[Discussion] alETH Rebacking

ov3rkoalafied

Purpose

The purpose of this post is to provide context to the current alETH situation and to discuss potential paths forward.

Mainnet alETH vs L2 alETH

There are two distinct situations involving alETH right now that need to be remedied:

Mainnet alETH - The July 30th Curve exploit drained the entire alETH/ETH pool. This results in a direct loss to backing of alETH on mainnet. More information here: https://medium.com/chainlight/curve-finance-analysis-and-post-mortem-ba55f2b26909.
Multichain - Earlier this year, Multichain collapsed (details remain unclear). It is unknown if the alETH held by multichain is recoverable. The multichain mainnet alETH is backing for alETH bridged to Optimism. Therefore, there is a direct additional loss to the backing of alETH on Optimism. More information here: https://forum.alchemix.fi/public/d/429-aip-95a-multichain-response-optimism-and-arbitrum

Mainnet alETH - Status/Accounting

The below numbers are from on-chain analysis but have not yet been verified.

Pre-Exploit State

At the time of the exploit, the curve LP had 15,939 alETH/ETH. The AMO owned 4990 shares of the pool. The entirety of the pool was drained.

Post-Recovery State

The hacker drained a majority of funds in the pool. They returned all of these funds, representing a net recovery of 4820.5 alETH and 7259 ETH. These funds are currently held by the Alchemix Multisig.

The hacker left 3854 alETH in the pool, which was effectively able to be stolen via arbitrage. These funds have not been recovered.

AMO Surplus

Per AIP 95a, the alETH AMO had built up a surplus of 382 alETH. There was an additional positive slippage of 27 alETh from the withdrawal of the 8,000 tokens, leading to a total surplus of 409 alETH. This is excess backing that can be used to re-back alETH (Ie, this is 409 LESS alETH that would need to be recovered)

Bridger Surplus

Some arbitrageurs got stuck in the connext bridge. As per AIP-96 the protocol will reclaim ₂₃₀ alETH from the arbitrageurs. This excess alETH will effectively act as recovered funds owned on Optimism that could be bridged back to mainne and bundled into the recovered funds to further re-back mainnet alETH.

MEV Profits

A few validators, most notably Coinbase, LIDO, and rocketpool, received over 500 ETH of tips from arbitrage bots after the initial exploit. In many cases, the majority of the profits these bots earned were paid directly to Coinbase among other validators. It is unknown at this time if Coinbase will return the ETH tips that represent the profit from the stolen funds.

Curve DAO

It is possible that CurveDAO will offer some level of financial assistance to LPers. The extent of this is unknown at this time, and Curve is still focused on the CRV/ETH exploiter who has not yet returned any funds.

Summary

Asset	Supply/Quantity/TVL	Link/Source	Notes
Line Item	Amount	Units	Notes
Stolen Funds	15,939	alETH/ETH	Estimated Quantity, Ratio TBD
Stolen Funds owned by AMO	4,990	alETH/ETH	Estimated Quantity, Ratio TBD
Recovered Funds	4280.5	alETH	-
Recovered Funds	7259	alETH	-
Unrecovered Funds	3854	alETH	-
AMO Surplus	409	alETH	-
Bridger Surplus	230	alETH	owned as OP-alETH
MEV Returns	TBD	ETH	-
Curve DAO assistance	TBD	TBD	-

External Factors

Curve is currently undertaking analysis to determine the balance of all LPers at the time of the exploit. This data will be necessary to issue any form of refund to LPers.

Optimism alETH - Status/Accounting

The Optimism alETH situation is largely summarized here: https://forum.alchemix.fi/public/d/429-aip-95a-multichain-response-optimism-and-arbitrum

In short, Optimism alETH needs to be collateralized with either 882 alETH bridged from mainnet, or ETH donated to the transmuter, or a mix of both. Rather than burn these funds / transmute them, the ETH and alETH will be paired to create an AMO on velodrome. If this is done, then alETH on Optimism will be “fully-backed” relative to mainnet alETH (ie, curve situation notwithstanding). The supply of alETH on Optimism would then be able to contract via typical Alchemix mechanisms (debt repayment, transmuter), the contraction of the AMO, and users bridging back to Mainnet.

