, add-ons, or support. Matt points out that this approach tends to favor the disruptive entrant. Conversely, some open-source companies open the complements while holding the core proprietary; Microsoft and IBM have both demonstrated this strategy.

But I think the most successful open-source companies have been built around another model: opening the infrastructure:

  • Linux is everybody's favorite example of a wildly successful open-source project, but the success of Linux is best measured by all the many different ways it's used, by someone else, to do whatever they want: it's infrastructure to web sites and telephones, laptops and supercomputers. Many companies have contributed to Linux, and are now profiting because it exists, by building something else on top of it. Linux is, of course, a big place! There are also Linux companies making money by more ordinary "open core / proprietary complement" strategies, but the success of Linux as a whole is far greater than even the sum of these.
  • JBoss is very similar "open infrastructure" example. Yes, there's a company there, working directly to profit from the product, but the success of JBoss is vastly more than that: it's all the companies that have built applications on top of it. Infrastructure. 
  • Apache httpd? No company there at all, but a community of companies collaborating to create infrastructure, infrastructure that completely dominates its market. 
  • CollabNet's sponsorship of Subversion is like this as well: much as we believe in the project, the resulting product is only a piece of the infrastructure for our collaborative development and distributed ALM product.

This has implications for inner-sourcing as well (naturally: everything does!): when you're considering how to benefit from open-source techniques within your organization, you're most likely to benefit from "open infrastructure," because that's the model that maximizes all participants' ROI.

Posted in Subversion
3 comments on “"Open core, open complement": something's missing
  1. Dirk Riehle says:

    Uhm, here is the link for the share of wallet wars and open source’s role in it:

  2. Dirk Riehle says:

    Dang, it looks like it swallowed my earlier comment?
    Anyway, I was saying that subsidizing an infrastructure component is not a revenue generator, it is a money sink. You only make money by taking revenue out of the software layer addressed by the component, e.g. Subversion, and shifting it into another layer, e.g. a software forge. You spend money on one layer to earn more in another layer. That’s why IBM is subsidizing Linux, etc.
    Here a longer explanation: http://dirkriehle.com/2008/06/18/open-source-in-the-share-of-wallet-wars/
    Thanks for the blog post and the argument, Jack!

  3. Well, I disagree that open source is purely a money sink: rather, it’s an investment. You get out more value than you put in.
    For example, we’ve several times tried to calculate how much we spent sponsoring Subversion, versus how much value resulted. That’s a pretty hand-wavy sort of calculation, of course, but we’ve done it several ways, at several stages of the project, and the answer consistently comes out about the same: CollabNet invested about 25% of the cost of developing Subversion. But we get 100% of the value (and so does everyone else, of course). We deliberately reduced the cost of a key component by 75%, so we could spend that building the rest of the product.
    To some degree, that’s covered in your “wallet wars” post, to which you’ve linked, but you show this “reduce the costs” phenomenon only for operating system and database, components that most of us out-source anyway. In this example, we applied the magic of open source development to reduce our *own* costs: it’s a “money source,” not a sink!

Leave a Reply

Your email address will not be published. Required fields are marked *