This article just appeared in Business Week. It is written by Ben Chelf, who is one of founders of Coverity. His firm has tested software from both commercial and OSS suppliers and makes two important point...

1. that the average quality (measured in terms of number of defects found through automated testing) is above the average for commercial code. (the "many eyes" phenomenon)

2. But the best commercial code is still significantly higher in quality than open source.

I would encourage everyone to read the article.

From my point of view it brings into focus the question of what changes we might see in the next decade of open source software in terms of quality beyond "many eyes". As Chelf points out, the key is automated testing at a level beyond what human testing would likely achieve. So the question is how would such testing enter the open souce world? The obvious direction would be for tools and training to be available to open source teams. But Coverity, as well as most other similar tools, are commercial products, from companies who depend on the revenue from the sale of the tools.  So any move to make this type of tool available in a centralized way would need a sponsor with deep pockets. Another possibility would be for the more established projects, like Eclipse, to make the tools available to their committer community. Maybe not such deep pockets needed here since the scope would be more limited. Finally, the OSS community itself could step up to create the automated testing tools - and again, Eclipse comes to mind. (BTW, I am impressed with the release process already underway within the Eclipse foundation so its not out of the question that they could take this on)

In any case, for OSS to continue to expand, there will need to be both training and tools to help projects with mission-critical initiatives to have the tools they need to create mission critical software.