Last week, I've attended Philly ETE conference, which I enjoyed very much. Michael Tiemann, Vice President of Open Source Affairs at Red Hat, delivered the key note on the first day. His presentation was titled Exonovation: Leveraging Innovation from the Edge.
The speaker focused on the advantages of open source -- positive sum game as opposed to zero-sum game. In a zero-sum situation a party cannot benefit unless the other party suffers a corresponding loss. In a positive-sum scenario the sum of wins and losses are greater than zero; competition is viewed as a complimentary thing.
Tiemann presented research summary and stats to emphasize why open source initiatives are thriving. One of the staggering statistics that was presented --- proprietary software having 50 - 150 times more defect density than the open source software. More than 90% of leading IT vendors failed to achieve "good" rating for value from at least 80% of their top customers (both defect density and satisfaction stats are measured 3 years in a row).
He cited his experiences on the discussions he had with Department of Defense on open source software model. One of his arguments against the proprietary model is --- you can't propose that something be a universal space and at the same time keep control of it (obviously, quite a few references to Microsoft; Vista is a favorite target, explaining how it inherited defects from XP along with its own issues).
It is not an understatement to say that we are already seeing open source software competing quite well with proprietary models in some areas, and perhaps even having a major pie in some areas. In that regard the speaker suggested an interesting paper [PDF] on MIT website. The paper is a case study (year 2000) on Apache web server development compared with proprietary alternatives.
Some points I felt are interesting from the study (authors warned that not to infer too much from one case study, but provided some hypothesis based on their research):
- In open source model code is written with more care and creativity, because developers are working only on things for which they have real passion.
- Participation of wider development community helped significantly in the defect repair (that explains the defect density claim mentioned above).
- The amount of code produced by the top Apache developers versus the top developers in the commercial projects --- the Apache core developers appear to be very productive, given that Apache is a voluntary, part time activity and the relatively lean code of Apache.
- Code ownership is more a matter of recognition of expertise than one of strictly enforced ability to make commits to partitions of the code base.
- While the user perceived defect density of the Apache product is inferior to that of the commercial products, the defect density of the code before system test is much lower. This latter comparison may indicate that fewer defects are injected into the code, or that other defect-finding activities such as inspections are conducted more frequently or more effectively. It is also possible that the diversity of backgrounds of the developers participating in the OSS project have reduced the probability of defects.
- In successful open source developments, a group larger by an order of magnitude than the core will repair defects, and a yet larger group (by another order of magnitude) will report problems.
- OSS developments exhibit very rapid responses to the customer problems. This goes along with another hypothesis that in many OSS developments the developers are also the users of the software.
That was research done back in 2000, many of these are more true, now in year 2009. What do you think?