[Agenda/Discussion] Scientific Governance and Risk #167 - Thursday, November 11 17:00 UTC

Information

This communication is provided for information purposes only. The views and opinions expressed in this meeting are those of people involved, and do not reflect the official policy or position of MakerDAO or any of its contributors, Core Units, or affiliates. This communication makes no representations about the enduring accuracy of the information or its appropriateness for a given situation. This content is provided for informational purposes only, and should not be relied upon as legal, business, investment, financial or tax advice. You should consult your own advisers as to those matters. References to any digital assets and the use of finance-related terminology are for illustrative purposes only , and do not constitute any recommendation for any action or an offer to provide investment, financial or other advisory services.

The zoom waiting room will be on, and a password is set to: 748478, please ping us in the Maker Discord’s #governance channel if you aren’t let in from the waiting room.

2021-11-11T17:00:00Z

Slides - updated before the call

Introduction

Governance Round-Up

Selected Updates / Discussions

  • SAS-001- Sidestream Auction Services Demo
  • @Oracles-Core-Unit - Oracles; Costs, solutions, next steps.

Other Discussions and General Q&A

  • (Time permitting, after general Q&A) - Ilk Hole size, Keeper health, liquidation ecosystem discussion.

Leave your questions in our anonymous question box and we’ll do our best to bring them up during the call.

1 Like

G&R #167 Snippet

This snippet includes Governance, MIPs, forum updates, and Core Unit team discussions from the MakerDAO Governance and Risk Call #167.

General Updates

Votes

Executive:

  • Last Week’s executive - PASSED & EXECUTED
    • Parameter Changes from MOMC
    • PSMs’ Fee In (tin) set to 0%
    • Core Unit Budget Transfer - DUX-001
  • Executive proposal is up for vote tomorrow. Will include:
    • Returning DAI from the Pause Proxy
    • Parameter Changes
    • GUNI, WSTETH, and D3M

Polls:

  • 5 Weekly Polls - PASSED
    • Add WBTC-B as a new Vault Type
    • GUNIV3DAIUSDC-A Parameter Adjustments
    • Recognised Delegate Compensation Increase
    • Increase the WSTETH-A Debt Ceiling
    • DIRECT-AAVEV2-DAI Parameter Adjustments
  • 1 Weekly Poll - REJECTED
    • NS-DROP Covenant Modification
  • 2 New Greenlight Polls (Vote ends on Monday)
    • MDI (MD Irradiance LLC)
    • OHM (Olympus DAO)
  • 8 Ratification Polls close 2021-11-22

MIPs

Weekly MIPs Update #61

  • The Ratification Polls for the November Governance Cycle are now up!

Forum at a Glance

Forum at a Glance: November 4th - November 11th

Team-led Discussions

Sidestream Auction Services Demo

Daniel from SAS-001

  • Daniel from SAS then presented a pre-recorded video showing their product.

Oracles Core Unit (MIP40c3-SP45)

Nik

  • Gas costs for Oracles are currently not included in the current Oracle budget.
    • This will come later.
    • Maker Foundation generously provided these gas costs to help provide Oracles CU for time on a best-decision approach.
  • Asking for a rolling 3-month budget.
    • $3,568,849.20, which includes operational and gas costs.
      • $3,033,522 if including refunded funds by Maker Foundation over gas cost.
  • Will submit their MIP soon.
  • Their MIP will provide a lot of perspective and data on spending costs for gas and Oracles and factors that affect Oracle costs.
    • Nik breaks down the medianizer gas costs and Oracle Security Module Uni LP + Spotter gas costs.
    • Provides gas distribution of the OmegaPoker contract.
    • Provides a quantitative observation and analysis of Oracles.
    • Estimated total annual Oracle cost: $7,236,833.10
      • This estimate does not account for rising gas costs.
  • $1,784,424.60 emergency fund for unique market or gas events.
  • 1/3rd of the cost comes from Oracles, and 2/3rd comes from the maintenance and security of Oracles.
  • Nik will be posting their MIP on Forums soon for more information and continued discussion.
1 Like

Semi-transcription

Episode 167: November 11th, 2021

Agenda

  • 00:00: Introduction
  • 01:31: Votes and Polls
  • 04:43: MIPs Update
  • 07:09: Forum at a Glance
  • 12:20: Presentation - Sidestream Auction Services Demo
  • 26:29: Discussion - Oracles Core Unit (MIP40c3-SP45)
  • 1:19:32: Conclusion

Video

https://www.youtube.com/watch?v=sylJ74tGSgE

Introduction

Agenda and Preamble

Payton Rose

00:00

  • Hello, welcome to the MakerDAO Scientific Governance & Risk Meeting #167 on Thursday, November 11th, at 17:00 UTC. My name is Payton, and I am one of the governance facilitators at MakerDAO. Feel free to ask questions during the call by raising your hand or submitting it in the chat. Thank you.

