CNET también está disponible en español.

Ir a español

Don't show this again


Resurrecting Marc Fleury

I spent four hours with Marc Fleury tonight, talking through lessons learned from JBoss and his future plans. Marc is, perhaps for the first time in a long time, very comfortable with himself and his life. He seems happy. Here's why.

Marc Fleury Heise

I was fortunate to spend most of my evening here in Atlanta with JBoss founder Marc Fleury. The second-to-last time I had seen Marc, he had cursed me (literally) for hiring away one of his employees (that employee joined us tonight--all is forgiven :-). The next and last time I saw him he told me he was going to bury my company (or, at least, compete with us).

Tonight, however, Marc was mellow. He strikes me as someone who is at peace, perhaps for the first time in years. He looks great. Seriously. Maybe that's the two weeks he just spent in Spain with family, but I also think he has managed to let money be good for him, rather than a cancer.

Yes, the "old Marc" occasionally flared up, and thankfully so. The industry was better for having had Marc in it. But now he definitely comes across as someone who is looking forward, not backward. He pointed out his "Wicked" T-shirt, but it's from the Broadway musical, not a death-metal band. He's changed.

We talked about a range of things, from his kids (he loves them) to his DJ'ing job (still at it). Oh, and we also plowed through his lessons learned from JBoss.

What are you up to these days?

Nothing. I do a little bit of science research (mostly biotech), a little bit of music, and a lot of time with family. Oh, and I spend a lot of gaming, Command and Conquer 3, mostly. 25 hours per week.

I'm reclusive. I don't even normally carry a phone. I turn down the ring on the phone so that I can hardly hear it. I don't like the distraction. I've disconnected almost completely. I'm spending my time thinking. I don't feel any need to rush back into things. I'm not looking to do JBoss II.

Nathalie and I are still in the same home. Same car. Same PS3. We'll probably buy a new house at some point, but we're not too worried about rushing it. We're happy.

Isn't that how JBoss started, too? You just sort of started writing a project that just happened to turn into a $350 million acquisition?

Not really. When we did JBoss, there was no real plan of how to make money, but it was definitely our aim to start a business. Back in 1999 I left Sun and started working on an EJB project. We burned six months on a horrible business model: hosted JBoss. A partner at Sequoia told me that it was not just a bad business, it was a horrible business. It failed. Completely.

Version 2 of JBoss, however, opened up real business possibilities. The project provided an innovative new feature that allowed for hot deployment. This, in turn, made the JBoss project interesting to outside developers, and they started downloading and using the product. Once the developers were there, we needed to provide them something. They needed to know how to use JBoss, and we knew how to use JBoss. Our training program was born.

What do you think the best new opportunities for open source are? Where do you think open source could go next?

I don't really know, and am not overly concerned with discovering the answer. But I think there are some clues to watch out for. First, follow the runtime. It's hard to build a large-scale business out of development tools/frameworks that don't offer the chance to sell maintenance. Runtime = operations = monetization.

Second, look for markets where the cost of innovation/development/integration is high. If you can piggyback on a government or other standard in this kind of market, you can drive monetization of the standard. (Marc suggested that the medical industry might be such a market.)

Third, it's no longer the case that you have to build community before you build a company. SugarCRM offers ample proof that a VC-backed company can build a viable community or, rather, can build a viable business around a smaller community of users than JBoss or other organically grown projects have had.

That said, companies like Alfresco and SugarCRM need to ask themselves (which have used this inverted community model) whether there's any real advantage to being open source. Distribution, yes. Development, yes (partner integration, etc.). But do these benefits outweigh the cost of not monetization a large percentage of one's user base?

Fourth, it's a legitimate model to find ways to "propertize" open source. Look at the Red Hat model: it's basically making open source proprietary. I've criticized this before, but I think it's a legitimate model for open source. It's a way of going from 3 percent monetization of a large user community to 100 percent monetization of a smaller user community. However, this model may only be feasible if you're working with existing communities. I'm not sure it works in the corporation-then-community model.

So, you've done this before. You've successfully sold your company. What's your advice for would-be open-source CEOs?

  1. Keep equity close.
  2. Keep cashflow neutral or better, and invest your cashflow in the company (especially people). If you manage cashflow well then you can keep your equity, rather than selling it to venture capitalists (VCs); and
  3. Hire the best. You want the equity so that you can give it to your people, not VCs.

Does "the best" mean "hire from Silicon Valley"?

No way. For development, you absolutely don't need to hire out of the Bay Area anymore. That's long gone. We had no problem hiring for our needs in Atlanta. We hired the best developers, wherever they happened to be living. Some relocated to Atlanta. The rest stayed where they were. Hiring developers is easy in open source, because you already know who the best ones are.

We always wanted to hire the top contributors to every project. We wanted people who wanted to contribute to open source full time. We wanted people who were both technical and business savvy. We took them wherever we could find them (geographically).

At JBoss we didn't outsource development, except very strategically. The company that outsources its development is making a huge mistake, because development--especially in an open-source company--and support of that development are the core values software companies can offer their customers.

Can you create an open-source developer? You were able to find them. Did you ever "build" one?

We tried once to create an open-source developer out of a normal developer, but it completely failed. We never tried it again. Truth be told, I had an aversion to it.

An open source developer is a self-starter. He's competitive - this is someone that wants to prove that they can do something better than you can. As such, it's a great recruitment/qualification vehicle, because you can see their work before you ever think of hiring them. You can see if they'll work out for the company. We definitely took that approach to hiring.

What about hiring the sales and marketing folks? The "business people," as it were? You didn't have the luxury of finding preexisting open-source business people. You must have had to create those.

For business people, we focused on the quality of the person. We needed someone not overly saddled with their proprietary past, but people that were open, smart, and could learn. We had a system in place for training new recruits so that we could teach them about open source. Matt Quinlan put it together, and it was excellent.

It helped that I was the founder of the project, but also our business leader. I had credibility with the developers, but I also understood the business side, so I could wear both hats. In the early days, I spent a lot of time talking to the sales guys about what the development guys were doing, and talking with the development guys about what the sales guys were doing. You can't let that tension get out of hand (and there was tension at times).

What was your biggest challenge as CEO of JBoss?

I wouldn't say that there was any particularly big challenge, per se, but rather that there were many challenging inflection/stress points. For example, we were seriously stressed when IBM declared war with Geronimo, and then HP got in the game against us, too. Red Hat and JonAS didn't scare us at all (really, not at all), nor did we worry about Sun's foray into the market. Oddly enough, of that group only Sun has managed to mount serious competition to JBoss.

Other stress points came when we moved to the professional open-source model, which angered some in our development community. Also, raising venture money created a great deal of stress because we suddenly had to think differently about how we grew. Each of these inflection points created stress within the company but we were able to work through them.

So, when are you going to do it again? When does Marc Fleury start the next JBoss?

I told you: I'm not planning to do JBoss II. I'm very comfortable right now. If I were to do another startup it would likely be in biotech, which I find very interesting (Marc was been working at Georgia Tech in biotech research), or another industry.

One thing I learned from open source is the power of the Internet. You can see it changing far more than just software. Even the DJ industry is being dramatically changed through the Internet. Open source is a passenger on the Internet train, but it's not the only one. There's plenty of opportunity out there.

But not today or tomorrow for me. Right now I'm enjoying time with my wife and four kids. And my PlayStation3.

Indeed, he is. You can see it in his eyes and hear it in his voice. Marc tends to give his unvarnished opinion. I believe him.

I think the industry could use his talents again. But I suspect his family enjoys them even more. His priorities are in the right place.