Proof-of-work crypto mining generates a lot of externalities. Yesterday’s post was about how not to measure those externalities. Today’s is about how to deal with them.
When miners use vast amounts of power, there are two broad categories of externality to worry about. One is environmental damage associated with power usage. Insofar as that’s a social problem, the way to deal with it is to tax power usage. That is the government’s job.
But the other kind of externality is more interesting from a crypto-design point of view. Miners are engaged in an arms race. Each miner makes the system a little more secure. Each miner earns an expected reward for that contribution. But there’s no reason to think that the miner’s reward is commensurate with the social value of the extra security that miner brings, and in fact there are plenty of reasons — at least plausible and possibly compelling — to believe that miners are over-rewarded. Currently, if you complete a Bitcoin block you earn a block reward of 6.25 bitcoins plus transaction fees that seem to be running somewhere around a tenth of a bitcoin, adding up to something in the vicinity of $300,000. It’s a fair guess that miners earn pretty close to zero economic profit, so their costs should be about $300,000 per block or roughly $45 million per day. Society is certainly getting something of value for that $45 million, but if it only takes, say, half as many miners to provide the same value, then we’re overpaying by $22.5 million every day (that being the value of the essentially wasted energy).
[it’s also entirely possible in principle that there is too little mining, in which case you can interchange all the implicit plus and minus signs in what follows.]
If you subsidize something and you get too much of it, the solution is to lower the subsidy. In this case that means lowering the block reward. If you don’t subsidize something and you get too much of it, the solution is to tax it. In this case that means a negative block reward. I don’t see any reason in principle why you couldn’t have a proof-of-work system with negative block rewards, and I think it’s at least plausible that such a system could be optimal. [The way this would work in practice is that the block reward would be subtracted from the transaction fees and then burned.]
A potentially nice side benefit is that you’d be taxing that environmental damage we talked about at the outset. You might believe as I do that if there’s a case for taxing that damage, it ought to be done by the government, not by the Bitcoiners. But if you believe your elected officials are falling down on the job, and if doing part of their job for them is an automatic side benefit of getting your own house in order, that’s a good thing.
I haven’t thought about any of this quantitatively — meaning that I have nothing to say about estimating the optimal tax rate, but there is, I think, a moral — namely:
It is often said that proof-of-work systems inevitably lead to a lot of social waste. The “inevitable” part can’t be true. It’s easy for the system to levy a tax that brings private rewards in line with social benefits.
Of course one possiblity is that with the optimal tax, transaction fees would not be enough to support the miners, in which case the system would collapse. But with such a small transaction demand, the system should collapse, so that’s not a problem.
I am sure that a lot has been thought, and a lot has been written, about optimal block rewards, and equally sure that I’m unfamiliar with most of it. So maybe a reader or two will contribute to my education here.
One thing that’s very interesting about Bitcoin’s proof-of-work secured blockchain is that transaction throughput is largely disconnected from costs to nodes and miners. That is to say that, within reason (and certainly at block sizes even an order of magnitude or two higher than today’s), increases in transaction throughput cause a negligible cost increase on the part of miners and node operators.
So, really, the bulk of block rewards and transaction fees pay for “security,” which is a pretty squishy concept. With the blockchain, there’s also a relationship between security and time. A transaction is equally secure in the following two scenarios: 1 confirmation at ‘x’ difficulty or 2 confirmations at ‘x’/2 difficulty (confirmations take an average of 5 minutes for the first and 10 minutes for subsequent confirmations).
I have a feeling that the level of “security” currently offered by Bitcoin is too high. It’s also being subsidized by the block reward (which is going down over time) for now, so that situation may reverse itself. However, I don’t have (or haven’t thought of) an objective way to judge this beyond the knowledge that when I transact, on Bitcoin (or Bitcoin Cash) I’ve never once seriously considered whether I feel that today’s hash rate/difficulty may be too low. In most cases, transactions that require “higher security” wait for multiple confirmations, and such policies *haven’t* changed to a large extent even as hash power has exponentially increased. So perhaps that’s an indicator of something, though it’s likely to be simple variability in the system more than anything else, and that problem isn’t solved by increasing hash power (“finding” a block happens at random against a probability distribution and I believe block time variability is roughly the same at any given difficulty level).
How about “Nakamoto meets Coase”? The problem with the Pigovian approach assumes that we can figure out what the optimal tax is. Why not sell or auction a fixed number of transferable “crypto-licenses” to potential miners instead?