It is linkbait because the article doesn't actually argue for his point in any meaningful way or suggest any alternatives, its just a headline written to catch attention, which we both admittedly fell for given our comments here.
The article suggests we can only determine market cap by a larger percentage sale of the company. I think this is a poor argument for a number of reasons:
1) Bid ask pricing is a very well established economic model. In active market, which I would argue Facebook has due to SecondMarket et al, it tends to closely approximate the value of the asset. Certainly there are market or sector wide bubbles, but that is an entirely different argument.
2) The counterexample used is poor. He uses the example of Digg which was almost completely illiquid at the time of that article. Additionally, had Kevin Rose actually wanted to sell Digg at that point in time he probably could have actually garnered that price. The 37signals $100Bn valuation is a straw man argument that is not based in reality at all.
The most recent investors would not have bought the stock if they had assumed that their investment would be worth less in the future, therefore, yes, they are valuing the company at $80bil or whatever the number is. This is unlike the "37signals is worth $1bil" link-bait stunt in that people expect to make money on their Facebook investment.
Of course, GSV is simply betting that Facebook will be valued by the public markets for more than $80bil at some point in the future, not necessarily for any extended period. They might very well be hoping to make money off Facebook in a greater-fool-theory play.
"The most recent investors would not have bought the stock if they had assumed that their investment would be worth less in the future,"
This is true.
"therefore, yes, they are valuing the company at $80bil or whatever the number is."
But this does not follow. They are valuing their investment at the price they paid. They aren't (necessarily) valuing the rest of the company at all.
For instance, they could believe that Facebook's total value will crash to nothing in the next couple of years, but that (due to demand for Facebook stock) they will be able to make a 500% profit by selling off their investment beforehand.
That's an extreme example, but illustrates a problem with extrapolating a company's value from the sale of a small percentage.
I'm sure others have studied this, but intuitively something like a rolling average seems to make more sense for that sort of extrapolation.
AFAIK, market capitalization has not much meaning for pre-ipo stock. Problem is that, IPO price isn't just the last price at which the stock was bought in secondary market. Also, there is no indication if what kind of stock was sold. If it is not common stock, then this definitely wont be indicative of the actual valuation.
This post by 37signals is criticizing the way market capitalization is often determined. Facebook's current valuation is just an extreme example.
Your choice to ignore the argument doesn't mean there is none.