Facing the open source firing squad

As Vadala noted, IBM maintains a team dedicated toassisting the Apache project. I have worked withApache's SOAP server for Java, a product originallywritten by IBM and donated to Apache.

I question, however, whether IBM views investment inopen source projects as a profit center. Think aboutthe types of open source products IBM spends its moneyon: Apache's Web server, the Java SOAP server, and fora time, the Mozilla browser project. I wouldcategorize these as software "plumbing," like TCP/IPstacks are to Web applications. It is base levelinfrastructure whose importance is derived from thekinds of software you build on top of it. Somecompanies make money from software "plumbing," butmost don't (including Microsoft) simply because thereis so much of it already in existence that hoping tosell it would be an exercise in futility.

Many of the most popular open source products fallinto the plumbing category, and most of the companieshiring developers to work on it never had plans tomake any money from the application domain covered bythe open source product. Rather, they derive revenuefrom their proprietary offerings, using the opensource products as lead-ins to sales of theseofferings. Think about the amount of money IBM spendson Lotus Notes or DB2 versus the peanuts spent ondevelopment of open source infrastructure. These opensource products are like marketing expenses, and theinvestment is recouped through sale of proprietarysoftware offerings (or in IBM's case, hardware).

In other words, proprietary software revenue wags theopen source tail, not the other way around. The factthat IBM hires people to work full time on open sourceis less of an indication of open source's inherentprofitability than a demonstration of the freedomavailable to companies with business models based onthe revenue generating power of proprietary software.

Vadala mentioned that IBM hired Steve Best as away to boost their credibility as a provider of opensource consultants. I'm sure Steve Best is paid well,but then again, the rock stars are always paid well. Most people aren't rock stars, and I wouldn't base theclaim that open source programmers can make money fromopen source projects by pointing to Steve Best. People like Anders Hejlsberg (inventor of Borland'sDelphi language and Microsoft's C#) and James Gosling(of Java fame) make a whole lot more money than mostproprietary programmers. I'm not going to point tothem, however, to guide expectations of financialcompensation among proprietary programmers.

Vadala repeated the claim made by many open sourceadvocates that consulting and support services willmake up for revenues lost from software sales. Thatargument has been deflated somewhat be recent eventsin the industry. Support is the essence of RedHat'svalue proposition, and they are still not profitable. Larry Augustin of VA Linux fame (the company thatstarted the whole Linux IPO craze) is taking hiscompany down the proprietary software route.

IBM certainly has talked the open source talk morethan most, and actively encourages open source productcompetency among its consultants. However, don'tascribe too much credit for IBM's recent success toopen source. IBM still makes its money fromproprietary hardware and software. Providing a teamof consultants that help customers to install opensource products is like sending out contract farmersto till the field in preparation for the sale of seed,tractors and harvesting combines. IBM views opensource as a low cost means to lead companies into itsrevenue-generating (and proprietary) stable ofhardware and software products, all of which arecompatible with the open source products that theyemphasize in their projects.

Lastly, given the track record of pure open sourceplays, I wouldn't consider Marty Roesch's successfulhunt for venture funding as evidence of a trend. Theventure capital industry is understandably shy aboutinvesting heavily in companies that see a pot of goldat the end of the open source rainbow. For everyMarty Roesch I can name a thousand wannabe LarryEllisons whose business model is predicated on keepingthe source code private.

In closing, a lot of people came away with theimpression that I dislike open source. That couldn'tbe further from the truth. As I noted in the originalarticle, open source products have advantages simplynot present in closed source alternatives. However,that doesn't means that I must accept that open sourceproducts are of necessity better than closed sourceones, nor that business models based purely on opensource are as viable as proprietary ones.

If Hegel were a programmer, he might say that theideal lies in a synthesis of the "thesis" underlyingproprietary development with its "antithesis" in opensource software. Such a hybrid model harnesses theproductive potential of two different segments of thedeveloper marketplace, those who produce code for nodirect compensation, and those that do.

Use the open source development model for what it doeswell, which is create flexible, widely used softwarewhose popularity is driven by its low cost andtransparency. Use closed source software for itsconsumer orientation, its speed of development, and togenerate revenue. That revenue is indispensable ifone hopes to attract the developers and entrepreneurswho wouldn't enter the market if "fun" were their onlyreturn.

John Carroll is a software engineer who lives in Switzerland. He specializes in the design and development of distributed systems using Java and .Net.