Other Presentations and Updates
Auctions Performance Presentation
- This table is covering auctions on Tuesday. To sum up a few stats, about ten million liquidated on Tuesday. Ten million debt is about 260 auctions. Maker collected more or less all the penalty fees, which is 1.2 million.
- Maker collected most of the penalty fees. 60,000 penalty fees were missing. Most of it was missing on ETH-B auctions. You can see here at the green line how much Maker had collected.
- Red means that Maker didn’t make a penalty fee, and there was no revenue. This line shows how much vaults got back from the remaining collateral.
- If vaults have a liquidation ratio of 130 on ETH-B and the penalty fee is 113, there’s 17% that vaults could get back. But since there is no considerable amount of competition amongst keepers, I’d say they make 50 to 20 percent of the money on auction vaults. Basically, we didn’t get anything back on the ETH-B vaults, and that’s because the liquidation ratio is much lower. There’s not much available capital above 113. It’s not necessarily that ETH-B auctions are worse. Keepers take a certain amount of capital, and since the liquidation ratio is lower, that means there’s less left over for vaults. That’s why we’re not proposing lower collateralized vaults. At least until we have liquidations, 2.0 implemented. It’s important to note that there are not many unique bidders. Here is the average number of bidders per auction. You can see that there’s only one or two the majority of the time.
- There are more bidders on renBTC. A bidder bid 50,000 DAI for zero renBTC. This vault made quite a high yield. In general, there is quite a lot of irrationality in some bids. You can see that a few bidders bid above market price for renBTC. Whereas if you look at ETH-B, some bids were fairly below market price. This is why keepers made a lot of money here. There’s a similar situation for ETH. It’s not as obvious because the liquidation ratio is higher on average. There are better bids on ETH-A verse ETH-B. One thing that’s positive about this and something that makes us reconsider the auction duration time is how much time it took to get healthy bids. We have a six-hour auction duration because we may get zero bids if it’s too short. Here you can see there’s still a lot of zero bids.
- On ETH vaults, 20% of all bids are all zero bids. That’s why we have a longer auction duration of six hours. You could see that most of the healthy bids came in earlier, within the first hour. Today, we proposed to shorten the duration from six hours to four hours. This is beneficial for the liquidation capacity. Currently, we can liquidate 15 million, which is the
box parameter, every six hours. If you shorten this to four hours, you can make six cycles a day rather than four. I didn’t mention that because the low efficiency of ETH-B vaults makes us propose a higher back value for ETH-B. We have already proposed a higher back value for UNI LPs due to low keeper participation. Suppose we increase the back value for ETH-B, which is the minimum bid increase from 3% to 5%. In that case, we may get more keepers participating or at least improve the bidding. There doesn’t appear to be any downside to this, as vaults already lose all of the value. I think it makes sense.
- The last proposal we had is to increase the
box parameter by a very small amount. Leeks made an excellent analysis; he checked all of the keeper wallets and estimated the amount of capital they hold, over 100 million dollars of capital. It is not all liquid or in stablecoin form. He’s suggesting that a 30 million
box would be appropriate. I think the same, but I’m a bit cautious because we are not seeing a lot of competitiveness here among keepers. Even though keepers have more capital, if we increase
box, we have a higher amount of liquidation. This means we would have even less competition, and the bids could go even lower. That’s why we proposed a modest increase in
box from fifteen to twenty million. This is our proposal analytics:
- To shorten the auction and bid duration for all
flips to four hours, increase blocks from 15 to 20 million, and decrease back for UNI LPs and ETH-B from 3% to 5%. That’s pretty much it, but there’s one other thing that we also need to address. It concerns gas prices. A lot of auctions got kicked with a certain delay. For instance, here is one vault with the total amount of debt liquidated at 1.5 million. It took almost one hour to start the auction, which is a considerable delay. It shouldn’t happen like that. This is happening because kicks are expensive due to gas prices. A kick can cost about $800 in gas. Now imagine this for 500 auctions. That’s about $150k, and there’s not a massive advantage of kicking an auction. There’s no incentive to kick it, and that’s why we have these delays. One idea to fix these problems is that we should have several actors performing these kicks. These people would have to get funded, and they would potentially perform these kicks to keep auctions ongoing.
- Iant: Primoz, I agree entirely with you, and I volunteer to do that right now. I spent $88,000 in kicks on Tuesday night. I’m happy to volunteer to help work something out. Thank you for your analysis here.
- Primoz: Thank you. I spoke with DeFi saver guys today. They said they saved a bunch of Maker loans. Yeah, I think they represent 40 million of debt, so they probably are protecting 200 million of that. They probably spent 400 ETH in fees, well, the users who pay from the collateral. Gas prices are getting really high to manage liquidations or protection.
- SamM: I just want to say that liquidations 2.0 solves this issue because there’s a built-in incentivization system. We need a solution until we get liquidations 2.0.
