I ended up reading this post because it talked about SourceForge (albeit the Enterprise software).
First, I don't agree that "serious" numbers are "multi-billion dollar" numbers. For one thing, of the 120k active information industry corporations which filed tax returns in 2002, only 264 of them received $0.25 billion or more in that year. (source) Personally, I can only think of 5 software giants which actually break into billions in terms of annual revenue - Microsoft, Oracle, SAP, Electronic Arts, BEA. There are probably some others but I'd be willing to bet it's no more than a dozen or maybe two dozen.
So one point I would make is that software revenue doesn't have to be concentrated in order to be serious. After all, the 119,476 information corporations that make <$50 million per year (under which VA Software falls) can account for, very roughly, $188 billion of annual revenue.
The second point I would make is that the nature of open-source is such that its value is hard to quantify with $. Different people may value an open-source software package at $1 or at $200 or at $2,000. But they all pay $0 for it, so it's hard to measure the value, but it's obviously not 0. I realize this has nothing to do with revenue, but recognition of this fact is a prerequisite to building a good open-source business model, which has everything to do with revenue.
Finally, one goal of open-source is to lower the cost of software to users. In principle then, open-source companies shouldn't be as big as proprietary software counterparts. The goal of open-source is to grow the economic pie while at the same time requiring a smaller slice for the software industry. This is directly opposite to proprietary software which desires more and more money to pour into the software industry - whether or not the economic pie is growing at all.
Basically, revenue-focused analysis is not simple in the open-source world.