General Updates

Votes

Payton Rose

Executive:

  • Last Week’s executive - PASSED & EXECUTED
    • Parameter Changes from MOMC
    • PSMs’ Fee In (tin) set to 0%
    • Core Unit Budget Transfer - DUX-001
  • Executive proposal is up for vote tomorrow. Will include:
    • Returning DAI from the Pause Proxy
    • Parameter Changes
    • GUNI, WSTETH, and D3M

Polls:

  • 5 Weekly Polls - PASSED
    • Add WBTC-B as a new Vault Type
    • GUNIV3DAIUSDC-A Parameter Adjustments
    • Recognised Delegate Compensation Increase
    • Increase the WSTETH-A Debt Ceiling
    • DIRECT-AAVEV2-DAI Parameter Adjustments
  • 1 Weekly Poll - REJECTED
    • NS-DROP Covenant Modification
  • 2 New Greenlight Polls (Vote ends on Monday)
    • MDI (MD Irradiance LLC)
    • OHM (Olympus DAO)
  • 8 Ratification Polls close 2021-11-22

MIPs

Pablo

4:48

Weekly MIPs Update #61

  • The Ratification Polls for the November Governance Cycle are now up!

Forum at a Glance

Artem Gordon

7:15

Post: Forum at a Glance: November 4th - November 11th.

Video: [Soon!]Forum at a Glance.

Team-led Discussions

Sidestream Auction Services Demo

SAS-001

  • Thanks for having us. I am Daniel Kremerov, the facilitator of Sidestream Auction Services. We are an SES incubation team that is going to form a new CU. We joined the incubation in the summer. Since Monday, we have been up for voting and hopefully will be a permanent CU as of December.

Questions

23:27

  • LongForWisdom: I remember when we initially put enabled flash loans auctions, we were not concerned, but we were aware that MEV might be an issue, having flash boats and things front running the flash loan auction transactions. Have you thought about that at all?
    • Daniel: Can you rephrase the question about if we thought about front running by bots?
  • LongForWisdom: Because the flash loan auction transactions are pretty easy to front-run. They are self-contained, and they do not require any capital. This UI looks awesome. But if every time anyone uses it, the transaction fails, then that is not ideal.
    • The main purpose of the UI is to have another backstab solution. If there is no participation that you can participate in through the UI, then the next step will be providing scripts or APIs to participate in these auctions. Currently, we do not see front running as a problem. But rather, it is good if someone is bidding on these auctions.

25:06

  • Rema: First of all, this is great for our keeper ecosystem. You said that you are only supporting Uniswap. Is it v2 or v3? Does the price in the UI include slippage as well? Price slippage when selling this collateral?
    • Lucas: It is a v2 for now. We are planning to upgrade, to support also automatic routing at some point. It includes the slippage, so it is calculated based on the amount sold using the Uniswap SDK.

Oracles Core Unit (MIP40c3-SP45)

Niklas Kunkel

  • Unfortunately, I do not have sexy presentations and videos like everyone else. I have MIPs that are pretty text-heavy, but that will do it.
  • We talked about this on previous calls. Right now, the gas costs for the Oracles are not included in the current Oracle CU budget. Those would come at a later date. Also, Maker Foundation generously offered to front the gas costs since August 1st to give the CU more time to figure out how to do this best. The tooling that we had back then to delve into, what Oracles cost us from an infrastructure point of view we are not very granular. We had the holistic view that we know how much ETH is going in and how much we are spending in total. We did not have a more granular or scalpel view of the ETH Oracle to be costing us this much, the ZRX Oracle costing us this much, and the Uniswap LP Oracles this much. We have recently started creating tooling and dashboards that give us insight into these types of things. It is time to finish the transition of the Oracle CU and the Oracle infrastructure out of the foundation and have it be completely decentralized. We are thinking of structuring this so that we will typically be asking for a rolling three-month budget. We will submit a MIP, a sub proposal, and we expect to spend this amount in the next three months. What is different from this first proposal is that it will include the repayment to the Maker foundation for fronting the cost since the beginning of August. Since that is a rolling timer, we assumed that the MIP or some version of the MIP should be passed due to how governance works by the end of December. We will say in the MIP that the cutoff for that is the end of December. If it gets passed, earlier or later, we will have to adjust the payment amount to the foundation.
  • What I want to get across today with this MIP is the analytics. Where they are coming from, and how we deduce the numbers in the budget for the MIP itself. These are dune analytics that we have hooked up to the smart contracts. We can provide the link for these. You can verify the data for yourself.

