I went to a presentation at Stanford VLAB titled “Bringing Great Open Source Ideas to Market” earlier this week. There were some interesting comments made, and I’ll get some of those down here. First I want to mention the overall message however: “2005 will be the breakout year for open source”. Good message, the spirit is right, but I don’t agree. I’m not at all an expert in terms of business predictions and industry tracking, but I have been writing open source since 1992 so I am familiar with parts of it. In particular I am familiar with this part of it. And as someone working within that community I can assure you that every year since about 1996 has been predicted as “the breakout year for open source”. The predictions are always made by people already working on or with open source, and every year something generally described as “business concerns” keeps it from happening. So although I’m always happy to hear it, I’m no longer moved by the prediction. What is happening, and what will continue to happen, is the accelerating adoption of open source. It might look something like a breakout to people within the particular industries that have picked up open source. However open source is already exploding on the market. Has exploded on the market even. It might seem like something new if you normally work with application servers and middleware, but go ask someone in embedded systems about Linux. It’s out there already. It’s just moving from lower levels like the kernel to higher levels like web service infrastructure (much as I’ve moved from embedded systems programming to working at Feedster, the parallel is not coincidental). What’s evolving is the business model to go along with it, which is exactly what Marc Fleury was talking about. So certainly kudos to him on that one.

I’m going to call out some of the points that I found most interesting. Mark at Indexed Forever has already posted with some info also. For me the most salient point was the generations of open source business model. He put down three of them:

  1. No business model - most projects

  2. Packaging - like SUSE and Redhat

  3. Professional - like JBoss and MySQL

He drew the distinction between the packaging companies and the professional open source companies by saying that packaging companies act as an interface to the open source community while professional open source companies ARE the community. Professional open source companies employ all of the major contributers to a project. They can do that because they are the ones that drove the project through critical mass. They might be the original authors of the software, but not necessarily. They do have to be the current wardens for the most popular branch. Because they do have all the major contributors to the project, and deep knowledge about the system, they can provide the kind of support that extends beyond answering user questions and gets into transforming the package to deliver greater value. Which is important because you can’t charge for licenses (past value created) but for support and maintenance (future value that could be delivered). Nice point, it lies pretty well in line with what I’ve seen over time. It certainly explains the great success that MySQL has had, and why there is so much buzz over JBoss these days. I would offer something of a counterexample in Ximian. Although I know some people would call acquisition by Novell success, while others would say it was the reason for the project not diffusing as far as it could have. I’m not saying that invalidates the principle however, just another datapoint to keep in mind as the model evolves.

He made a few other points about scaling the project also. First of all, if you do have a project that has critical mass and you are positioned well, you’ll need to have a good way to filter leads. As an open source based company you’ll have a lot of users, and only a small subset of those will be customers. Traditional companies aren’t structured like this, and it can actually be a great asset to be sitting at the hub and facilitating all the discussions that go on about the project. But you have to have a much different salesforce because of that, not everyone understands the transition. You’ll need salespeople who can filter the huge amount of info and figure out which are leads to profitable customers. That’s much different than the salesperson who understands going out and hunting down a license sale. He also made a point about choosing metrics to measure your progress. That can be number of downloads, cvs commits, number of support emails. The metric should be chosen based on what particular goal you have for the stage your at. Although I don’t think this is what Marc said directly, I took this as calling out the measurable marketing you have at your disposal when you’re running as an open source company. You can see how the community is growing, how your message is making it out there, how people are using the system and what they want done. The increased transparency of open source goes both ways. Not only do your customers have more insight into what you’re doing and how you do it, but you should be able to get more insight into what they need and how they’re looking at solving their problems. If you can take advantage of that it’s a very good thing.