Potential Solutions

Mainnet alETH backing was lost. Not all of the funds were returned. Therefore the solution will require balancing the interests of restoring the backing (which benefits everyone with alETH including LPers), and returning funds to LPs (which benefits LPers directly), from the funds that were recovered from the exploiter and through other methods.

Note that the relationship between LPers and curve does not include Alchemix as an intermediary in that the Curve contracts were exploited, not Alchemix. In that manner, the Alchemix AMO is one of many LPers. Therefore, the baseline expectation is that absent other externalities in a plan, Alchemix does not need to use DAO resources (such as the treasury) beyond what has been recovered to make LPers whole. Alchemix’s priority with any non-recovered funds is to make alETH whole.

If all LPers receive an equal share of recovered funds, the net return is (7259+4280.5+230 aleth from bridgers) = 11769.5 alETH/ETH. The missing funds are 3854 alETH, of which the AMO owned 1206 alETH and LPers owned the rest. Subtracting the AMO surplus and a pro-rata share of the 230 alETH from bridgers, the remaining outstanding missing AMO balance is 725 alETH. The total refund would be 73.8% of what was lost. This will be a baseline for a few different calculations in the various solutions.

“Simple” Solution 1

Pro-rata to all LPers, use the treasury

If pro-rata is given to all LPers, then the treasury would still need to front 725 alETH (or ETH), as well as 882 alETH for Optimism, for a total of 1607 alETH or ~$2.6m.

In this solution, if more funds were to be recovered or gained later on that exceeded the amount the treasury spent to re-back mainnet alETH, there would need to be a follow-up distribution to LPers.

A few treasury positions could be utilized for this. Note that some assets are already being used for AIP95a, alUSD portion, and have been excluded:
Locked CVX unlocking 12/06/2023, worth $1.1m
Locked AURA unlocking 12/13/2023, worth $455k
ALCX/ETH protocol owned liquidity, ETH portion worth $270k
Velodrome alETH/ETH LP worth $1.8m (can/should only be used for Optimism portion)
Locked VELO - Worth $843k, but would likely OTC for much less

The sale of these assets would significantly dampen the growth of Alchemix especially with many DeFi tokens at/near lows relative to ETH. Many of these assets directly contribute to the revenue of alETH and circulating supply, and thus may have a positive feedback loop effect, where the loss of these assets requires alETH to shrink even more after the backing is restored. This situation is possible but is not recommended unless all other solutions have been explored and deemed unacceptable.

“Simple” Solution 2

Prioritize AMO first, LPers second

In this solution, the AMO receives the necessary funds first. This would ONLY apply to Curve exploit funds, NOT Optimism alETH.

In this situation, the AMO would receive 4990 alETH/ETH at the ratio of returned funds. This amounts to 3139 ETH and 1851 alETH. The AMO, and therefore alETH, would then be fully backed on Mainnet. Optimism alETH would be made whole through the velodrome treasury/bridging, per AIP 95a (which would go to an adjusted re-vote).

The remaining funds to LPers would be 4120 ETH and 2429.5 alETH. Additionally, the LPers would get the 409 alETH AMO excess and the 230 alETH recovered from bridgers. This would represent a total return of 7188.5 alETH/ETH. This would represent a return of ₆₅.7% to LPers instead of ₇₃.8%.

From here on out, all recovered funds would go to the LPers. Ie, any bridger or MEV recovery and/or Curve DAO assistance, if possible, would all go straight to LPers. (This assumes if CurveDAO assists they would pay in assets. If they paid in voting power or by other means, a different solution would need to be found). The DAO would need to determine the right balance of batch refunds. Additionally, the DAO could commit some treasury funds and the AMO surplus up front to decrease the gap between this scenario and the pro-rata option. Lastly, the DAO could commit to fill the gap between actual distribution and pro-rata distribution with the treasury in the future.

