‡ Melmint: Trustless Stable Cryptocurrency. [PDF]
Yuhao Dong (University of Waterloo), Raouf Boutaba (University of Waterloo)
(‡ accepted for both conference and journal)
Paper summaries from the reviewers:
“This is a paper describing a Seigniorage Shares Style stablecoin called Mels. The stabilization is provided via a set of auctions.”
“The paper proposes an interesting approach to effectively implement demurrage in a cryptocurrency by presenting mechanisms to vary the supply of a coin over time, allowing it to contract as well as expand.”
“The paper presents a minting mechanism to create a trustless stable cryptocurrency using Themelio blockchain.”
Comments on the strength of the paper:
“I think this is an interesting idea, especially in the context of demurrage, which should perhaps be made more central to the proposal. Nobody really knows what would happen in practice over the long term, the experiments that were made in Germany/Austria during the depression were of relatively short duration, but appeared to be highly successful. A cryptocurrency is potentially a useful testbed from this perspective.”
“The paper provides a thorough evaluation of the proposed mechanism.”
“The author(s) employ advances economic modeling techniques, it's persuasive and the analysis seems extremely thorough.”
“The idea is well described, some simulations are presented of price relationships, although not in the context of any form of realistic economic simulation, especially with respect to behavioural choices.”
“Rather than attempt to justify in the context of fixed exchange rates/currency boards, where the economic theory is highly debatable, and the lesson from history is that failure over long periods of time is inevitable (Keynes comments on that notwithstanding), it would be better to simply present it as an attempt to implement demurrage.”
“[With] such a system, implementation details are very important. The paper needs a lot more auction theory citations, sadly much of this work has yet to make its way into the blockchain space, this is perhaps the best place to start.”
“The author seems to have confused blockchain front-running with traditional exchange front-running. This is a completely understandable and unfortunate misunderstanding making the citations [4, 17] invalid. I suggest reading this paper. Further, the blockchain level analysis doesn't properly take into account griefing.”
“The assumptions regarding cost of attack in the Mechanism-level attacks don't seem to be based on the actual costs associated with attacking such systems. Although these are references to blockchain level attacks, but they are an empirical upper bound on the cost of mechanism level attacks, generally speaking. This paper reviews a number of the attacks referenced above. Bearing this all in mind, the protective measures specified in the final paragraph on page 7 seem insufficient.”
“The game-theoretic analysis that is referenced in the Blockchain-level attacks section is the analysis I actually want to see, there is also probably a multiplicative effect from applying a mechanism-level attack with a blockchain-level attack that warrants such work. The analysis would require a working implementation and a robust simulation framework, these are not readily available to the public so I understand why such an analysis was omitted."
“The unique design quality of Melmint seems to be the use of DOSC sequential proof of work but this does not seem to show up in the evaluations?
“One way to significantly improve the paper would be to include the alternative mechanisms in the theoretical & simulation validation for comparison, and to connect those to the novel design qualities of Melmint.”
“The paper presents simulation results of some of the expected price behaviour, but does not model what might be the behavioural reaction to such a currency, which could be expected to be quite varied I suspect.”
“The paper mentions demurrage in passing, but does not actually reference ‘The Wörgl Experiment’.”
“The arguments around currency boards, and fixed value currencies are not particularly compelling, and it should be noted that although this is a property that is desired of money, it is not by any means a property that can be imposed on money, within the context of a larger monetary system based on market price determination. Variations in value serve a similar role as elastic shock buffers in other systems, bearing in mind there is no practical way to control the distribution of money across the economy, attempts to fix this risk endangering the fault tolerance and other properties of the larger system.”
“Statements such as "such extreme volatility is due to a combination of volatile demand and perfectly inelastic supply" should be tempered a little, the reality is that the reasons for the volatility are not completely understood, and deliberate market manipulation - for which there is evidence - should not be overlooked. There is also the issue (as alluded to) that to this point in time monetary expansion of the cryptocurrencies has been extremely rapid - comparison with similar rates in the main economy would be to periods of quantitative hyper-inflation. What will happen as the supply stabilizes is not yet known.”
"Currency board stablecoins have an advantage in that as long as the bank is trustworthy, no economic shock of any size can disturb the peg. Even if all users suddenly dump the pegged coin, the bank always has enough assets to sell to maintain the peg." This claim is incorrect, unless the bank is operating as a full reserve bank, which they typically don’t. Attempts to peg currencies can succeed in the short term, and as long as there is’t a serious mismatch in the relative quantitative expansion of the two currencies, and no serious exogenous event (trade war, internal revolt, etc.) intervenes. But over the time scales that currencies operate on (decades) sooner or later all fixed currency schemes fail - see for example 19th century gold standard, Bretton Woods, pre-EURO monetary band etc.- even if the counter-parties are trustworthy, due to the underlying mathematical forces involved.
“I'm curious about their validation in Section IV a) the theoretical model, and b) the simulation, and especially how they are justifying their proposed mechanism as an alternative to other ones like makerdao and seignorage shares.
For a), their equation seems basically to be relating the expected revenue from fees, to the time value of money used as reserves. Would this be equally applicable to any stablecoin that has some way of collecting transaction fees?
For b), similarly how would the other alternative designs fare under the same simulation settings?
In my opinion the best way to improve the paper would be to include the alternative mechanisms in the thoeretical & simulation validation for comparison, and to connect those to the novel design qualities of Melmint.”