Daniel Lyons interview of Larry McVoy might raise the wrath of open source purists, but he raises some good points about the difficulty in building a software business around open source ideals. Unfortunately, his conclusions based on his own experience with those difficulties don't quite stand up to inspection. Open source is just as innovative as closed source, and it is definitely harder to sustain a business with it. In essence, Larry is both right and wrong. Just not about the same things.
To say that open source as a business model in isolation is unstatinable is probably right. It is extremely difficult to maintain margins on product that keep the business profitable and growing when the product itself is free. I have looked, in the past, at starting my own software companies based on open source. It is incredibly difficult to build an application or tool or other software component that you give away – and let others maintain on their own if they choose – and find a steady revenue stream based solely on that component. But Larry likens his business – BitMover, which makes a singular product (or perhaps a small suite) – with red hat. Where pundits like Larry go wrong when comparing a software business to Red Hat is in assuming Red Hat is in the business of creating new components. They aren't. They are an integrator at their heart. Instead of integrating one or more components with an operating system – the old Microsoft model – Red Hat integrates thousands of componets including the operating system to create a distribution. It's not an apples to apples comparison, so Larry's conclusion that Red Hat's services model assumes Red Hat delivers crap in order to sustain the model falls apart. Services are not limited to fixing buggy code for Red Hat, whereas for a small development house services are very much centered around bug fixes and feature enhancements. Integration is different than pure development. Larry doesn't appear to see it that way, however.
Larry also concludes that innovation doesn't happen with open source and that most innovation happens in closed source development. This argument really depends on your definition of innovation. One definition I found online says simply “the act of introducing something new.” Is a source code control system something new? Or is the way the system implements that functionality new? If the latter, Red Hat innovates constantly by implementing desktop features, server services and customer support in new ways. But is a new implementation really innovative? Its different, but is it what we really mean when we say it's innovative? I don't see bitkeeper as innovative, but that doesn't mean its not a good solution to the problem it tries to address. Innovation in source code control systems – to me, at least – would mean that source is no longer “controlled”. Think of a chaotic system that perfoms the exact same functionality. I can't imagine how this would work – so if someone implemented it, I'd have to say that was innovative.
The problem here is that, if you take my definition of innovation, you can't find too many innovative products in either closed or open source communities. Why? Because by this time in the software world no one stands alone in their work. We all stand on the shoulders of others. Much of the software today is extension or reimplementation, but little is really new. Innovation, then, is rare. Almost as rare as quality in software. So if you want to compare closed source to open source, innovation isn't the place to look. Look toward quality. At least there you can find some quantitative measures for comparison that rely far less in interpretation. To say that innovation would go to zero in all open source world assumes that truly new ideas come only from behind closed doors. In fact, truly innovative ideas often happen as accidents. To say that working with open source can't incur accidents that lead to innovative ideas is just plain silly.
Besides accidents, if we talk about BitKeeper being innovative from the perspective of introducing a “new” way to manage code, then we have to include many innovative open source projects: mythtv (which is starting to evolve into a commercial solution with various business partners), PHP (which has temendous support from Zend), and of course Mozilla. Each of these are innovative in their own way. Throw in the plethora of languages (Perl, Python, Ruby and others) and the list of open source innovations gets quite long. Again, we're looking at this from Larry's point of view: we assume a new source code control system is innovative because of the way it implements souce code management. These other projects are innovative, then, because they introduce new ways of doing the things they were designed to do.
Larry's contention that building an open source company that can be sustained with open source based revenue can't be done doesn't hold up either. The obvious companies – Red Hat and mysql, for example – are surviving at a minimum and actually thriving in some cases. I've already mentioned Zend, which provides add on features like debuggers and IDEs for the PHP environment (not to mention working on the language engine itself). The fact that the number of well known successes is small could stem from the difficulty in finding a way to make an open source business a success. But that doesn't mean it can't be – or isn't being – done.
Futher on this interview Larry comments about open source being like blueprints and a doctor's bag instead of a doctor's service to a patient. In a sense, that's true. At it's core, open source provides raw materials. The problem is that Larry assumes that those raw materials are the only way the patient can get the service. Red Hat, Novell, and IBM are showing that's not the case. So this argument, too, just doesn't hold up. What is different is that, unlike medical services, if you really *want* the raw materials you can easily get them. Freedom of choice is something too few open source advocates point to when speaking to the public. I wish we could get over the “free beer” vs “Libre” stuff. But that's fodder for another day.
The last thing in this interview I took exception to was Larry suggesting that open source innovation would dry up if commercial funding stopped. There is a lot of funding coming from hardware companies, that's true. But to say open source innovation would go away if that funding went away is ludicrous. First, there are many successful open source projects that are not hardware bound that have little to no funding. The gimp is probably the most well known example. Pull funding from that project (currently limited, for all practical purposes, to some funding for development of GEGL) and the project would continue unfettered. They are not dependent on it. And GEGL is innovative, even from the point of view of being a completely new idea for how raster graphics applications should be designed.
But what really struck me is how one sided Larry sees the funding issue. Let's suppose he's right and open source is completely dependent on such funding. If it goes away, the project goes away. So what makes him think this never happens in the closed source world? I worked with Nortel, dell, Samsung, and many otheres too. When funding goes away in closed source environments, innovation goes away much faster than in the open source world. The reason? At a company you go in to get paid and when they stop paying you don't keep working for them. Open source developers still scratch itches. Innovation survives funding droughts in open source. It doesn't in the closed source world.
When you read this whole interview and think deeply about what Larry is trying to say you realize that Larry is upset that making money with open source is hard. He's right about that. But having come from three startups in the past 5 years that were completely closed source, I can tell you that it isn't easy in the closed source world either. No one said running a business would be easy. I can't sell – I don't like the personal interaction that much. So I don't run a business. But to blame difficulties in business on open source in general is really digging for excuses.
If it were easy Larry, everyone would do it.