In this example, the 1206 alETH loss to the AMO is being transferred to LPers. Subtract the 409 alETH surplus and 230 alETH bridger surplus and the debt to LPers becomes 567 alETH. If the assumed AMO size is 8,000 alETH/ETH (based on estimated alETH to ETH ratio), at an estimated 10% APR, 1 year of AMO harvesting would be 800 alETH/ETH. So repaying the LPers could take ₉ months.

Note that this solution is only possible because the exploiter returned funds to Alchemix instead of Curve. This solution would prioritize the backing of alETH and health of the overall system first, and LPers second, thus may not be as popular of an option for LPers compared to the first option. This is an important point for the DAO to consider, as LPers are necessary for the protocol to operate efficiently.

Solution 3

Refund LPers pro-rata, slowly re-back alETH

In this solution the AMO and LPers would all get a pro-rata refund of what was recovered. alETH would then operate as partially backed, where a portion of profits earned by the AMO would go towards re-backing alETH until full backing is achieved.

In this scenario, alETH may exist at a discount relative to the historical price of alETH. However, if the DAO commits to continuing to incentivize liquidity for alETH even in the event of an unwind using the treasury, then the last alETH holders would still earn substantial rewards which could effectively make them whole.

Other Options / Complexities

One option is to simply wait to see what Curve does, as the presence or lack thereof of financial assistance from Curve DAO could guide the decisions. For example, in “Simple” Solution 2, Lpers are estimated to have to wait ₉ months to get what they are entitled to right now. However, if Curve DAO were to offer financial assistance, this 9 months could become a more reasonable figure.

When refunding LPers, it will need to be determined if LPers get alETH + ETH separately, or in the form of LP tokens.

Separate Distribution

This is easiest (provide alETH and ETH separately), as no LP math needs to happen. However, it does make it more likely that LPers will no longer LP (additionally, it creates gas costs to re-establish the positions). It’s possible new LPers fill in the void, but in the current market conditions it is difficult to attract new TVL.

LP Token Distribution

This is difficult as the recovered funds are skewed towards ETH and the balance of the pool is unknown, especially in a situation where alETH is not immediately re-backed. The DAO would be able to mint new alETH off of ETH backing to issue the correct backing if needed, but the accounting would take time to sort out. This approach is beneficial in that it is easy for LPers to remain as LPers (which is good for the protocol). This option is more worthwhile to explore in a situation where alETH is immediately rebacked. If alETH is not immediately rebacked, the pool balance is more difficult to predict and it may be less desirable for LPers to continue LPing.

Next Steps

The next steps are to determine if it is more worthwhile to await more information / potential fund recovery, to develop a more complex/robust solution, or to aim to proceed with one of the three proposed plans.

msplint

I'd really weigh the principles of the project more heavily on this. Take a shot at a great long term outcome rather than 'make amends' for what happened. Some of these details probably won't matter but a strong core structural solution that's self correcting, much like a auto-repaying loan might not fail just because it's creative.

KW710

Personally, I feel like any solution needs to prioritize the restoration of the alETH backing, so I'm inclined to take Solution 2 as our starting point and iterate from there where nuance and compromise is required.

My reasoning is as follows:

1) The backing has a not-so-subtle impact on the token parity between ETH and alETH. Acknowledging that alETH is not intended to have a hard peg except within the internal Alchemix ecosystem via the initial lending mechanic, the transmuter, and the loan repayment mechanic, approximate parity is still vital to incentivizing both borrows and LPs. Even if LPers were to be made whole by Alchemix instead of Curve, there's a strong disincentive to LP as long as the possibility of significant arbitrage in the LP pool exists. In other words, prioritizing the needs of LPers doesn't actually solve the underlying challenges facing the protocol and the DAO. It ultimately leaves us in the same position as if LPers had not been made whole and trades long term confidence in the DAO's ecosystem for fleeting, extractive good will.

