As you might expect of someone who studied under Professor Lessig, Matt’s both thoughtful and considered in his arguments, while holding strong opinions on the question in general. Further, he has the benefit of having experience on both sides of the open source divide, having been with Novell prior to his current position at Alfresco. In other words, I may disagree with him, but I’ll certainly do so respectfully.
Here are the principle areas where I’d take exception:
Does The Definition of Open Source Matter?
One of my primary difficulties with the current debate, as outlined in my earlier piece, is the relevance outside of those communities with immediate and direct interests in open source. Matt took exception to this, saying:
Customers do care. They’re not stupid. They derive real, tangible benefits from open source. Else they wouldn’t be pumping cash into open source companies and into internal development of open source expertise and code. Open source matters in fundamental, enterprise-changing ways.
Don’t believe me? You can ask the IT executives themselves. Come to this year’s Open Source Business Conference to hear the CIO of H&R Block (an active user of a range of open source projects, from Alfresco to Zimbra to Linux to…), the CTO of E*Trade, the VP of Product Development at Washingtonpost.com/Newsweek Interactive, and other IT executives talk about tangible benefits they receive from open source that they cannot get from proprietary software.
As I read this, Matt seems to be rejecting mine and others’ contention that customers aren’t interested in a philosophical debate – they’re most interested in a solution to their problem. That the very same customers that employ open source as a solution to those business problems – the H&R Blocks, the E*Trades, etc – would care enough about open source to follow the debate over so-called hybrid approaches, attribution licenses and so on.
I disagree for a few reasons. First, I think it’s interesting that one of the firms mentioned – H&R Block – is using a product, Zimbra, that purists should have a problem with. Second, I think OSBC has to be considered a self-selecting audience; by definition those are customers are already interested in open source, as opposed to the population at large which if they’ve heard of open source probably have a poor understanding of the economics, business models, and so on. Which brings me to point three: my experience with the enterprise population at large is that the level of understanding and education with respect to open source is poor.
For every E*Trade, there seem to be a dozen or more firms that either a.) rely on open source without realizing it, b.) think it means that everything is free, c.) are anti-open source, or d.) have never heard of it. Talk to open source software vendors that are considering a migration to the GPL, as an example, and see what they tell you: odds are good that it’ll be that their customers are terrified of the GPL. Usually for no legitimate reason, and in spite of existing investments in GPL products such as Linux or MySQL, but terrified nonetheless. As a result, I find it difficult to believe that the enterprise population at large is even a fraction as interested in debating the philosophical differences between open source business models as the folks that follow it closely. Their primary business in life, after all, is not open source (unlike many of us) – it is their business.
Put more simply, I think trying to pin down an actual definition of what open source is is an exercise of limited utility if the end goal is to sell to enterprise customers. If the goal is to determine a session agenda for an open source conference such as OSCON or OSBC, of course it’s relevant.
Should The Definition of Open Source Matter?
If we can set aside the question of does the definition matter, the greater question is should it? Enterprises, after all, are much like people: they don’t always know what’s good for them. Matt, indeed, seems disappointed in those of us questioning the importance of this discussion, saying:
I’m actually a little disappointed that otherwise highly astute people would presume that this is an immaterial question. Freedom in IT matters, and how you define freedom (or, in this case, open source) matters, as Michael Tiemann reminds us, or else you’ll never actually reach your destination.
I suppose it’s easy to forget all that open source is designed to rectify in enterprise IT: lock-in, misalignment of vendor/customer interests, inflated costs, buggy software, payment for real value and not merely licenses, etc. All these things go away the minute you close the code or, at least, become more difficult to achieve. Period.
Irrespective of whether this objection was aimed directly at me, I feel compelled to respond, or at least clarify. While I personally don’t see tremendous value in this debate, it’s not because I don’t think the questions matter. I tried to be clear on that, saying:
Q: Accepting for the sake of argument that customers won’t care about these issues, do you think that’s smart? Should they care?
A: To the extent that it impacts what they can and cannot do with their software, yes, I think they should. But frankly, customers have enough trouble with even basic open source principles: asking them to digest the implications of attribution licenses, as an example, I think is too much to ask at the current time.
In other words, I do think Matt is correct, and that the question is in fact important. I just don’t think it actually is that important to everyday customers, and that more wouldn’t see the distinctions between hybrid and pure open source firms than would. Or that most customers would see little cause for concern in an attribution licenses.
In my experience, anyway.
Is 100% Open Source Always the Best Approach for New Firms or Projects?
In a separate piece examining Alfresco’s transition from a hybrid model to 100% open source, Matt raises an interesting question – one I feel is quite relevant to discussions of what is, and is not, open source. He says in part:
Perhaps protection was important early on. This is the big issue that I don’t know how to resolve. We felt over time that we didn’t need the proprietary trappings. But it’s possible that this coverage was just enough to help us launch and get into “orbit.” I think for companies that take a more organic-growth approach, as Larry Augustin espouses (and with which I largely agree), early protection is less important than early community growth. And I want to believe that community growth is always more important.
This is, to me, a critically important question in terms of deciding where you stand. To some extent, I see some of those on the purist side of the house attempting to define open source in very strict terms, as if open source is somehow all or nothing.
I personally don’t agree with that approach, because it doesn’t seem to recognize that projects or companies may occasionally need to evolve towards pure open source rather than begin there. Take OpenSolaris, as an example. It’s a question of license rather than degree of openness, but the principle is the same. I’m on the record as saying that their choice of the CDDL – while unpopular amongst GPL advocates – was essential to the projects success. Being GPL, the equivalent of being pure open source in this example, would have – in my opinion – jeopardized the viability of the operating system (many would disagree). Now? I think the project is established, and could move with relative impunity to a GPL based license. Being slightly less open, then, had benefits in the short term without precluding longer term openness. This also seems to have been the case with Alfresco. And who’s to say that SugarCRM or Zimbra, once they are better established and more mature, won’t make all of their code available rather than just the majority of it? That’s Second Life’s plan, according to all reports.
Denying firms that contribute to the greater open source ecosystem the benefits that accrue to the open source brand until such time as they’re quote unquote pure open source seems shortsighted to me, given the ancillary benefits they can provide along the way.
If 100% Open Source is the Superior Approach, What’s the Problem?
The last objection I’d raise is simple. Matt, along with a host of others, have been very outspoken in their optimism and predictions of success for a fully open source approach, advising proprietary companies to burn their boats. In this piece, he summarizes a recent win for Alfresco that is directly attributable to the distribution advantage they enjoy as a result of being open source.
Given that level of confidence, I’m not sure what the basis is for the apprehension regarding hybrid or attribution licensed open source. If the approach is indeed superior – and I’d agree that in most cases it is – what does it matter if certain companies don’t embrace the purist mindset? If the theory is sound, the non-pure open source firms will either migrate in that direction of their own accord once they perceive the benefits – much as Alfresco has done – or they’ll eventually be disrupted themselves by competitors that are willing to go that final step. The worst case here just doesn’t seem that bad to me, but perhaps I’m missing something.