Episode 72: February 06, 2019
00:00: Intro with Rich Brown
2:37: Mkrgov.science demo with LongForWisdom
29:11: Recap of Forum Topics with LongForWisdom
29:46: Forum Discussion
44:44: Migration Analysis with Primoz
57:48: Questions and Discussion
SCD: The Single-Collateral Dai system
MCD: The Multi-Collateral Dai system
CR: Collateralization Ratio
DC: Debt Ceiling
ES: Emergency Shutdown
EV: Executive Vote
GP: Governance Poll
SF: Stability Fee
SLP: Secondary Lending Platform
DSR: Dai Savings Rate
CDP: SCD CDP
Vault: MCD CDP
Introduction & Governance
Summary & Introduction
- Welcome to the February 06th edition of the Governance and Risk call. LongForWisdom will walk us through a new iteration of the Governance dashboard. I’d love for this group to evaluate it together. We love feedback and suggestions for improvement, so please feel free to tell us.
- Give us feedback about the call.
- DISCUSSION HAPPENS IN THE FORUM
This is the Governance Analytics dashboard created with the team at Protofire.
- The home page has several metrics that bring insights into Maker Governance data. Each of these metrics views are interactive. E.g., staked MKR or votes. Hopefully, we can start to illuminate Governance operations or overhead.
- We’ve also added help text/context to each of these metrics(they’re the
i in the top right corner.) The insight includes links to the sub-query.
- One common complaint is the widget has a slow loading time. It’s often querying the blockchain, we’ve had this discussion before and will come up with improvements.
- MKR responsiveness measures how quickly votes go live. On average, 25K MKR responds to an executive the first day it goes live. Responsiveness gives a baseline idea of how much MKR we could rely on for emergency responses. Also, if an executive is popular, you can see the visual indicator.
- It’s an average figure, and there is some nuance with that calculation.
- Similar view for polls. Breakdown of responsiveness per day.
Voters has a leader board. You can expand it to a larger list with ‘view all.’
- If you click on an address, you can see how it’s voted, including links to polls or executives.
- Executives overview including
uncast, which helps to discern if MKR is gathering on an old executive.
Time to pass the average time an executive takes to pass, you can see they’ve mostly passed in a single day.
MKR distribution by Executive Helping display the hat.
- List of most recent polls.
- Deeper view of executives, including the current hat.
- Similar view for polls, you can also search by name.
- A deeper view of polls information.
- Individual poll details here to dive deeper.
- Vote breakdown is now more visible, you can see whales voting patterns, or when people change their votes, it makes it easier to see capitulation.
- The leaderboard here shows who has contributed most by address size.
- Executive votes also include filtering and sorting. One point is the metadata only pulls title and status, coming from the Foundation’s dashboard pending a new API endpoint.
- Page two of this executive vote. Including the breakdown vs. current hat.
Approvals by address size is interesting, though it’s not Sybil safe. We’ve complained about whales, but now that trend is more visible, and voting is more distributed than we thought.
Questions and Discussion
- Rich: We could split polls up by hours though right since the cap is usually a week?
- L: I suppose we could, sometimes polls can be longer than a week, though. A cap might be a good idea.
- D: The data is mixed since some polls have a longer duration.
- D: Those are vote proxy addresses, right?
- L: Need to double-check. They don’t attach to the contract, so they may not be proxies. The dashboard might be missing cold or hot wallet interactions. Regardless, if you click on the address, you can see how it voted.
- D: Do we have the same level of detail for Executive Votes? I know executives don’t auto-populate their information as polls do.
- L: The data lives on GitHub, and a hash on chain links back to the poll, so we don’t have the same for executives at the moment.
- D: Conceptually, in the future, couldn’t we just vote directly from this page?
- L: Yes!
- R: You can see the volume and also the distribution. As long as it’s reasonably flat, you can assume an alignment.
- L: Perhaps.
- R: It doesn’t say what they voted for.
- L: That’s an option we don’t have, but we could look into displaying which vote by address size.
- C: How could we use this to make Governance more efficient? What voting information we should address in future governance.
- L: Much of these metrics are for current and not historic state. We don’t track efficiency, though health and distribution is kind of close.
- E.g., 62k on the hat could be larger and, thus, safer. Tracking what increases voter participation is more complicated.
Total votes kind of tracks efficiency.
Rate of change on votes adds insight too.
Matthew: Any way to check about sitting on an old executive? Showing what might be theoretically lost?
