Quote from: moneygames on November 21, 2013, 01:11:47 AM

A common criticism that economically inclined folks have leveled at Bitcoin is that it is prone to deflation or deflationary spirals. The Keynesian approach to this perceived problem is to increase or decrease money supply until inflation reaches a target rate. Individuals who believe this is necessary for a useful currency have argued that this is only possible with a centralized controller of the money supply.

It occurs to me that it might be possible to create a keynesian decentralized virtual currency somewhat like Bitcoin, where the difficulty adjusts to keep coins generating at a specified rate which is in turn adjusted based on information about coin value over time. The difference with this new currency would be that rather than having the target coin generation rate follow a predetermined schedule, the coin generation rate would be tied to coin purchasing power in some clever way. One way to achieve this would be for the protocol to intelligently scrape information about global exchange rates and try to regulate the average exchange rate between different world currencies to some constant. Alternatively purchasing power information is fed by miners somehow and there is a reward in coins depending on how close the information matches the average information being fed from around the world. This way there is an incentive to not skew the data, and it would be difficult to form a >50% group willing to pump inaccurate purchasing power information into the network.

Both kinds regulatory mechanisms listed above are rather naive and could result in the network being fooled by attackers into generating coins at strange rates.

[disclaimer: I realize I am referring to a "distributed network" as being capable of various things, this is hypothetical for the sake of starting with conceptualization.]

A better mechanism starts with the concept of implied volatility. If the network could "sell" derivatives on coins for coins to network participants, self-interested network participants will buy these derivatives or options contracts at different prices depending on their beliefs about the future vs current purchasing power of a coin.

A simple example(network issuing "bonds"): consider a contract where the network sell a fixed number of bonds for 1 coin with a 0.1 coin per year return, and some of bonds for 1 coin with a 0.09 coin per year return and so on down to arbitrarily small returns. If the coins are inflationary, say purchasing power is decreasing by 5% a year, participants will be willing to put many coins into the 0.1 coin return per year but much fewer coins in the 0.05 or lower return bonds(as they would rather spend the coins than buy a bond that loses value). If the coins are deflationary, users will be willing to put more coins in a lower return bonds. The network monitors the distribution of bond purchases at different prices and determines something about the expected future value of coins compared to their current value. If participants are buying up the bonds available at all return levels a deflationary state is implied, and the network decreases difficulty until the purchase rate of lower interest bonds starts to decrease, the network difficulty is basically adjusted with some kind of negative feedback control with respect to the deviation from a certain reference bond buying distribution that corresponds to the desired inflation rate.

A similar and perhaps simpler mechanism is for the network to allow participants to borrow coins in exchange for slightly decreased mining difficulty for a certain address. This creates similar incentives to the bond scenario, letting the network hold onto your coins in exchange for small returns in the form of more coins mined tells the network something about how the value of coins is changing over time depending on the equilibrium purchase rates of different lowered mining difficulties, and again the network can apply negative feedback to drive the equilibrium on "mining difficulty reduction purchases" to a price which implies the desired inflation rate.

I believe that better solutions exists involving a more sophisticated set of derivative trades between participants and network, which would give the network much more reliable information about value projections of participants, allowing cleaner feedback on coin generation rate, along with providing mechanisms for eliminating coins.

What does everyone think? Is there economic sense to these ideas,(assuming the reader is a keynesian, for the sake of argument) and if so would it be possible to implement them in a distributed protocol?

No, Keynesianism is wrong in general, but the price of money is the list of all goods that money can buy. To try and give it a specific number value is impossible and price indices are just silly in general, for example say oranges cost $1 and apples cost $2, so because prices are ratios we have $1/orange and 2$/apple. To add [ ($1 / orange) + ($2 / apple) ] together you must equalize the denominators [ ($1 / orange) * (apple / apple) + ($2 / apple) * (orange/orange) ] which leaves you with [ ($1*apple) / (orange*apple) + ($2*orange) / (orange * apple) ] now adding them together we have our answer { [ ($1*apple) + ($2 * orange ) ] / (orange * apple) }. What does that mean? It means we had better figure out what an apple dollar and orange two dollar are and how to divide them by an apple orange or we have just done a bunch of work for nothing.

However, the general idea that money should increase in quantity every year is correct. The Ideal Money is a unit of value, as mentioned by Mises around 5 times in Ch 2. of his Essay Economic Calculation in the Socialist Commonwealth (that Chapter is only like 9 pages but I don't want to cite right now, the pdf is online). Furthermore, the price of a suit in 1500 and 1913 was one ounce of gold, meaning gold kept stable value over 400 years, and the price of wheat in Greece and Rome, measured in silver was basically the same and these civilizations were also centuries apart. This means the market has consistently chosen goods with stable value as money, so you can improve upon Bitcoin by creating a crypto-currency that tries to retain stable value, but the only way to do that in a decentralized manner that will work with the Bitcoin system is modeling the rate of coin creation after the rate of economic growth. Unfortunately you can't measure economic growth in a meaningful way for the reasons listed above, so the best you can do is make educated guesses.