Is there a natural divide between open source and proprietary software market opportunities?

I'm reading through Si Chen's excellent paper called "An Economic Model of Open Source Software Adoption." I got lost in the calculations Si uses, but the conclusion hit home:

[T]he hurdle for adoption of open source software is a function of the relative feature sets, size of user base, scalability of features, and profit margin of competing commercial products. Where existing commercial products enjoy a large user base, can address a high portion of user needs out-of-the-box, and the profit margins of the vendor are reasonable, open source software must offer comparable or better features to gain mass adoption. If, on the other hand, user needs are highly heterogeneous, the commercial product's vendor has excessive profit margins, or where the software is fundamentally not scalable, then an open source product with significantly lower feature sets could gain mass adoption.

These conditions suggest that open source and commercial software models may be well adapted to separate niches of the software industry. Open source software may fundamentally be more suitable for software whose features are not scalable, whereas commercial software may be better suited for mass market software with highly homogeneous user requirements and features.

The equilibrium conditions shown here also suggest that the two models serve to check and balance each other. Commercial software sets a threshold of features and a ceiling on costs for new open source products. Open source software, in turn, serves to limit excessive profits and uneconomic feature bloat in commercial products. (11)

I don't know that I agree with this, but find the conclusion fascinating, all the same. Si is suggesting, I think, that open source serves a heterogeneity of user interests, but proprietary software may well afford better economics for serving a wide body of homogeneous users. I'm not sure why this would be true - why can't I develop and sell a browser, for example, and license it as open source? I get Si's point that the economics behind doing so make a lot of sense for a proprietary/commercial vendor. But why don't these same economics work for the open source vendor?

One thing that does strike me as 100% accurate is that open source serves to curb outsized proprietary profits. We see this throughout the industry right now: open source in Business Intelligence, Enterprise Content Management, Email, etc. Where proprietary companies try to wring too much profit from their monopoly rents, rest assured that an open source vendor will arise to bring sanity back to the market. And to customers.

Give Si's paper a read. Definitely food for thought.