Add attestation correlated penalties spec #3780
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
igorline
added a commit
to igorline/lighthouse
that referenced
this pull request
May 26, 2024
igorline
added a commit
to igorline/lighthouse
that referenced
this pull request
May 26, 2024
ensi321
reviewed
Jun 4, 2024
| pending_balance_deposits: List[PendingBalanceDeposit, PENDING_BALANCE_DEPOSITS_LIMIT] | ||
| pending_partial_withdrawals: List[PendingPartialWithdrawal, PENDING_PARTIAL_WITHDRAWALS_LIMIT] | ||
| pending_consolidations: List[PendingConsolidation, PENDING_CONSOLIDATIONS_LIMIT] | ||
| net_excess_penalties: Vector[uint64, TIMELY_HEAD_FLAG_INDEX] # [New in correlated_penalties] |
There was a problem hiding this comment.
A little finicky. Wonder if we can set the length to be dynamic len(PARTICIPATION_FLAG_WEIGHTS) instead. There is no precedence on this though
| def committee_slot_of_validator(state: BeaconState, index: ValidatorIndex, epoch: Epoch) -> Slot: | ||
| committees_per_slot = get_committee_count_per_slot(state, epoch) | ||
| for slot in range(epoch * SLOTS_PER_EPOCH, (epoch + 1) * SLOTS_PER_EPOCH): | ||
| for committee_index in range(committees_per_slot): |
There was a problem hiding this comment.
if validator_index in get_slot_committees(state, slot):
return slot
Will be cleaner this way without the extra for loop
| elif flag_index != TIMELY_HEAD_FLAG_INDEX: | ||
| # [New in correlated_penalties] | ||
| slot = committee_slot_of_validator(state, index, previous_epoch) | ||
| penalty_factor = compute_penalty_factor(state, slot, flag_index) |
There was a problem hiding this comment.
penalty_factor, _ = compute_penalty_factor(state, slot, flag_index)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Implements correlated penalties proposal from https://ethresear.ch/t/a-concrete-proposal-for-correlated-attester-penalties/19341 with multi-dimensional net_excess_penalties
Quantitative analysis with past mainnet data show that known large operators see their cumulative penalties increased. This proposal is meant to either incentivize operational distribution or reduce the profit margins of scaled operations.
This notebook includes some charts to visualize the behavior of this gadget https://colab.research.google.com/drive/1mOsM-cEibQAqQBtlIdlpS6h5sahOzXEw?usp=sharing
Participation penalties of offline validators are now scaled by an extra
penalty_factor. As participation drops (exaggerated in this chart for reference)penalty_factortemporarily increases to > 1.net_excess_penaltiesis a moving average that asymptotically tends to some value untilpenalty_factor = 1.I would like to consider splitting the participation bits into correctness and timeliness. Otherwise a proposer playing timing games can cause a fraction of attesters to suffer increased penalties without any fault of their own. We need more research to understand what behaviors can be attributed to individual setup operations and which can be induced by other network participants.
Joint effort from of EthBerlin04 with @nerolation and @igorline