- Kurt: Not to rain on your parade, but it’s not actually confident that it solves it because those transactions are done in the clear and can be front-run by generalized front runners. You’re still at risk of the commons’ tragedy where you know to determine a vault is under collateralized requires you to do work; you have to run a continuous database of vaults and keep track of all of them as well as servicing the ones that are undercollateralized. However, if you always get front-run when you try to liquidate something, that could still eat away the incentive no matter how large it is because you can be bid up to whatever the value of the incentive is on the gas price. So it’s not entirely clear whether or not this solves the problem.
- BrianMcMichael: Even without the front runners, the gas is a fixed rate plus a percentage of the vault. When we see two thousand or three thousand gwei transactions, we could easily get to a case where the recompense doesn’t cover the gas to kick it off. We might still need a backstop for liquidations 2.0.
- d_zap: To chime in really quickly and add to what Primoz said, if the other advantage of passing this over to Maker is that it will be organized or coordinated, somebody could write a contract to kick lots of auctions at once, which will save money. With every individual kick, you’ve got a 21,000 gas baseline for what the transaction costs. In contrast, obviously, if you string together 100 kicks, then you will save that much from every auction. That’s definitely more efficient if you have a huge number to get through.
- BrianMcMichael: That’s actually a super simple contract to write. You still need somebody to handle the database and monitor the blockchain and outstanding vaults, which is still a resource-intensive task. There is still some expense that isn’t directly reflected on-chain there.
- Primoz: It’s good to have more than one actor doing this. We don’t want everybody to rely on this one actor. If something fails, it’s a problem. We need a backup as well. A few people doing this would make sense.
- Kurt: We’re researching ways to possibly mitigate that front-running threat on the kicks. In 2.0, so far, there’s no clear solution that makes the problem go entirely away. We have solutions that decrease the severity. We’d also prefer to not have a single centralized entity that relies on kicking auctions.
- BrianMcMichael: Maybe we don’t want to do liquidations when gas is 3,000 gwei anyway. Even if we kick them off, who will bid on them? It might be its own kind of stock market circuit breaker that prevents us from getting zero bids anyways.
- LongForWisdom: One other point is that if we end up getting EIP-1559, that changes the fee market, right? I don’t know if that’s positive or negative, but that probably affects how high the gas can spike for how long.
- Christopher Mooney: We might get an opcode where we could maybe use the opcode to compensate people.
- LongForWisdom: I’ve heard that before as well.
- Kurt: The hope for EIP-1559 is it reduces the volatility in gas. The average price will still be the same, but if it makes things more predictable, it would probably help compensate for it.
- LongForWisdom: One minor note about the kick incentive is that I don’t know if it has to exceed the cost of gas for it to be worthwhile because you still potentially get to all the auctions. It may not always need to exceed the kick cost but just offset it by 80 or 90%.
- BrianMcMichael: We need to have the incentives aligned. We don’t pay people to call drip right now. They’re incentivized to do it. Maybe we need to think about the incentive mechanism behind it.
- Primoz: Auctions should get kicked, and that’s why we have this. However, one vault wasn’t kicked for a few hours, and he was able to save himself. It’s not something we would want to rely on, of course. Many vaults are active; if you look at the largest ones in the last six hours, they can save themselves.
- JoeQ: I just want to add, something we can consider is a lot of this is happening at three or four o’clock in the morning. I know we’re transitioning to four-hour durations, and I’m really happy about that. Still, we have it extended out for specific timeframes because some people go to sleep, and the market goes way against them. That’s something to think about.
- Kurt: While we’re on the subject of making certain liquidations happen reliably, various external projects are working on solving these problems with a sort of cooperative game theory. One example that people might be familiar with is b-protocol which is actually built on Maker protocol. They have dedicated liquidators that line up and volunteer to liquidate things in a predetermined deterministic fashion. You don’t have everything going to the miners or just no guarantee of return. I think that’s an interesting avenue to explore as well.
- Christopher Mooney: Joe, you’ll work the mandated actors, and we’ll establish something even if we have to use the multi-sig that the mandated actors set up. I just want to make sure that this problem is solved as soon as possible.
- JoeQ: I’ll coordinate with Primoz and the community to figure something out. Primoz had mentioned there should be a backup actor. There’s so much money at risk. I believe that’s necessary.
- Christopher Mooney: I’ll volunteer to be a backup.
- Iant: I’ll volunteer independently as well.
- LongForWisdom: Great to see some concrete actions come of that.
- LongForWisdom: I believe the crux of the question concerns how Maker holders can judge how well core units are doing. You have many Maker holders not paying attention to what’s going on.
- Juan: One of the solutions that Matt brought up during the last call was to make sure that it’s not the default, yes. Someone can stay complacent forever. As you mentioned earlier, there’s the budget, so as long as we make the budget non-default, yes, then it should be fine. The coordinates should fight a bit to keep providing value to the DAO and show that the budget is useful. It’s essential to find the right balance. You don’t want to hire someone and tell them they have employment security for the next day and a half, but let’s see what happens after that.
- Christopher Mooney: We have to find a model, right? I agree that it’s going to be difficult. How will we effectively fire teams for not performing? The easiest thing I could think of is an annual review where MKR holders either re-elect that team for the next year or the team gets axed. That’s probably the easiest one to start. For measurement, we should set out our goals for the year, and MKR holders will see what percentage of the goals have been met.
