Zombie Indexers are hurting the protocol, but it’s too gas expensive and there’s almost nothing to gain for someone to force-close stale allocations. Wouldn’t it make sense to give a reward to whoever closes stale allocations? Think like Maker liquidations. The cleaning job would be done by Searchers! (“Searchers” is a term used by Flashbots to describe the role of searching for and submitting liquidations, arbitrage, and other forms of MEV. “Keepers” is another good term and they have a similar role to Searchers but in the Keep3r Network.)
Q1: Where would the reward come from?
A1: It could be paid from the delinquent Indexer’s allocation stake. To initiate the conversation, we propose that 100% of the Indexer’s allocation stake be paid to the Searcher who force-closes a stale allocation.
Q2: If the force-close reward were to be paid from the Indexer’s allocation stake, wouldn’t this be a way to move GRT from a vesting contract (the delinquent account) into an account where the GRT would be liquid (the MEV reward account)? Or, If a malicious participant suspects that their vesting contract would be revoked (as may be the case here if they are not serving queries) then wouldn’t they be motivated to extract any fraction of their unvested stake using this method?
A2: It is highly unlikely that a malicious Indexer will be able to take advantage of force-close. The expiration date of allocations is publicly known as soon as an allocation is opened. Furthermore, the value of force-closing a potential future stale allocation is known as soon as the allocation is opened. The moment an allocation expires, Searchers will compete to submit force-close transactions. These transactions will result in a GRT reward to the Searcher during the same block that the force-close transaction is submitted. This reward can be atomically swapped for ETH using a decentralized exchange (DEX) like Uniswap. That ETH can be used to pay for transaction priority. In other words, this is capital-free MEV that can be performed in a single transaction. Because it’s capital-free, this will result in aggressive MEV competition, with most of the GRT being deposited into a DEX GRT-ETH pool, and most of the ETH being spent on transaction ordering priority.
Q3: Can you quantify the impact of stale allocations?
A3: We can estimate the reduction of indexing rewards that honest Indexers have experienced historically because of the existence of stale allocations. To estimate this, we would need to identify all instances where a delinquent Indexer was allocated to a subgraph deployment at the same time an honest indexer closed their allocation on that deployment. We’d then add up the indexing rewards the honest Indexers actually collected during those circumstances. We’d then subtract that number from the sum of indexing rewards that honest Indexers would have received had the delinquent Indexer not been allocated when the honest Indexer closed.