The other day my 16-year-old daughter came down the stairs in tears. She was holding the new version of the Scholastic Attitude Test and complaining that it was unfair.

Look at this question, Dad. Who could answer a question like this?"

I looked at the question and had to agree that it was a tough one:

27. Lamb: wolf;
   Open Source Project: ________________?
a.  Microsoft
b.  Enterprise software vendor
c.  419 scammer*
d.  Jack Messman

(For additional information on 419 scammers, refer to www.419eater.com/html/trophy_room.htm.)

"Honey, it has to be A. Everyone knows that Microsoft is the opposite of open source."

"But it's not," she answered through moist blue eyes, "It says I got the answer wrong."

For those of you who have been following our journey through the various open source business models, you can probably guess the right answer. For those who are joining us mid-stream, it makes sense to back up a little and set the stage.

Background
(Skip This If You Have Read My Previous Articles)
The power of the open source movement is based on the Internet. With the Internet, communication costs dropped to near zero. This economic phenomenon enabled developers from all over the world to collaborate on projects for "free." In an earlier article, we called attention to the Debian world map as a "proof point" of the global reach this "free" resource has made possible (www.debian.org/devel/developers.loc).

The original open source organizations were based on informal groups of volunteers who got the money to fund their project through Donations. Two very successful open source projects that are based on Donations are the Apache Software Foundation and Debian. However, there is a natural limit to the amount of resources the Donation model can bring to an open source project, probably about $5 million per year.

As the open source movement grew and attracted the attention of large powerful corporations like IBM and Intel, a second open source model emerged based on Membership. Open Source Development Labs, ObjectWeb, and Eclipse are examples of successful organizations based on the Membership model. But the Membership model also has its natural limits, probably at about $15 million per year in funding.

Realizing that the open source movement was very popular, some entrepreneurs decided to create media sites that sold Advertising and sponsorships. Believe it or not, Red Hat in 1999 told Wall Street that a big part of their business model would be selling banner ads to open source geeks visiting their Web site to download "free" software. But the advertising model ran into its natural limits. Banner ads are too cheap. You can't hit a home run like Google selling banner ads. Today, the most successful proponent of the Advertising model is VA Software, which runs SourceForge.net, the world's largest open source development site and OSTG (Open Source Technology Group), the world's leading community-driven media network that services 19 million unique visitors per month. However, VA Software's media-related revenue has plateaued at about $20 million per year (the equivalent of 13 McDonald's outlets, not very impressive for Wall Street).

The smart money led by Red Hat transitioned to the Conversion model. In the Conversion model, you give something away for free and then convert the consumer of the freebie to a paying customer - somehow.

The first feeble attempt at the Conversion model was the Media Kit. This business model entailed "giving away" the software but charging for the box, CDs, and documentation. SuSE Linux is the market leader in the media kit market. It loads lots of stuff into its Linux distribution (i.e., SuSE Linux Professional supposedly includes 3,500 packages) thus making it too large to download, so that it can sell 250,000 media kits per year for $89 each. The problem with the media kit model is that it also appears to have a natural limit in the annual sales range of $25 million (the equivalent of 15 McDonald's outlets - by way of comparison McDonald's has 30,000 outlets worldwide; interestingly Starbucks goal is to have 30,000 outlets worldwide) see Figure 1.

Fresh Content
(Resume Reading Here If You Have Read My Previous Articles)
Recognizing that the media kit model didn't have the juice to get it into the big leagues, Red Hat innovated once again in 2002 and invented the Enterprise Server - Maintenance and Support business model. Red Hat "gives away" its GPL Linux distribution but sells maintenance, support, and training linked to its distribution. The Enterprise Server model has worked very well for Red Hat, propelling their business to an annual run rate of $260 million (equivalent to 175 McDonald's outlets, larger than the McDonald's franchise in 20 states - not bad). SuSE Linux, a rapidly vanishing part of Novell, has imitated the Red Hat Enterprise Server - Maintenance and Support business model with moderate success. (All indications are that Red Hat is trouncing SuSE in the marketplace.)

As the open source movement matures - as shown in Figure 2 - its opportunity frontier is shifting from the operating system to middleware to applications. Red Hat, as we have mentioned, is the clear winner at the operating system layer and as such its maintenance-based conversion business model is being imitated by venture-backed companies in the middleware layer.

JBoss appears to be a successful example of an open source middleware company that is pursuing the Enterprise Server - Maintenance and Support model. It's giving away its application server, which competes with the IBM's WebSphere and BEA's Tuxedo, and selling maintenance, support, and training. According to Joe McGonnell, director of marketing, JBoss is achieving about 250,000 free downloads per month across the JBoss branded products and Hibernate. Their Web sites are being visited by 260,000 unique visitors per month. This compares to 2.3 million unique visitors as measured by hosts per month at Apache. (One way to double-check these numbers is to look at the popularity statistics page on www.freshmeat.net. There you will find JBoss ranked 249 [http://freshmeat.net/projects/jboss/] and Hibernate at 943 [http://freshmeat.net/projects/hibernate/]. In comparison, Apache ranks 11.)

The company's vice president of product management, Shaun Connolly, told ZDNet UK that JBoss has about 500 customers who each pay about $35,000 per year for maintenance, representing $17.5 million of revenue per year. In addition, JBoss sells support and training, which probably increases its annual revenue to $30 million per year. JBoss has managed to break through the "glass ceiling" of the older open source business models by imitating Red Hat's Enterprise - Maintenance and Support model, evidence that the business model is viable for middleware.

In terms of go-to-market, JBoss has not been very forthcoming. They declined to answer our questions about the size and composition of their sales force. The danger of not giving information to a journalist is that he or she is then forced to guess. Nick Herring and I guess that JBoss is selling almost exclusively through a direct sales force calling on large enterprises, especially if they only have about 500 customers. So, in short, when you strip away all the open source trappings, JBoss is just another enterprise software vendor.

Which brings us back to my daughter's SAT question. I guess the answer is (B) enterprise software vendor, after all. Beware of wolves wrapped in the skin of open source projects.

Next month we will finally - we have been promising this for a long time - look at the Conversion model as it relates to selling good, old-fashioned enterprise software.