@uly
That's exactly the direction I was thinking in.
Let's say I manage a goods exchange site.
Let's say I do the following: you can buy roctos for x dollars/euros/yen etc. on the site for (1+c(x))*x dollars. In other words, you buy roctos and pay a small commission, some kind of function of the amount you buy ((log x)-.5) rounded down to cents and with a minimum purchase of 10 dollars/euros would be an example. I store the money and sign a contract with all buyers and sellers that at any time, they can exchange roctos for their original currency and get (1-d(x))*x dollars/euros for their roctos, d being a similar function to c (although somewhat more favorable for the customers, in other words, it should be a little bit easier to get rid of redcoin than to buy it). This is very similar to currency exchanges.
Basically I store any amount x dollars/euro in bank accounts, because I need it when people want their money back: the system would be completely non-fiduciary in the sense that I legally oblige myself to have, at any time, the same amount of money stored as I have issued roctos. This means that you'd have different kinds of roctos in circulation: dollar-roctos, euro-roctos and so on.
The amounts x*c(x) and x*d(x) are for site maintenance and - if I choose to launch a profit-based currency - for profit. But that's not the point: I'm studying this problem as an intellectual exercise.
At some point, when no new roctos are issued, I guess the value of the redcoin can appreciate, but never go below the value of what people put in it, because any rocto can be converted back to its original currency.