6-13-2023 This is an archived work. It’s preserved in memory of Nikolai Mushegian, co-founder, core contributor, and dear friend. Some parts may be out of date. Updated docs will be posted soon.
Rico is a spiritual successor to Dai and Rai. But why a third system?
The short answer is that Rico was the intended design. The teams building Dai and Rai made choices that make those systems behave differently from one of the earlier points in the design space, which is potentially a better choice for a credibly neutral, credibly decentralized, and economically scalable credit and stablecoin system.
Dai chose to become a dollar-pegged asset and de-emphasized automation. This might have been a necessary step, otherwise collateralized stablecoin systems might not have gained traction at all. Most people are incredulous to learn that a system with 150% minimum collateral ratio survived a collateral crash of nearly 90% without dropping below 200% collateralization. This gave a live demonstration of the scale-invariance of the system.
The next major crash did not go as smoothly. As a whole, the system “worked as intended” in terms of appropriate failsafes kicking in, but some localized failures revealed weaknesses. The limitations of the dollar-pegged design became evident as an artificial crunch was caused when some auctions failed to process which resulted in some collateral being sold for 0 dai. The result of this loss of collateral and liquidity crisis caused dai to become overvalued. When lowering all interest (quantity) rates was not enough, MakerDAO operators had to spring into action and implement a number of new contigency mechanisms, including introducing USDC as collateral.
At some point, every theoretical issue with your mechanism will become real. It is like gravity. The market finally processed the issue with the naive dollar-pegged design, and critical mass for Rai was formed. Meanwhile, the Dai system depends on USD collateral to scale safely. What do you do when there is no more USD available to you at a price of 1 USD? Where does one go when they need new real dollars? That future is already inevitable unless Dai renegs on the dollar peg. This is possible, since the functionality is in the CDP engine to have a moving target price, but the repeated decisions of MKR governance shows this is not likely to happen.
Rai was formed as a reaction to the cracks revealed in the Dai system. The first change was that Rai regressed to using only ETH as collateral as a reaction to the general problem of centralization.
The more important feature of Rai was that it re-enabled the Target Rate Feedback Mechanism (Vox module),
the entire component that deals with automating the monetary policy of the system.
The market was shown how a price rate is distinct from a quantity rate, how both could be used for monetary policy but price rate is more general, and how core aspects of monetary policy can be automated with simple mechanisms, even aspects that some dogmatically believe must be done by a panel of experts. The easter egg here is that making the price rate explicit allows the central bank to offer a negative effective borrowing rate while still having positive cashflow, which is necessary for the system to still “pay for insurance” and have a real equity backstop.
Rai also emphasized “un-governance”. The history of why MKR is called a “governance” token, and why MKR-voting is emphasized, is a story of how bad marketing decisions stick. The result is a self-selection of holders towards those who believe MKR entitles them to control some aspect of the system. Try to imagine where that leads, in light of the idea that ‘anything that can go wrong, will go wrong’ in cryptoeconomic systems.
The first major issue with Rai is that the regression to single-collateral is just that, a regression. It seems like it might make sense as a reaction to the idea of centralization, since all other assets have some varying degree of centralization (at least their representations on ETH do). Later I will write more about why this is a fallacy. With an appropriate collateral onboarding and debt ceiling mechanism, new valid collateral types can only strengthen the system, and are not qualitatively distinct from the risk that WETH introduces because it might have a bug.
Note, the following paragraph is outdated, as the Rai team has updated their controller to be more like the original design. It no longer functions like a wrapped perpetual swap like described here. The next major issue is that Rai uses the wrong controller. “Wrong” might be subjective, since the result is a system that is useful for different things. Rai is essentially a USD/ETH perpetual swap, where the ETH is the cash and the USD is the synthetic, and where the interest rate is baked into the price of the synthetic. This might be good for some things, but it makes it impossible to implement the kind of interest rate arbitrage strategies that enable stablecoin systems to scale like fiat currencies.
In retrospect, this path may have been better. Dai and Rai serve as useful conceptual stepping stones. It probably would have been too much to introduce all the concepts at once back in 2015-2017. It might have been inevitable that some parts end up getting built wrong because of how many new primitives there are. If that’s true, then this was a relatively controlled progression.
Rico can be described with 2 major differences from Dai and Rai:
There are also a minor number of ‘cultural’ differences: