Enough of this "open core" confusion!

There's quite a flurry, in the blogosphere, of discussion about the notion of "open core" software. Some see a threat, like Simon Phipps' "Open Core Is Bad For You"; Matt Aslett urges some perspective on the debate, reminding us that "Open core is not a crime," and in a comment there, Mark Radcliffe worries that venture capitalists might be put off by "open core demonization." 

These are all smart, experienced guys, but they seem to be running in different directions. Makes you wonder if they're all thinking of the same thing when they say "open core"! Maybe we need to work that part out, first.

I think what we're missing here is an exploration of the delicate line between "crippleware" and "added value." If your so-called "open core" product isn't quite up to the job at hand, if customers really need to buy the commercial license in order to get functions they need to execute the basic goal of the software, then you deserve the "faux-pen" label and shame.

But, on the other hand, if the open parts accomplish their goal fully and well (and, of course, if they're truly open, to contributions and priorities coming from the community), then the open-source product deserves to be assessed on its own terms. If there are also commercially licensed, or even proprietary/closed things associated with it that together accomplish some larger goal, that's a different product, not a betrayal of the open one.

In between, there's naturally some gray area we can debate to exhaustion, but a clearer idea of the purest distinction might help.

Obviously, CollabNet occupies a particularly clear-cut corner of this space. The Subversion project, initiated and supported by CollabNet for nearly a decade, is unquestionably "open source." The project sets its own direction, listening to CollabNet's needs but completely open to other ideas. And while much of the guidance and labor have always come from CollabNet, much as well (and in truth, more) comes from the many non-CollabNet contributors. There's no "faux" in Subversion's "open source."

At the same time, though, it's undeniable that Subversion is crucial (dare I say "core"? šŸ˜‰ to CollabNet's mission. And in addition to Subversion and a number of other open products, CollabNet certainly does have non-open products, often built directly upon these open-source foundations. Does the existence and completely proprietary licensing of CollabNet TeamForge make Subversion somehow "faux-pen"? Hardly!

What I think would better frame all this is an expansion of the "open core / open complement" dichotomy (for which, I believe, Matt Aslett may take some justified credit). The category of "something crucial to the larger product, and yet also fully useful on its own" simply doesn't fit neatly into the "core/complement" split. Lumping these cases in where they don't belong just creates confusion. And when, like Simon Phipps, you have strong things to say, you want to aim them with clear categories.

Tagged with: ,
Posted in Subversion
4 comments on “Enough of this "open core" confusion!
  1. Henrik Ingo says:

    Hi Jack
    This is an interesting post. It is interesting because as far as I can tell, Collabnet/Subversion is not engaging in any of the objectionable activities that the open core debate is criticising some vendors for. I blogged about the details of these practices in my blog here: http://openlife.cc/blogs/2010/july/so-if-he-doesnt-call-himself-open-source-vendor-then-everything-fine
    …but to make it short: AFAICT Subversion is purely an open source project (it is not even owned fully by CollabNet?), there is no proprietary software using the Subversion brand (since that is what you call CollabNet), and there is no proprietary software being distributed at subversion.org.
    The fact that CollabNet then creates proprietary commercial products upon subversion is a separate space, and the fact that CollabNet has been a strong supporter and leader of Subversion could even be seen as a positive excercise. (A “purist” could of course oppose any and all closed source software, but that’s a different topic altogether and should not be confused with the ongoing debate.)
    So while you of course have any right to stand up and defend the likes of MySQL, Zimbra, JasperSoft, etc… there seems to be no reason for you to defend Subversion or CollabNet, since nobody is accusing you of anything. (It is left as an open question whether your model could be in the “open core” bag, but in any case it would not be of the objectionable kind of open core.)

  2. Thanks for your thoughts, Henrik. Im glad you find nothing objectionable in either Subversion or CollabNet!
    I wasnt particularly interested in defending either CollabNet or those other vendors you list. Rather, I was trying to move the conversation a bit towards how to do it right, and away from confrontation.
    If I may restate the objections raised by your blog and others, the offensive thing here is, as I understand it, obscuring the line between open and proprietary source. Your blog focuses on one form of obscurantism, mixing the two types into a single product and brand. But the obscuring also happens in another way: labeling all mixed-mode work with the same name. You left as an open question whether [CollabNets] model could be in the open-core bag, but thats the question Im opening. My belief is that open core is the most appropriate way currently available to describe the CollabNet / Subversion relationship, but I wouldnt choose to use that term because of the negative associations with other models within the space. I want a new term, not only to stitch onto my own golf-shirts, but to recommend to other companies looking for a successful and inoffensive model.
    Instead of excoriating people who use terms badly (when no better terms are available), lets set the terms of discourse so they identify and reward constructive models.

  3. Henrik Ingo says:

    Ah, I understand your thinking now.
    I do see the need for that too. In the database space we have EnterpriseDB based on PostgreSQL in the same situation. It is essentially open core, except that EnterpriseDB does not own PostgreSQL and the brands are separate. Since this is a significant distinction, it may be practical to have another “label” for this model that would also include Subversion/Collabnet. This would of course be in your interest now, since open core is at risk of becoming a negative term (for those that care about open source).
    Matthew Aslett also made a great point in an earlier post, that the fact that IBM, Oracle and the lot use Apache software in their proprietary products, do not make these companies open core companies. This makes sense, and in that case the same argument would apply to CollabNet.
    (Btw, “objectionable” could of course be many things. I could come back here arguing that people should use only 100% open source / free software, there are people that do that. But that is not what the current debate is about.)

  4. Yes, thats the sort of thing I was shooting for. Good examples, thanks. Im not familiar with the nuances of EnterpriseDB/PostgreSQL, but I think Collabnet/Subversion is considerably closer to the open core / open infrastructure line than {IBM,Oracle,PrettyNearlyEveryoneOnEarth}/Apache: CollabNet doesnt own Subversion, but CollabNet is and has always been the clear and principal corporate sponsor of Subversion, whereas the Apache Foundation model has a longer list of rough equals.

Leave a Reply

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