- Juan: Yeah, but you can do the same with the budget. That’s what I meant. Each core unit is proposing a budget for three months. We could extend it for longer.
- Matthew Rabinowitz: Maybe it’s possible to not only set it as binary zero and one but to have the same budget and some scale between zero and two hundred percent chunked into 25%. You can combine what Chris outlined as you maybe only met 75% of your targets, but perhaps that was just the situation. Maybe the budget shouldn’t change, or perhaps it needs to be increased for performance.
- Juan: Totally, and another tool that we’re using for this is that a core unit could potentially have several budgets. As an example, we’d like to try with Amy bootstrapping a grants program. One of the budgets that we’re going to propose is purely for grants. If governance doesn’t think this is a good idea, the rest of the core unit could vote on the general budget and vote down on this grant budget that we will try to incorporate.
- LongForWisdom: I think that it should be interesting to see how it would work. Forcing an annual vote may not work in some cases. For example, the governance team is not doing well. There’s an annual vote; it will be difficult for governance to vote to ax the only team or the only smart contracts team. There needs to be some way of governance expressing dissatisfaction where an alternative can put themselves together. Having no one is worse than having someone who is not doing good in some cases.
- Christopher Mooney: You could bake that in at the end.
- Juan: This was less of a controversial topic than we thought.
- Iant: In terms of this specific problem in this structure, we use moving forward to analyze it or build a solution. Is it on com dev, or is there no real team that is quote on quote responsible for figuring this thing out? The way I see it, this will require research and then hopefully a testing phase of some sort.
- Juan: The short answer is no. this is not assigned to a specific team. This is something that we could work towards. However, because this is a DAO, I trust a lot in the smart people I’m working with. If something is not working, I will find a way to fix it. I think looking for every contingency in advance may not be the best solution. We will find a way, and we will share the best practice as we go along.
- LongForWisdom: I think what you’d find is that the other domains are the first to notice if one of the core units isn’t doing a great job. What I expect to happen is that when it comes around to annual renewal time, the rest of the core units would have organized an alternative on the down-low so that governance has a choice which would be the smoothest reaction. This may lead to some bad feelings for the existing people but to some extent, that’s unavoidable.
- Juan: I think we’ll find an organic solution. As long as we keep attracting the right talent, we should be fine.
- Derek: Juan, I’m thinking that there shouldn’t be any surprises for a team throughout the year if we’re diligently focused on our expenditures, tasks accomplished, and tasks that haven’t been accomplished for certain reasons. It should be known as we go throughout the year to stand with the community and other domain teams. I’m hoping that will be quite open, and if we’re falling below the mark, we know we need to step up.
- Juan: I agree. Maybe some core units are scratching an itch that won’t be there a year from now. I think you’re thinking about the essential core units. The most essential ones will always be there. Some other core units are more experimental. For example, if there is a goal of expanding DAI into the Antarctic and then three months later, this isn’t going as expected, or it’s going great, that’s why organically, we will expect a lot of reporting. Seb has been working on that. There are various initiatives on reporting transparency.
- LongForWisdom: There are sub proposals to add, remove, modify or replace core units, right?
- Juan: There is the possibility.
- LongForWisdom: There are governance processes to do these things, but in practice, we haven’t done them. We don’t know what the fallout will be.
- Juan: In the coordinate MIPs, we tried to cover these cases where someone can propose, for example, they want to offboard this guy or onboard someone, and then there are emergency procedures as emergency core unit facilitators if something is not going as planned. We are trying to think of worst-case scenarios and make it work. Amy is saying that the challenge is who does the challenge. That’s very interesting. Sometimes, the issue with human beings may be that we don’t want to be the bad guys in the room, and therefore we don’t give the most honest feedback.
- Matthew Rabinowitz: At the end of the day, MKR has everybody working on this together. Collectively speaking, it’s the money of all MKR holders. Therefore, the vote of that matters.
- Juan: Any other comments or questions?
- LongForWisdom: There’s a lot more that I could say, but we’re getting close to the time, and it makes more sense to discuss on the forum.
- Christopher Mooney: When we’re thinking about budgets, we have the real risk of running into the same issue that governments run into where the voters vote with their emotions. There may be a point in the future where our income falls like it was in the depths of the bear market. This would require a contingency for what we do with teams and our budget if we begin to run out of money at that point. We have to be careful with what we approve.
- LongForWisdom: You’re right, Chris. that situation is inevitable, and we will cross that bridge when we come to it. Let’s wrap this up now. Thank you, everyone, for coming and discussing today, and I look forward to seeing you on the forum and next week’s call.
Links from Chat
Common Abbreviated Terms
DC: Debt Ceiling
ES: Emergency Shutdown
GF: Governance Facilitator
SF: Stability Fee
DSR: Dai Savings Rate
MIP: Maker Improvement Proposal
LR: Liquidation Ratio
- Artem Gordon produced this summary.
- David Utrobin produced this summary.
- Denis Mitchell produced this summary.
- Jose Ferrari produced this summary.
- Everyone who spoke and presented on the call, listed in the headers.