The only explanation I've come up with that sounds plausible is that a new currency, even just Bitcoin with a different genesis block, should be as valuable as Bitcoins, but fails the test of scarcity -- it's too easy to counterfeit them by double-spend attacks because of the smaller power of the network.
I don't trust this idea, though -- it smacks of post-hoc reasoning. I guess by extension the value of a proof-of-work-backed currency in aggregate should be bounded by the expected cost of a 51% attack given rational actors; doing this math for the Bitcoin network would either give strong evidence that the current pricing is irrational or weak evidence that it is rational.
Most often, when gold and silver were used as currencies, it was because some dude punched his profile on one side of the coin, and because doing that as a private citizen was difficult and/or punishable by gruesome torture, the value of the money was related to how powerful the head on the coin was because that related to the probability that the coin wouldn't turn green.
This doesn't really answer the question you posed. Now I'm totally in the realm of post-hoc reasoning, though, and I can't really think of a concrete way of falsifying this argument, but I'll go for it anyway. Gold and silver occur naturally near the surface in relatively pure forms; gold especially since it doesn't really allow with anything. As a result, these metals were present in antiquity, and as systems of money grew up, they were good candidates because people thought they understood the scarcity -- an understanding that turned out to be proven incorrect several times in the history of coinage, with reasonably disastrous impacts on savers in those metals. Of course, "being used in antiquity" was not one of the properties I defined above. Arguably fungibility is a factor if the metals were not easily identifiable or distinguishable from other metals.
Which is exactly why its so difficult to replicate "easy functionality" such as Facebook, Twitter, Instagram et al.
