Mainnet still has 910 validator records with exactly 1 ETH. They are all pending_initialized, their median age is 404 days, and none can activate unless somebody supplies the other 31 ETH. EIP-8205 exists to stop protocols from creating this awkward little balance in the first place.
EIP-8148 says the rate of validators moving to 0x02 compounding credentials has been very low. That is true if a validator means a key: on the complete UTC day of August 21, 0x02 was 17,780 of 901,667 active indices, or 1.97%. Those keys carried 14.05 million of Ethereum's 42.33 million active effective ETH, or 33.19%.
Ethereum has scheduled a fivefold expansion of the validator exit lane for Glamsterdam. At mainnet's current active stake, EIP-8061 would let 1,279 ETH exit per epoch instead of 256 ETH. Deposits do not get the same treatment.
EIP-8363 arrived in the EIPs repository yesterday with a cushion: double the base reward factor at activation, then decay it back to today's value over 18 months. That sounds like a gentle start. At Ethereum's current active effective balance, it is already a 16.0% cut to the net consensus reward curve.
The reason is simple. The proposal doubles gross duty rewards, but it also burns 58.0% of each doubled reward. A fully performing duty nets 84.0% of what the current curve pays.
canonical_beacon_block_attester_slashing stores two attesting-index arrays for every proof. Summing those arrays across mainnet gives 245,290 validator-index positions. The protocol intersected them into 555 overlap positions, covering 554 unique validators. The arrays are evidence, not a validator-slashing counter.
The consolidation queue looked busy if I counted it the lazy way. On Jul 5 UTC, canonical_beacon_state_pending_consolidation had 7,927 deduped queue-position rows. The latest canonical state snapshot had seven pending consolidations.
Same table. Different denominator.
Rocket Pool's rpl_staked event looks like it fell off a cliff at exactly the wrong time. The last decoded stake event in mainnet.int_rocketpool_node_event was 2026-02-17 23:20:23 UTC. The first decoded megapools started about 40 minutes later, on Feb 18, and the withdrawal side kept moving for months.
That does not mean "nobody staked RPL after Saturn." It means the old decoded stake-event surface is a bad post-Saturn RPL-staking ledger unless you know what changed.
canonical_beacon_state_pending_partial_withdrawal looks like a tidy queue table. It is not a validator count, and it is very much not a daily intake table. In the latest mainnet snapshot I checked, the queue had 87 positions, 81 validators, and 1,457.9 ETH pending.
The new withdrawal-request table has a rude little footgun: amount = 0 does not mean "withdraw zero ETH." It means "full exit this validator."
That matters because a naive sum over canonical_beacon_block_execution_request_withdrawal.amount quietly drops the exits. In the May 27 to June 26 mainnet window I checked, those zero-amount full-exit requests showed up as 0 ETH on the request side, while the same validators later produced 23,812 ETH of canonical withdrawals within seven days.
The last MaxEB committee post had an uncomfortable-looking number: large validators were about 25.6% of active effective stake, but only 0.65% of ordinary attestation committee positions. That sounds like the accounting forgot the stake.
It did not. The committee assignment surface counts validator indices. The reward surface remembers the balance.