That all rough estimates are rough for a reason, they streamline the variables considered as "known" and make many assumptions based on those "known" unknowns.

I'll agree the Metcalfe's, Reed's and Zipf are all flawed. They should not be used as the "Standard Reference" on decision making, but another piece of the puzzle called: Cost Containment.

Cost Containment(or lack-of) is what killed more dotcom-ers than anything else. As the the saying back then, "Cost is no object", even the VCs were buying into that mantra. Quoting an "unproven" law, that was showing the evident sharp edges, but ignoring them.

Time has taken its toll on Metcalfe's and Reed's, they aren't aging well. I'd be curious to know how well this paper is recieved, and the comments on it as well.

I like the format you chose, it made a very good read. I'd only change a few things in your proof (such as the (2n)2 thing... you are going to need to back that up more than you have, even though the Biblio exists, it would help to have it in the paper keep the chain-of-thought/proof going. But then, too much and it becomes drool.

Nice job overall. Like I said, it did make a good read.