Oh yeah the 37 cent coin thing

I was just abstracting in my post below about some negative aspects of new experimental economics, and it reminded me of this piece I read a while ago:

Do We Need a 37-Cent Coin?

I mean, what? Are we really asking ourselves that? The answer is a definite no but for entertainment purposes let's:

First, a note : The probability of a transaction of value v is not uniform from [0, 99].

From a purely human perspective (as we are not very good adding machines), it helps that our current four coins have a common factor of 5. 5 is nice for many reasons, but they can all be alluded to by merely saying “5 is nice because it is more-or-less half way between 0 and 9 and it relates indirectly to our preference for base 10.”

I mean, seriously, a coin whose value is a prime number? Prime numbers are messy (there-in lies their joy…)!

If you’re measure of efficiency is number of coins per transaction, your root worry is probably weight of coins or size-in-pocket. Solution: smaller, lighter coins (This is why the silver dollar is impractical - trying carrying twenty dollars in silver “ones” around with you).

I think a MUCH better measure of efficiency would be speed of transaction. In this case, I doubt the efficient solution would involve 37 cent coins.

Unless I completely missed something…

Regards,
Nate

So this is primarily an example of optimization with the wrong constraints. The proper constraint is not coins-per-transaction. Like I said, obviously not because it gives us the absolutely ludicrous answer of "37". The proper constraint is speed-of-transaction.

I worry that if people in new experimental economics continue with these sort-of nonsensical approaches, economics is going to lose what little popularity it's managed to gain since "Freakenomics" was published and the populace is going to go back to treating us as boring but vaguely threatening. And communist. Or socialist, or whatever.