- L: Perhaps. One measure I’ve been evaluating is “how active is MKR.” The idea of that is daily, monthly, and quarterly activity would help us spec out what is available in an emergency. Diving into that metric would be helpful.
- D: It shows us who the protocol politicians will be too.
- L: Once we have vote-delegation, then yes, people will point to votes and say, “rely on me for consistency.” Assuming, of course, we have delegation in the future.
chat: If the address holding the MKR has other non-voting activity we can know if they didn’t get hit by a bus right?:
- L: Potentially a “time since last active” for addresses that are larger maker holders. Tons of ideas we could do to confirm activity.
- Kenton: Does anyone know the status on mkr.tools? If there might be a v2?
- R: Pretty sure he needs a robust API, so he doesn’t have to scratch it out for MCD. Taking the opportunity to plug the community grants and team. The models are the same in Governance and with the community grants. We are interested in evaluating proposals that look like this dash.
- R: That’s how Protofire did it. They came with an idea and went through the grants process. Low and behold, we have this as a result.
Forum Activity Recap
- David: BAT SF question, is it ever possible to have an SF lower than the current DSR? My understanding is no because you could just issue that DAI and put in the DSR for that spread.
- C: We’ve talked about subsidizing other collateral and locking it in the DSR. If it’s small amounts. It’s technically possible, e.g., when the Sai SF was lower than the DSR. People would mint Sai, migrate, and lock it up; nothing broke, although that’s not the best behavior.
- D: For example, BAT SF is 5% while the DSR is 8.75%; in this situation, although they would be able to mint Dai and earn on the spread, some portion of the borrowing would be healthy, not just there to capture the spread.
- C: If we wanted to do that for BAT, I’m not sure why we’d do it specifically for BAT, but it’s possible.
- L: The main benefit is diversification or subsidizing for diversification. Dai that’s created and locked up is essentially parasitic since it will go back into the DSR, gaining only diversification.
- MR: If you lock in DSR, it’s pure risk. Especially since the downside is on a volatile price change and liquidations. It costs the Maker community capital to subsidize the collateral.
- Cyrus: The Sai supply is still going down, so Aaron’s thread is relevant but not urgent. Until we see the resulting interactions slow down (E.g., closing CDPs and SFs, etc.), it’s still a premature discussion. There is still migration every week. For example, a large DeFi platform migrated $1-2 million Sai. The peg is drifting and getting weaker, though.
- L: I saw a large trade on Uniswap for $.90
- C: That provides incentives for CDP holders to repay.
- S: Unless they’re waiting for ES, where they get fees forgiven due to peg drift.
chat Does it make sense to publicly say that the Sai peg is not reliable, or is becoming less and less reliable as the supply dwindles?
- C: People selling Sai right now are bearing the cost of forgiven fees. If it drops to $.90 for a more extended time.
- D: Who is the buyer?
- C: Who is the seller? Why are people selling Sai?
- Matthew: Going back to our core objective, if the risk profile stays the same, and we keep lowering the debt ceiling to amount outstanding. Risk is hedged, and the system is exposed. The risk doesn’t go away, but it just shuts it off. The SFs get waved. When I went through this cycle, asking for opinions from other professionals, we asked: what advantage do we gain if the risk doesn’t change? The only thing that changes is recouping fees in MKR. If the peg drifts no one is using it much right, it’s unusable?
- C: It’s not great for the Maker brand to have an old version and see the peg drift. Secondly, if ES isn’t going to happen soon, that hurts the peg further.
- M: Trying to find a perfect solution isn’t going to work. It’s a communication issue.
- D: Does it make sense that if the migration contract exists and significant liquidity is in it and available, that we’re still providing an “out” and taking care of the end-user?
- L: We’re assuming bad actors, but there hasn’t been enough liquidity for someone as large as 3088.
- M: Tools exist for partial transfers.
- D: Also, we’re not here to guess motives. We just provide a solution where possible.
- L: 3088 could be waiting for ES without having to pay a ridiculous amount. Trying to source his balance on the open market, it’s also not in the migration contract, is going to be impossible.
- Dai Supply went up about 5 million to a total of 111.7 million Dai.
- Net minting was not that high, ~2 million Dai over the last week.
- There is a lot of Sai being migrated to MCD, about 3 million over the last week.
- Two days ago, a large sum was migrated from Nexo, roughly 1.4 million Sai.
- About 1.7 million of the 3 million Sai was used by CDP’s to migrate to MCD as well.
- Two larger CDPs migrated last week, paid about 100,000 in fees.