2) If they have not happened already, blunt conversations with Curve need to occur about what part they will play in making LPers whole. Until that occurs, it would be irresponsible for Alchemix to preemptively move forward with any plan to do so on its own.

As correctly stated above, Alchemix is not the faulty party in reference to the losses sustained by LPers. This applies to both the protocol on a contract level as well as the DAO on a social one. No fiduciary duty was breached. There was no negligence on the part of the DAO or protocol team. The fault belongs to Curve alone as it was their contracts which were exploited. As such, and as there is no obligation to do so, any aid that Alchemix does provide is, by definition, an act of charity.

I am all for being generous and helping those in need. I absolutely think that the members of the DAO should collaborate on creative solutions to provide as much aid as possible, especially since many of the LPer recipients of said aid are also DAO members. However, charity can't jeopardize the long term viability and sustainability of the protocol.

3) Less pressing, but still a valid consideration, is the accelerating development of the US regulatory paradigm around blockchain assets in general and stablecoins in particular. The US legislature is actively discussing potential frameworks with which to regulate stablecoins. Regardless of whether it is actually close to passing meaningful legislation or not, the fact is that the topic is sufficiently top of mind to fuel debate around the subject. We can be confident that any pegged or backed asset is going to face scrutiny on multiple fronts as this legislative effort gains momentum in the US. When that occurs, how is it going to appear to opportunistic regulators looking for a punching bag that one of Alchemix's assets went under-backed for a significant period of time? In my view, it would be far more advantageous to the project in the long term to demonstrate both to the market and to future regulatory inquisitors just how quickly the backing was restored.

And before anyone chimes in that US regulations don't apply in Europe and Asia and that the US isn't the center of the universe, let me beat you to it. The US is not the center of the universe.

HOWEVER, as long as significant portions of the project's human capital are US residents and/or US citizens, then the regulatory impact of the DAO's decisions remains very relevant. There's no stopping various agencies from attempting bad faith actions in the court system as we're all aware (unless you've been living under a rock the past two years). But let's not make it easier for them. Let's not do their job for them.

LogrisTheBard

Speaking as an LP, if I have a choice between walking away whole today and leaving a burning wreck behind me vs walking away whole in a year or two with a healthy Alchemix I'll take the latter. Koala's post is focused on prioritizing who wins and who loses. In terms of priorities for the Alchemix protocol, we need to prioritize the necessary actors of the system first. To me that means vault users first, followed by LPs, and prioritizing ALCX dividends last. This prioritizes not only doing the right thing which maintains reputation but also maximizing growth over profits today.

Defi has seen hacks before. I'll just cite my own article here: https://tokenomicsexplained.com/when-things-go-wrong

Reading it again now, it has aged well and it offers expanded options on the solution summarized above. Rather than just saying restore the AMO first and LPs last, we can expand on how to repay LPs eventually and what to do with the funds in the meantime.

One idea from my post is to issue a debt token like GRAIN. We don't even need liquidity depth on it, just optionality on what each LP wants to do with it. LPs can exit right away with the ₃₅% loss or they can leave the money with Alchemix and the AMO will manage the funds. LPs who stick around can receive extra rewards from the DAO. It doesn't need to siphon off all revenue, it just needs to make the APR higher than comparative opportunities in Defi. This could be done with ALCX issuance, AMO profit, the harvest skim, treasury asset yield, etc. Eventually we can make everyone whole using DAO profit as an income stream without any major disruption. For this cost the DAO maintains reputation and it gets some sticky hostage liquidity (at inflated cost).

Other wild ideas include creating a Convex/Concentrator/Curve LP ALchemix vault and having the DAO take an alETH loan to pay out LPs with a loan on the AMO funds. It would be beautiful in a way if we could solve this problem using a self-repaying loan backed by the revenue of the protocol.

