Proposal: Reimburse Risk Assessors for Undue Loss from Claim 113

Hey community,

I’m the director of Ease/Armor and believe Nexus Mutual stakers did not get fair treatment with the passing of a recent claim. This is not a proposal to refund full loss from stakers (the point of staking is to payout funds fairly), but a proposal to reimburse funds that were taken over the amount owed.

Background

After the Rari Capital Exploit, Claim 113 was submitted for a cover policy of 5MM DAI. The member who submitted the claim did experience a genuine loss of about $1.4MM USD worth of assets–about 28% of the cover value. With Protocol Cover, a 20% or greater loss is required, as defined by the term “Material” in the wording of the cover policy.

The claim was passed, resulting in an excess payout of ~3.6MM DAI. Given the massive overpayment to the claimant, Nexus Mutual risk underwriters suffered a large unjust loss. I ask for the community fund to be used to compensate affected Rari Capital underwriters for an amount equal to what was taken as overpayment: 43.94k wNXM.

Abstract

  • Risk assessors lost millions of USD in NXM more than actual loss.
  • A partial payout was possible.
  • The policy terms state a partial payout should happen if possible.
  • The policy was automatically accepted with a small % of quorum before discussion concluded.
  • This event hits risk assessors extremely hard if more than the fair amount of funds are taken.
  • Other protocols are currently working on ways to compensate their risk assessors for unfair loss from other events, resulting in very bad optics for Nexus when capacity is the biggest bottleneck in growth.
  • The community fund has a history of compensating users for shortcomings of the technology.
  • The community fund should compensate risk assessors for their undue loss.

Policy Wording: Supplementary Claims Amount Guidance

While the claim was being assessed there was active discussion in the Claims Discussion discord channel about the need for the Mutual to make good on this wording, as both policy holders and risk underwriters are making their decisions to provide cover and be covered based on the official documentation provided to its members by the mutual. Given policy wording, an underwriter would not expect to be liquidated 5x the amount necessary to make a claimant whole. The argument by some is that Nexus currently does not have an ability to pay partial claims, as they are programmatically designed with a binary “yes”, “no” system. Given the “Material” loss requirements were met, a “yes” was voted, and the claim passed, even though quorum was not even close to getting reached.

However, a workaround was identified that would have provided the ability for the mutual to make partial payment. The purchase of a new policy by the claimant equal to actual loss + policy cost. That policy could have then been immediately submitted for assessment. This would have allowed the Mutual to make a partial payment equal to the real loss of the original policy. There is no programmatic issue in place that prevents a new policy from being immediately submitted for a claim. Given this, the possibility of partial payouts is possible by the mutual at this time, and the steps should have been taken to do so. Since they were not, risk underwriters deserve to be compensated for the unjustified loss that was outside of the terms of the policy wording they agreed to underwrite.

Despite this, the claim was automatically accepted with ~6% of claim assessors quorum and ~0% of “all members” quorum and before a full discussion could be had. For consideration, affected stakers likely would have voted the other way with ~66% of quorum had the discussion had time to finish.

Precedence: Cream Hack Restitution

There is past precedence for the fund to take action to help mutual members when the programming of the system falls short, as seen with the unanimous passing of paying out 10 ETH to an affected member by the CREAM hack. The argument used for the community fund to pay this is that the code does not currently have an ability to pay partial claims.

On claim 113 the “Material” loss requirements were met, and a “yes” vote was passed. However, this caused a similar situation to the issue suffered with Claim 102, except with the roles reversed. Risk underwriters were liquidated for $5MM USD in NXM for what was equal to a $1.4MM USD loss, almost 5 times what was necessary. Given that the Mutual’s discretion in the past has directly sought to compensate its users for the failings of its programming, that same discretion should be used here, especially with V2 on the horizon.