Questions with Niklas

50:11

  • Aes: Was your assumption on doubling costs based on the actual costs for the past 99 days? Or is it based on the current run rate?
    • It is based on the current run rate. The sampling since August is not that relevant. If you zoom out from the chart, it is very volatile. Using that type of limited data set, trying to establish some kind of trend is quite foolish. I would not focus too much on the 2X number; it is a spitball number. There is not a big data set to come up with anything more algorithmic or deterministic.

51:12

  • Aes: I am looking at the average gwei assumption used for the forecast. Right now, the average gwei is around 175 a day. Would that imply about 350 for the doubling?
    • It depends if you want to use the today value or write the sampling over the last period. I do not take much stock in the gas price today because a week ago, I saw like 60 gwei.
  • LongForWisdom: It takes a while for the gas prices to adjust to the new ETH price and stuff.
    • These are the numbers that governance should consider when thinking of onboarding a new collateral type. For example, ETH has cost us historically 330k since August. If now the gas prices double, you are looking at 660k. Any Vault that is going to make less than 660k you should not bother. You are coin flipping based on if the gas price will not go up and screw us. It would be best if you thought of only the big-ticket Vaults that can generate at least 2 million of stability fees, at the very least. You are still paying people very high salaries to onboard those things. You should be thinking higher.

53:35

  • Aes: Is the plan also to buy a lump sum of ETH essentially? How do you think about the technical funding of it?
    • I did not put that implementation in the MIP because I wanted to give the CU more flexibility in trying out different solutions in that regard. We would want some of it in ETH, some of it in Dai. The rough number off the top of my head would be like a 50 50. You want some exposure to ETH but not all of your exposure in ETH. Ideally, in the future, we get actual gas features that we can use to hedge our gas costs, like airlines use oil features to hedge their jet fuel costs. We will try a few different types of models to see what works better. The goal is not to get the ultimate lowest cost for running the Oracles; it is more just to bound those costs, to make them more deterministic. The more we can constrain the range of costs, the better off we are.

55:19

  • Aes: We can hedge it simply by holding ETH on the balance sheet. The more ETH goes up, our balance sheet assets will go up, which can also be used in the feature to fund Oracle fees. Something that we can discuss on the forums.
    • Let us discuss it on the forum. I do not want it all in ETH because of these Black Thursday-type scenarios where everything is crashing. Prices are going down. And yet, gwei could be obscenely high, with people willing to pay $5,000, $10,000 transaction fees. If we are completely in ETH, we would be screwed by that type of event.

56:23

  • David Utrobin: I always imagined that the Maker protocol could use the surplus buffer directly like a split solution, where some of it is held in the Surplus Buffer, a little goes on MKR, and then a little bit is used to buy ETH dollar-cost averaging over time, and making a native pool within the protocol of ETH that can automatically pay Oracles. What do you think about that vision?
    • It is something that community should discuss more. My initial inclination is that we should not do that. The purpose of the CUs is to delegate the responsibility of things. Governance does not have to be an expert on all of this different stuff. Taxing governance with the inventory management for the Oracles would be a mistake. If governance wants to have exposure to ETH and diversify some of the Surplus to ETH, have at it. I would not do it specifically because of Oracle ETH inventory management.

58:13

  • Prose11: In terms of the Uniswap Oracle, you mentioned the v2 ones and their costs. Are they any cheaper when we are already running Oracles for the underlying assets?
    • This was purely for the gas cost of running that Uniswap Oracle smart contract. If you need to create new medianizers for the component tokens and those, then go ahead and add the medianizer costs on top of that. We do not currently have that for any of the tokens where we onboarded a Uniswap LP token pair, and we did not have the medianizers already, quote-unquote, for free because we would have already onboarded those assets in Maker.