Either way, I trust the devs are doggedly engaged with the Curve DAO. To be very clear, Alchemix did nothing wrong here, and doesn't technically owe us LPs anything but the recovered funds back. The reason we're having this discussion is because, as usual, Alchemix wants to do the moral thing. If Curve shares that sentiment at all hopefully that DAO will find a way to come up with the shortfall funds and cleanly resolve the matter.

KW710

LogrisTheBard Couldn't agree more.

ov3rkoalafied

LogrisTheBard re: "vault users first, followed by LPs, and prioritizing ALCX dividends last"

Vault users benefitted from this exploit technically, so they are fine. Agreed that ALCX holders are last in this hierarchy. So that would put LPers top priority. Ultimately they are who care the most about the value of alETH.

So do LPers want their share of recovered funds, where the alETH they receive is partially unbacked?
Or do they want less then their share with an IOU, where the alETH they receive is fully backed?

So far I've seen about 50/50, in which case I don't think we can really take from LPers that want the assets they are entitled to. Especially when the protocol does in theory have other means of making alETH whole.

jamiewoo

IMO,Rebacking $ALETH is always the top priority.
if average refund all LPers and AMO pro-rata,the $ALETH market price is almost 0.75eth(worst mood-case)，and LPers get the remaining returned aleth amount is 73.8%,so LPers get actual fair value $Aleth is 0.75*73.8%=0.5535eth in Solution 3.

If we priority refund AMO first, LPers second.the $ALETH market price is almost 0.99eth when AMO is full backing(at almost all times in history),and Lpers get the remaining returned aleth amount is 65.7%,so LPers get actual fair value $Aleth is 0.99*65.7%=0.65eth in Solution 2.

So the conclusion is Whether we prioritize LPer’s benefit or not,The premise of all optimal solution is rebacking $ALETH market price first.that means we have to make AMO full backing again first in all situation.

ov3rkoalafied

Documentation: The primary MEV bot that acted after the exploiter has returned the 43.3 ETH profit they earned: https://etherscan.io/tx/0xfe4b8d479405499710999f2c2140e48b09e5f7546dbfb2a2787859c63af2a8a1

LogrisTheBard

So do LPers want their share of recovered funds, where the alETH they receive is partially unbacked?
Or do they want less then their share with an IOU, where the alETH they receive is fully backed?

I don't think the DAO can ethically withhold the funds it has in any sense from an LP that wants them. To me this makes Solution 2 unviable even if I'd be fine with it for my funds. My starting sentence was me personally disfavoring Solution 1, it would be ethical enough but I don't want to burn the DAO down behind me.

My suggestion of an IOU token gives each LP a choice. If they claim now, they can take their assets and walk away (however unbacked they are). If they leave it there, after some time the funds will go to the AMO to restore the peg as much as is possible and start earning fees. In exchange for staying there we'd direct some form of additional revenue to those who stay.

To be clear, I'm suggesting an extension of solution 3. Wrap the funds returned in an IOU token. Give it a few weeks for LPs to withdraw their pro-rata share if they want to. After those weeks the funds will go to the AMO and begin compounding the AMO revenue + some stream of extra revenue so that the APR is a few percent higher (e.g. carve out some ALCX issuance). This extra revenue option is only available to affected LPs and only until they redeem.

Basically the DAO is buying the liquidity at a slightly higher rate out of charity until the LPs are made whole either by Curve, Convex, Coinbase, returned hacked funds, or as a last resort a stream of income from Alchemix that isn't destructive to the DAO.

ov3rkoalafied

LogrisTheBard I think a simpler form of this may just be to say that any unclaimed LP tokens just get staked and revenue is used to reback alETh.

If the refunds are offerred pro-rata, then I don't think the DAO has any obligation to pay extra for people that "remain in". Ie, I don't see why the DAO should pay extra to LPers because of waiting for Curve, Convex, Coinbase, etc. The DAO is one of those LPers, it is also waiting.