Act of Good Faith Toward Underwriters

It is widely known in the DeFi coverage industry that one of the biggest hindrances to growth of the space is capacity. If underwriters are at risk to be liquidated for a maximum of 5x the amount necessary, it will cause a large lack of faith in the system. This loss largely impacted arNXM users, who currently provide ~40% of the staked capacity and now have a -10% return from staking with Nexus for the past 1.25 years. A lack of action to compensate them for unnecessary losses could have severe consequences on the mutual’s future ability to provide capacity.

This is especially important given that the mutual already acknowledges this shortcoming of the system through the planned implementation of v2 and with the community fund paying out to circumvent this shortcoming for policyholders in the past. Nexus Mutual risk underwriters are just as important as policy holders, and equal measures should be taken to look out for both parties, as all are members of the mutual.

Furthermore, the underwriters that bore the brunt of this payout are particularly useful to the protocol because the Armor/Ease core team works closely with the Nexus Mutual core team to stake funds in the most useful ways. When the Nexus team receives requests by large clients to purchase coverage, we’re often contacted and asked to add capacity to the protocol pool for these clients; this relationship gives Nexus Mutual a clear avenue to satisfy client needs.

Comparison to Other Protocols

Other DeFi coverage protocols, such as Risk Harbor and InsurAce, have announced plans to compensate underwriters that were liquidated more than anticipated in the recent UST de-peg event.

Risk Harbor: Discord

InsurAce: UST De-Peg Update – May 22 – InsurAce.io Blog

Because they feel that their underwriters were mistakenly over-exposed to risk, InsurAce is taking major steps to ensure stakers feel comfortable and remain in the system to the point where they may be securing capital specifically for this purpose. Nexus Mutual has a community fund that’s been used for similar purposes and can easily afford the compensation to achieve the same goals. If this is not done, the optics look very bad for underwriters thinking about staking with Nexus Mutual.

Potential Against Arguments

Programming working as intended:

Some users stated underwriters had an understanding of risk taken on when staking into the platform. I believe that the risk understood by underwriters is not equivalent to the loss incurred by underwriters in this event. As pointed out by the policy wording, an underwriter would expect the claim to be partially paid out if possible. This thought process can be see in some new users on discord even:

It’s important for the mutual to payout exactly according to the terms of the coverage, as should be expected by the terms agreed to by policyholders and underwriters alike.

Second, as shown with the precedence of the restitution provided to CREAM users, the mutual has previously stepped in to compensate affected members when the programming falls short of what is needed.

Precedent Set for Partial Payouts in the Past/Future:

I believe, if terms are adjusted for the time between now and V2, partial payout precedent will not affect future cases. If a similar extremely impactful case happens again, that can be looked at independently. Regarding the past, I’m not sure when the partial payout language was implemented, but there are no events where underwriters were hurt to the same extent. As seen here,

https://twitter.com/NexusMutual/status/1527017855308353536?s=20&t=n5BlC34PHCI6qvLkVsEO1A this payout was unprecedented, and accounted for $5MM of the $8.1MM Nexus has paid out to date. For any partial payout issues prior, the total amount of paid out events is limited and can be dealt with fairly easily if needed.

In Summary

Nexus Mutual was created to protect users from undue loss. This claim created undue loss multiple times larger than the initial loss. This occurred simply because of current limitations of the product, which are in the process of being fixed, but it will forever be looked back at by potential underwriters as an example of how the mutual functions and where its priorities lie.

It’s extremely important to be fair and follow the terms in this situation. Not just from a moral and legal point-of-view, but also for the future of the protocol. Institutional capacity will likely be much harder to bring into this industry than institutional clients, as institutional underwriters are wary of catastrophe coverage overall, much less catastrophe coverage with little historical data on its risks. We do not want to give future underwriters the impression that their funds matter less to the mutual than paying out claims.

It was possible for Nexus to do a partial payout for this claim. Nexus had a duty to partially payout according to its terms. I believe that claims assessors would have voted for a partial payout had there been adequate time. Not rectifying this situation will be very bad optics for Nexus. The community fund has all the means to compensate stakers. I believe stakers should be compensated for their undue loss.

