He points to Sun's open-sourcing of its SPARC designs, Tim Berners-Lee asking that ordnance data be open-sourced, and the approval of the Creative Commons license in the Netherlands as pushing the boundaries of where we thought open source might work.
Here is what I think.
The real limits to the business model lie in provable commercial advantage. So long as a company gains a net benefit from hiding its code, the code will be hidden. Once that is no longer the case, it makes perfect sense to go open source.
The new balancing test is between protection and stagnation. Code that is protected depends on its owner to progress. This is a continual, and rising cost.
In discussing new open source projects with folks from companies like Computer Associates and Sun over the last year, this point has been made explicitly. We made it open source so it can become better.
Think of it as Moore's Law in action. Hardware loses its value as it sits on a wharf, or in a warehouse. So, now, does software, thanks to open source.
Sure, there are costs and risks in this approach. Someone may learn your code base better than you know it. They may take it in a direction you did not intend. They may start gobbling profits you thought should belong to you.
But despite the high (and rising) cost of software development, software is not really a capital good. You don't write off the costs of developing software over 10 years. Generally you expense it. So when the risk of obsolescence rises past the proprietary value of the code base, open source becomes actuarily sound.
Put more simply, in software "mine" can become more trouble than it's worth.