59:36

  • Nadia Alvarez: About onboarding new collaterals. It would be great if you could add the cost of that collateral to your risk assessment. I was surprised by the Gelato LP token.
    • Those are expensive. It is 377k for the GUNI one, minus 5% uncertainty because the sample size is relatively small. That is the annual cost.

1:01:03

  • LongForWisdom: The stability fee is currently more than that, nice.
    • Until we just rugged it because of PSM fees. We will have to see going forward how much it generates.

1:01:20

  • Prose11: Would that be similar for other v3 implementations? Or is it unique to Gelato?
    • I do not know. For example, Kurt is creating the curve LP Oracle. We could do the stake ETH ETH curve pool as a collateral type. People can get staked ETH yield and the curve Lido yield. We do not know beforehand how much demand there will be for that Vault and how much Oracle will cost until we see it running. There will be cases where it is not clear-cut for governance on the costs and the upside. We have to make some educated guesses at some point. For this other stuff, the standard Oracles, a typical collateral type, we know the cost very precisely, to a degree of uncertainty with gas prices. We have experience in knowing how much this will cost. It is better not to bother if the collateral type does not generate 2 million instability fees.

1:03:27

  • PaperImperium: Do you know how many hours it requires to build the Oracles? Can we add those to the estimated cost of onboarding?
    • Once you have built it, you have built it. The costs to deploy another one is relatively low; they are incremental costs. Building it out the first time, getting it audited, and feeling comfortable that it is secure is costly. There are things that the CU can add to its roadmap to limit some of the costs that we spend on Oracles. We can, for example, offboard some of the unprofitable Vaults. That is gas that we would not be spending on those vaults. If governance does that, we might not spend this estimated budget. We would be spending less. We can make tuning parameters on the Oracles to make them cost a little bit less. Not 50% less but 20 or 25% less. That is a discussion that has to be had with Risk. When decreasing the sensitivity of the Oracle, how much more risk does it put on the Maker protocol in terms of holding assets longer than it would otherwise. It comes with the decreased operation cost for the Oracles. What happens with L2? That is still an open question. We suspect that L2 Oracles will be cheaper than Layer1 Oracles. If there were to be some Maker, either the full protocol or some shell of a protocol, on Layer2, L2 could support collateral types that L1 cannot support because of the Oracle costs. I want to caution of thinking of L2 as this. To think that is going to fix all of our problems. I want to caution against that for the following reasons. L2 segregates computational costs from storage. You do not have to do computation on L1, which is expensive. You do the computation on L2. All call data, including the data passed to the smart contract on Layer2 to do those transactions, needs to be recorded on Layer1. Oracles are very heavy on the call data; at least, the medianizers are. LSM, and spotter not so much. The reason is that you need to get a consensus of a bunch of feeds. If you have 26 feeds, you need at least 13 to reach a quorum of ETH’s price. For each of those feeds, you need the price, the timestamp, and the signature saying that I have verified through my identity the time, price, etc., which you need for each of the 13 feeds. Much data gets passed to these functions. We will have to do some experiments to see the realized costs on L2. I do not want to promise the world and then have it come short. There will be cost savings. I do not know to what magnitude yet.