@robert, thanks for the post and putting this detail together. I will try and keep my response as succinct as possible.

  • I disagree on the framing that the mutual has the ability to pay partial claims. This is an extreme interpretation of a very technical workaround that breaks all existing workflows and requires coordination with every claimant and all claims assessors before any claims are submitted.
  • Armor (and other stakers) have benefited from not paying claims under the 20% threshold in the past, so there are two sides to this argument. You can’t have it both ways.
  • Nexus V2 is close, so any precedent set by agreeing to any one-off reimbursement will be limited.
  • It’s unclear at this stage if the claimant is willing to refund any of the claim amount and/or if Fei will reimburse users.

Overall, I see both sides of the argument here and as such I am open to partial reimbursement of burnt NXM for all impacted stakers on Rari after the situation with the claimant and any potential reimbursement from Fei is finalised.

Using Armor as an example, but noting this should apply to all Rari stakers:

  • Armor’s extra burn is approx 44,000 NXM
  • Use any reimbursement from the claimant and/or Fei to purchase wXNM on market and refund stakers, including Armor, proportionally. For the sake of argument, say this amounts to a 20,000 NXM refund for Armor. (Note: this could be zero)
  • Use the Community Fund to refund some portion of the remaining amount, eg 50%. So in this example 24,000 NXM extra burn remains after reimbursement so 12,000 NXM could be paid from the Community Fund.

I think the Community Fund refund proportion (the 50% suggested) should be discussed after understanding the situation with the claimant and Fei.

Appreciate your perspective and understand your points. Thank you for taking the time to outline your point of view on this claim payout. I just wanted to clarify a few things mentioned in your post:

Actual Loss vs. Claim Payout

  • The member lost $1.4m due to the Rari Fuse exploit, which represents 28% of the total cover amount. The $5m payout was 3.5x the actual loss, not 5x. Important to be accurate on this point.

  • In order to pay the claim from the Community Fund, members would have needed to deny a legitimate claim and pass a separate Snapshot vote to compensate this member from the Community Fund.

  • Denying a legitimate claim sets a bad precedent for the mutual. The 20% requirement provision has been in Protocol Cover wording since launch and members voted to launch Protocol Cover through an on-chain vote. This is noted in the Annotated Protocol Cover wording in the docs.

  • The Supplemental Guidance section in Protocol Cover wording is future proofing for when partial claims are possible once the technical capabilities are present in the smart contracts. Again, this is coming with Nexus v2.

Claim 102 vs. Claim 113

These two claims are not equivalents, and it’s important to note why that’s the case.

The member who filed Claim 102 experienced a genuine loss but did not meet the 20% loss requirement present in the Protocol Cover wording given partial claims are not yet possible in Nexus v1.

Because Claim 102 did not meet the technical requirements to receive a claim payout through the traditional Claims Assessment process, members declined that claim but created a community-driven proposal to compensate this user for their loss from the Community Fund.

Claim 113 did meet the requirements laid out in Protocol Cover wording, and it met the 20% or greater criteria as stated in the “Material” section of Protocol Cover wording. Members discussed this case and voted through the Claims Assessment process.

When Claim 113 was filed, the mutual released an announcement in Discord and on Twitter and Telegram announcing that the claim was open to voting:

After 72 hours passed and the voting weight in NXM did not exceed 5x the cover amount, the vote when to a full member vote, as outlined in the documentation: https://nexusmutual.gitbook.io/docs/claims-assessment/claims-assessment#claims-assessor-voting

The mutual made an announcement of the transition to a full member vote as well:

To quote from the Member Vote section of the docs: “As per the Claims Assessment vote, the voting period lasts for a minimum of 36 hours and ends when either the total voting stake reaches 10x the Cover Amount or 72 hours have passed, whichever occurs first. A simple majority outcome is used to determine the claims outcome. Should the minimum 5x voting weight not be achieved in the member vote, the claims outcome is then determined by the earlier Claims Assessor vote using a simple majority basis.”

