dataforge-labs/ethereum-attestation-pool

Ethereum attestation pool

Attestations seen waiting, against attestations that reached a block.

On chain you can see that a validator's attestation is missing. You cannot see whether it was never produced, produced but never propagated, or propagated and never included. That difference is only visible while the attestation is still pending.

Contents

name	one row is
e18_attestation_pool	one slot: attester slots seen waiting, attester slots included, and the difference

What this is not

Attester effectiveness is well served elsewhere. Participation rate, correctness and inclusion delay are all derivable from chain data and several public tools publish them. None of that is here, because none of it is scarce.

Reading it

attesters_seen_in_pool counts attester slots observed waiting for that slot. attesters_included counts those that reached a block. attesters_net_never_included is the first minus the second.

The difference is signed, and negative values are ordinary: a proposer can include attestations this vantage point never held, so a negative number means the network carried more than was seen here. Clamping it at zero would hide the asymmetry the column exists to show.

Read the median rather than the total. Across 1,319 slots with a fully observed inclusion window the aggregate gap is -2.0%, but that is pulled by a tail. The median slot is -0.07%, p10 is -2.77% and p90 is -0.01%. A typical slot agrees with the chain almost exactly and a minority accounts for nearly all of the difference.

No slot has yet shown more attesters here than on chain. If you see positive values in quantity, suspect the collection before the network.

Before you build on this

• One vantage point. This is what a single consensus client held, not what the network held. An attestation absent here may have been present elsewhere. Treat it as a lower bound.
• It counts attester slots, not validators. A bitfield population tells you how many attester slots it represents, not which validators they were. This answers how many were left out, not whether a particular validator was.
• Attestations remain visible for roughly ten minutes. One that arrived and was included between two observations is invisible here. n_polls_seen is on every row; low values mean low confidence.
• window_closed marks slots whose inclusion window was fully observed. Rows without it have incomplete inclusion counts by construction. Filter on it before drawing conclusions.
• Slots near the end of a collection window are finalised early and will show window_closed as false.

Partitions are parquet, one file per collection window, under dataset/YYYY/MM/. Every dataset here carries a FIXED 7-day sample WINDOW starting at its own first day of collection, together spanning 2026-08-28 to 2026-09-03, so you can check schema, coverage and quality before asking for more. It does not advance, so there is nothing to gain by re-downloading it. The full history is held privately, available on request.

from huggingface_hub import snapshot_download
import pandas as pd, glob

path = snapshot_download("dataforge-labs/ethereum-attestation-pool", repo_type="dataset",
                         allow_patterns="e18_attestation_pool/**")
df = pd.concat(map(pd.read_parquet,
                   glob.glob(f"{path}/e18_attestation_pool/**/*.parquet", recursive=True)))

Coverage

e0_run_manifest lists every collection window with its poll counts and failure counts, and is published in full rather than windowed. Gaps between windows are real, cannot be filled in afterwards, and nothing here is interpolated.

License and contact

ODC-BY: use it freely, credit "DataForge (dataforge-labs)". Questions and requests for the full history via the discussions tab.