- The total supply of Dai+Sai is still increasing at a similar rate. Only a net 2 million Dai was printed last week, which is low considering the recent ETH price rally. Historically, large increases in ETH correspond to increase DAI minting.
Defiexplorer.com/stats shows aggregate repayments and draws for collateral being added in MCD. This can be used to check how much Dai was generated in MCD.
- You can see a large withdrawal of ETH; this pushed the amount of ETH locked in MCD bellow 2 million. I’m guessing that this is happening because the CR is very high, at 381%, CDP owners are withdrawing the unnecessary capital from collateral.
What surprised me is how high the Sai CR is at 427%, as it is usually lower than the Dai CR. I’m guessing this is due to general inactivity in SCD - Half of the debt is in CDP’s that are inactive for more than six months. These CDP’s are simply not migrating. They aren’t even withdrawing collateral.
Sai DC is at 30 million. As the Sai supply is at 23.2 million, the SCD DC leaves about 8 million.
- Cyrus: Wouldn’t the Sai CR be one of the few indicators of CDP owner activity? That behavior shouldn’t differ that much from MCD to SCD. You can’t go by the Dai supply because they won’t be adding or wiping debt due to risks from the peg and potential tax.
- P: Exactly, this was my point. If you think about it, we’ve seen even higher CRs. The ratio itself is related to leverage. One theory is that they are not active. The other is that they are ok leaving ETH in collateral without removing it and converting it to something stable. This might tell you that they are really bullish. I’m leaning toward the first option that they are inactive.
- C: If they wanted to leave it in deliberately, then so would the MCD CDP holders. Why would there be a difference?
- P: True.
There is a buffer of 8 million available for Sai minting, including Sai burned in the migration contract. This should be enough if there is a sudden SAI liquidity shortage if new Sai needs to be minted.
- On the other hand, there are debates that perhaps we don’t want to have too high of a buffer. We don’t want to see unnecessary Sai printing.
- Because of this, I checked if there was any Sai minting in the last few weeks. From the top 100 CDP data from MakerScan, you can check how much is minted. The numbers are actually low; I haven’t noticed any new SCD CDP’s trying to mint a significant amount of Sai.
- At most, you can see some CDP’s using SCD and minting a few thousand Sai.
- Dai in DSR at an all-time high at 55 million Dai, probably because the DSR is also at an all-time high at 8.75%.
- I noticed that on Compound, the DAI lending market surpassed the USDC lending market.
- Just last month, I saw the Dai supply in Compound go from 20 to 30 million. I’m also guessing this ratio will increase in favor of Dai vs. USDC. The last time we saw Dai lenders dominating USDC was five months ago, even before the migration.
- I believe the Dai rate will dominate because the supply rate is much higher for Dai. The unutilized part goes to DSR. In Compound, this isn’t the case.
- Sai still has some liquidity in Compound, not sure why this is not reducing. In the last few weeks, remaining Sai liquidity has become sticky.
- cDai supply in Compound increased to almost 30 million in USD.
- This graph shows grouped addresses by the amount of Sai held. There are three categories: those holding 10,000 Sai to 100,000 Sai, ones holding 100,000 - 1 million Sai, and the ones holding 1 million+ Sai. These 300 addresses hold 22 million of the Sai, which is 90% of the Sai supply.
- 300 addresses are represented by these wallets.
- Only 100 Sai holders represent 80% of the supply.
- Top 100 CDP holders reflect a similar metric, with 85.58% of Dai held.
- We are waiting for 100 Sai addresses to match with the 100 top CDP’s. If they migrate all at once, we would only be left to 10-15% of supply available currently at a few million Sai.
- Estimating 200-300 people who we are waiting for migration of the last few million Sai.
- I’ve checked which CDP’s have migrated. This table showed two weeks ago when I wanted to estimate how the debt is structured in CDP, including who is wiping the debt and minting the debt. I wanted to see what changed in the last 17 days.
- I’ve noticed that migration is still going on. Since Jan 20th, 8 CDP’s have migrated for a total of 4 million Sai debt repaid and 336,724 in fees paid.
- These CDP’s who have migrated were active during the migration period. We are still not seeing any activity in the formerly inactive CDP’s, even in the last few weeks.
- Around 10 CDP’s migrate daily, proving that migration is still under way. I’m not sure whether this page shows partial or complete migration.
- Here I’ve noticed a large increase in migration contracts. This is the spike I noticed two days ago when Nexo migrated.