This process worked as designed, and the vote was in favor of a payout, which occurred according to the Claims Assessment design.

My Personal Views

  • I disagree with your assertion that partial payouts are possible. They are not and won’t be until Nexus v2 is live. If a claim meets the requirements in the Protocol Cover wording and members vote “Yes” for a payout, then the claim is paid out.

  • As Ease/Armor is acting as a pseudo-staking delegation module for users, it’s your protocol’s responsibility to review the risks associated with each protocol/platform you stake against. When claim events occur and members vote, all outcomes are decided by members of the mutual and the risks involved with Risk Assessment are noted in the docs.

  • Notice was given to members regarding the initial filing of Claim 113 and the transition to a full member vote. The full member vote operated as documented and as planned, per the parameters set in Claims Assessment. Deadlines can’t be stopped so individual members can coordinate votes.

  • Armor/Ease does help manage open capacity and benefits from premiums when covers, large and small, are purchased.

  • In the case of Armor/Ease’s Armor stake: no request was made to open capacity on Rari Capital, so the choice to stake was solely Armor/Ease’s.

Past Claim Events

I would also note that Armor/Ease has benefited from past claim payouts where proof of loss wasn’t required to file an arNFT for a 1000 ETH payout, which was claimed by Armor and used to offset losses for the Yearn yDAI hack payouts and to add to the Armor reserves to offset future losses in the event of future payouts, like the Rari exploit.

2 Likes

Thank you for the responses! To clarify/respond to some things:

  1. I can totally see the interpretation of “possible” going both ways. I read it as, “within the mutual’s capabilities,” rather than, “within normal functionality,” and it seems other users do as well judging by the post from Discord, but I can see it meaning within normal functionality as well. I do believe, however, that if there’s a disagreement stemming from a lack of clarity in the cover wording that the responsibility should lie on the mutual; if not, potential members may be wary of coverage.

  2. Agreed that it’s 100% stakers’ responsibility to account for risk—regardless of if we’re asked to stake in a protocol. I believe there was actual loss and have no problem with the initial payout; the only problem I have in terms of fairness is the enormous extra loss from idiosyncratic protocol functionality that hurt stakers very badly and could have been avoided.

  3. I disagree that paying out the actual loss will be bad optics for Nexus as long as there’s a note on the claim saying it was paid out individually. I think there’s currently a lot of emphasis put on being generous with claims because of feedback that people are worried that a DeFi coverage protocol may not pay out, but I believe the worse optics at the moment will be not giving the same attention to stakers, as it’s likely that institutional capacity will already be much harder to come by than institutional clients.

  4. The relevance of claim 102 is just that in that case the mutual acknowledged that they shouldn’t be bound by technical limitations and the community fund was used to make up for those limitations.

  5. In regards to Past Claim Events: if the cover will be turned in one way or another, the way Armor functions is that it turns it in, purchases NXM/wNXM, and returns 100% of those to stakers immediately, thereby ensuring stakers aren’t giving away free money. That tweet is incorrect in mentioning a reserve other than anything extra stakers get that functions as sort of an implicit reserve. We’ve had other circumstances (Cream hack) where we could have submitted a claim, but didn’t think it would be fair to stakers since we didn’t have our own loss and the alternative was that stakers wouldn’t need to payout.

Many good points here. Making payouts in excess of actual loss creates moral hazard and can be gamed. The stakers in the RARI pool should bear the amount of actual loss. The rest of the payout reflects a generic risk that affects the whole V1 protocol, and the cost of the associated RARI payout over actual loss should be socialised across all NXM holders via a series of appropriately-timed purchases of wNXM from the open market, unwrapped and distributed as rewards to the RARI pool stakers who made the timely and high profile payout in excess of actual loss. Given that this problem disappears in V2, no precedents will be set, and the whole episode can be chalked up to marketing, including preserving achieved yields for distributors such as Armour.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.