1:08:19

  • MakerMan: Do we have a number per collateral what it costs? What amount of ETH do we need to operate the Oracles per collateral type? That should be roughly linear. A year and a half ago, everyone wanted to add lots of collateral types. I was saying to be careful. If they do not make us money, it is going to cost us money. It is costing us money. We have got a fixed cost ETH per collateral type. So we can target the amount of Dai that we think we will earn by putting up a collateral type.
    • I agree. Gas costs used to be a lot lower. From an economics point of view, things that made sense back then now make less sense or do not make any sense. We can stop pursuing things of low economic return just for the sake of doing them. We can focus on bigger targets. There was this whole drama with the Matic token because there was an agreement that they would mint a lot more Dai, and we onboarded them based on that agreement. They mint some Dai, but they did not mint enough according to the agreement that we had. Given how much they have borrowed, it is an unprofitable vault. Unless they can be convinced to uphold the original agreement and mint significantly more Dai than they are doing, we should offboard them because it is unprofitable.
  • MakerMan: When gas was one gwei, and ETH was like $200, we had a completely different ecosystem and network than 100 gwei with ETH at $5,000. Unless something radically changes, we could have ETH at $10,000 and be at 200 gwei in a heartbeat. Do we know how much ETH it takes for a particular collateral type? So that we can target and try and get to levels. I was looking at the MATIC Vault. There were 79 million in collateral. They were minting about nine, and their cap was 10. They do not feel like they can move. They wanted to go beyond 10. We can address these issues. We need to think about a future with a higher ETH price and higher gas costs and work against what Vaults would have to maintain profitability. We might get to a point where we have to run on profitably on Oracle’s because gas prices might be high temporarily. We do not want to close Vaults because they are slightly unprofitable, not if we can make it grow. If we cannot, then we need to ax them. There are several Vaults that we need to look at carefully to ax. Is there a number on ETH for one collateral type? What does it cost us an ETH for collateral type? We can run these numbers in the future.
    • There is no consistent ETH number. It is not like it is 100 ETH to run this oracle, and if the price of ETH goes up or down, it does not matter; it is 100 ETH. That is not how gas prices work. They are highly volatile. The causality here is questionable, but there is a relationship between ETH going up and gas prices going up. Maybe it is that markets are going up there, it generates more interest, more activity, and therefore gas prices go up. There is some correlation, but it is not absolute. There are events like Black Thursday where ETH can go down, and yet gas prices go inverse. However, gas prices could go sky high. We need to have a plan in place. Maybe it is okay for Vaults to be a little bit unprofitable for some time. We need to define what is unprofitable that we are willing to ETH and what we mean by some time. We have to make sure that there is a straightforward course of action when we violate those thresholds. So we do not spend two, three months of bureaucracy trying to decide what to do. There should already be a plan in place, and we execute on it.
  • I am trying to get a handle on cost per collateral type within medianizer. If we do “N,” what is the cost for N+1. There is a total cost for the whole thing that is probably fixed to some element. But then there is a cost per collateral type. What does it cost for us to have the system and add individual collateral types? All this other stuff comes out of that.
    • Those are the numbers that I presented in the MIP. I will be publishing this. You can take a look at those numbers more closely. But they are just what we have seen in the past. They are not indicative of what we see in the future. You have to look, for example, at the number 330k and think how it scales, not fixating on the number. If we assume an awful scenario where gas prices triple, then it is a million. But that is what it costs us to onboard a collateral type today. It is not what it might cost us a month, three months, or a year from now.
  • LongForWisdom: We cannot do this because the cost is denominated in gas units. The price of gas changes as gwei changes, and the price of that change is based on ETH. We could figure out the total gas units each thing costs per month, a fixed amount. But you cannot figure out a fixed amount in ETH terms or dollar terms.

1:16:42

  • Joshua: I was living in this fantasy world where I thought level-2s would solve our problems. The stock exchanges have 1000s of stocks, like Russell 3000 and New York Stock Exchange. Do we have a plan for supporting that throughput?
    • It remains to be seen. Sharding, which is this fairy tale unicorn thing that people talk about, is still years off, which helps with the data component. L2s are supposed to solve the computation component. Shards solve the data problem. Are there some things that maybe we can do? It is all trade-offs and compromises. It is what compromises are. Is Maker governance willing to accept in other areas in order to reduce things like Oracle costs? For example, we looked at the medianizer cost being a third of the total Oracle cost of this Oracle Security modules spotter combination. If governance is willing to go an Oracle attack on certain Vault types that are maybe smaller in debt ceiling size, you could hook up a medianizer to a Vault. Is that a good trade-off to make? I do not want to comment on that, and everyone will be very opinionated on that. There is no magic bullet. It is where you want to pile the risk up.

Conclusion

Payton Rose

  • Thank you, everyone, for coming and engaging. Please let us keep the discussion going on the forum. We did not get to a few topics today; we will include them in a prompt governance call.

Suggestion Box

Common Abbreviated Terms

CR: Collateralization Ratio

DC: Debt Ceiling

ES: Emergency Shutdown

SF: Stability Fee

DSR: Dai Savings Rate

MIP: Maker Improvement Proposal

OSM: Oracle Security Module

LR: Liquidation Ratio

RWA: Real-World Asset

RWF: Real-World Finance

SC: Smart Contracts

Liq: Liquidations

CU: Core Unit

Credits

  • Artem Gordon produced this summary.
  • Kunfu-po produced this summary.
  • Everyone who spoke and presented on the call, listed in the headers.​

The full call is available for review on the MakerDAO Youtube channel:

1 Like