Hmm. Well it’s hard to tell what’s the right perspective. In any case, it’s all a choice between the right definition of a corner case. So it’s not particularly consequential. What you’re saying is that the case I outlined in 2. above is not the perfectly elastic case. It instead analyzes what happens if one actor is willing to sell batches large enough so that the average price is p. Under this scenario, it does seem like a directly incentivized market can reward buys and punish sellers relative to an unincentivized market. That said, there’s no doubt that in the perfectly elastic case as you’re describing it, there’s no hope for direct incentives. If sellers are only willing to sell until the marginal selling price is p, then in a directly incentivized market, the price for buyers will be worse than in an unincentivized market as long as the marginal mint is smaller than the marginal burn.
I don’t know enough about markets and trading to know what perspective is more consequential. In any case, I think we can conclude that for any pair of direct incentives where the mint is smaller than the burn, there is at least one important condition under which the unincentivized market is better for buyers than the directly incentivized market. This whole discussion has got me thinking about what the right definition for a “good pair” of direct incentives could be. Throughout yesterday, my intuition was that a definition for a good pair of direct incentives would capture the idea that a good pair of direct incentives produces markets that are better for buyers and worse for sellers than unincentivized markets, regardless of the nature of the buy and sell pressures. You’ve largely convinced me that this is impossible. Now my intuition is that a definition for a good pair of direct incentives would capture the idea that a good pair of direct incentives produces markets that are better for buyers and worse for sellers than unincentivized markets with probability greater than 1/2 + \epsilon over all possible combinations of buy and sell pressures. Needless to say, for such a good pair not to be gameable, it would have to be pseudorandom. And at that point we’ve moved rather far from the original idea. It’s still an interesting conversation to have: do good pairs of direct incentives with probabilistic guarantees exist? But I think I’ll move that conversation elsewhere.
Thank you for discussing this with me. Much of what you’ve said has changed my mind. More relevantly to this thread, I’m on board with the idea of reweights that effectively help maintain the peg by creating properly incentivized arbitrage conditions. I haven’t paid close enough attention to the idea you propose here, but it looks like there’s a developing consensus that your idea will achieve this! I hope it works out.
What’s observable? And what would we want to build on it? It looks like it’s a tool for data visualization. I’d definitely be interesting in working on some visualizations that help us better understand how the system is operating.
I’m supposed to be occupied as well! But I haven’t been able to get my mind off of all the fascinating problems that have arisen here. It’s possible that I’ll get excited enough about making visualizations that I won’t be able to help doing them before the weekend. That said, if I were responsible I’d also wait until the weekend.
I see your point. My worry is that reweight mechanism would not work well in a scenario of high sell pressure. Dont you see a risk of bots frontrunning the reweight and being the first to sell?
The difference in the undo, is that ppl are also returning TRIBE. So PCV goes down but also TRIBE mkt cap is reduced. In the case of using just reweight, only PCV goes down.
But I understand that is important not only to provide a short term solution, but also show that the system is prepared to work well in a sustainable way.
Arbitrageurs will buy FEI in advance of the reweight until it is trading at or near $1. At that point (or at whichever point they prefer), any regular user can sell their FEI in advance of the reweight. After the reweight, it doesn’t matter if bots are first to sell because, as long as there is a predictable future reweight coming, arbitrageurs will again buy FEI in advance of the next reweight. There will now be buyers to go along with the sellers so the post-reweight price should not dip too far below the peg for too long. The more predictable and the more confident arbitrageurs are in the reweight mechanism, the more they participate and the smaller the actual deviation from the peg at any time.
The protocol doesn’t have to burn those TRIBE, they could be added to the 40% DAO portion, or given to pre-swappers.
I do.
That did not happen last week, despite the arbitrage opportunity AND the mint incentives. I think this proposed rework of the reweight trigger criteria is well-designed (in that it addresses most of the problems with the current reweight), but it should be used in addition to a “reweight queuer” or other mechanisms to guarantee an exit at peg.
The psychological aspect of being able to do risk-free arbitrage is an important tool to keep the peg.
Yes, these are better ideas I think. Maybe letting these TRIBE in a staking pool to reward long term holders o TRIBE.
I agree that without a sth like “reweight queuer”, more reweights only means more bots draining the PCV. This was already proved last week. Bots will not be tired to drain the PCV.
I read all the posts about “reweight queuer” but it is not clear to me, If we can guarantee that everyone would be able to sell at the peg or If would be a order to follow.
@Eswak@Bruno
I agree that the reweight queuer or an auction could benefit FEI holders even more. But there is a fundamental difference between expected and unexpected reweights. If reweights are expected in the future, bots will buy FEI in advance so that they can sell after reweights. FEI holders can simply sell to these bots and exit at prices pretty close to 1 before the reweight actually happens.
This did not happen last week because reweights were never expected. No bot could have anticipated that the reweight condition would encounter a bug, so they had no reason to buy up FEI beforehand. Technically, the first two reweights weren’t bugs and they were triggered because the reweight condition was met, but due to the nature of the reweight condition, it was still hard to predict them, and there were no guarantees of regular reweights.
Great articles, by the way. I am surprised that people thought they could create a stablecoin with no collateral…
With high sell pressure, the problem is that is not guaranteed to sell at peg $1 as after reweight the price quickly go down. With this uncertainty about what price you are going to sell is difficult to arbs work. That’s why I think that without a solution as an auction or reweight queuer, we could have the same problem even with expected reweights.
@Bruno@Eswak I’ve been thinking about this some more, and realized I may not have appreciated the full benefits of auction/queuer. I think they are better than AMM reweights mainly because they don’t waste gas. Gas auction from reweights takes money out of the protocol into the hands of miners, whereas auction or queuer lets sellers compete with lower prices or longer wait times, which both benefit the protocol. This in turn means that regular sellers can get a higher price. Once the protocol is through this crisis, a next step could be to develop and refine these gas-free mechanisms that provide more efficient ways for the protocol to buy FEI. For example, we would have to choose how much ETH to be auctioned off in each auction, as a function of the accumulated sell pressure.
So I am mostly with you here. Perhaps a remaining point of discussion is that I think it is more about gas efficiency than fairness to regular investors. Because there will be many bots competing to buy from regular sellers and sell to PCV, these bots won’t make much profit in equilibrium. But they still need to recoup the gas fees that they pay in gas auctions, and this will mean lower prices for regular sellers when they sell to bots.
Also, I think the gas cost of reweights would be something we are willing to take in return for a quicker return to the peg. As long as frequent reweights are expected, bots don’t need to compete in every gas auction. A patient bot could buy at any price below 0.99 across multiple intervals, and then sell a week later, when the gas auction is not that competitive anymore. This means sellers can sell to this bot at 0.99. So I agree with @Terzo that it’s about time value, and that we will see patient investors buy from impatient investors. As long as some investors are patient, the gas cost won’t be huge.
That said, if I were responsible I’d also wait until the weekend.