Robert Murphy writes:
Some in the Bitcoin controversies asked me to read this essay (by Niels L). They claimed that Rothbard botched things, and needed to go back to Menger, when it came to understanding the demand for media of exchange. The only real issue I see in this article, is the claim that when a commodity comes to be demanded as a medium of exchange, this wouldn’t actually increase the total demand for it. (In the standard Rothbardian treatment, there is a snowball effect as more and more people increase their demand for it.) The reason, according to van der Linden, is that you would just be substituting one person’s demand for another. For example, if I acquire a goat not because I want to consume it, but because I will trade it away down the road, then sure, my demand went up by “one goat” but then the person to whom I trade it will now reduce his demand from the rest of the community by “one goat” (since he’s getting the goat from me).
I see what the issue is here, but I don’t think it works.  People hold media of exchange for a length of time, and sometimes not having specific future exchanges in mind.  Think of it this way: If the whole community is walking around with gold coins in their pockets, and this was facilitated in part because there are a fraction of them who wear gold as necklaces, it is nonetheless true that the community is holding more total gold than would have been the case had gold not been adopted as money.  It’s not the case that each gold coin is merely a shifting forward of that gold’s destiny as a commodity.
 Any good that you accept as a medium of exchange is speculative, in that you do not know with certainty what you will be able to get for it. Market conditions change and prices change. Any time a good is accepted as a medium of exchange, individuals have specific goals in mind with it. Their goal is to trade it away for something they actually want to consume (or give it away to someone they care about). During the time that they kept it it was not consumed and remained consumable by others.
 I would agree that more people would hold a good if it were used as medium of exchange, but I don’t see how they would necessarily hold more total amount of the good because it is used in such a way.
A good that is used as a medium of exchange is a good that has a low cost of retaining its value compared to other goods (example: gold doesn't rust when exposed to air). Therefore such a good tends to also be suitable for building up a reserve (unlike bananas) if it’s cost effective to produce considering its future value. Sending a space mission to extract gold from another planet is momentarily not cost effective, but some mining initiatives may be.
After looking at some numbers it appears the yearly amount of gold produced versus consumed is in the same range, and there is somewhere around 1-to-40 of yearly gold production versus total gold available for consumption (of which 20-25% held by governments).
Gold has not been used as money for quite a few decades. If it had, would the ratio of yearly production versus available keep growing? More and more people would buy it, irrespective of the amount used in production?
According to my theory that wouldn't be the case, but according to Murphy I am wrong . If he has praxeological reasons for it, I would like to know (and understand), but I don’t believe simply stating hypotheticals that assume it furthers the problem.