Type-4 transactions are already mostly two gas stories

Type-4 transactions are easy to count and easy to overread. From Jun 20 through Jul 3 UTC, Ethereum had 257,388 deduped type-4 transactions, which was only 0.824% of all transactions in the same blocks. That sounds like a small EIP-7702 adoption number until you look at the gas: one address labelled Poisoner and a pile of plain ERC-20 transfer(address,uint256) calls consumed 86.5% of type-4 gas.

That is the trap. transaction_type = 4 tells you the envelope is EIP-7702's set-code transaction type. It does not, by itself, tell you this is a clean smart-wallet action, a user operation, or even something interesting at the authorization-list layer.

EIP-7702 defines SET_CODE_TX_TYPE = 0x04. The outer transaction still has a destination, value, data, access list, and an authorization_list. Xatu's canonical_execution_transaction table exposes the transaction type and outer call fields, but not the individual authorization-list entries, so I kept the claim deliberately narrow: this is a top-level type-4 envelope and calldata story.

The first check was just the type distribution. I resolved the canonical execution block range for Jun 20 through Jul 3, then deduped raw transaction rows by transaction_hash before grouping. That dedupe matters in this window: raw canonical_execution_transaction had repeated rows, while the deduped total matched mainnet.fct_execution_transactions_daily FINAL exactly at 31,246,133 transactions.

-- clickhouse-raw, after resolving Jun 20-Jul 3 UTC to blocks 25355072-25455433
WITH tx AS (
  SELECT DISTINCT
    transaction_hash,
    transaction_type,
    success,
    gas_used,
    from_address,
    to_address
  FROM default.canonical_execution_transaction
  WHERE meta_network_name = 'mainnet'
    AND block_number BETWEEN 25355072 AND 25455433
)
SELECT
  transaction_type,
  count() AS txs,
  countIf(success = 0) AS reverts,
  round(100 * count() / sum(count()) OVER (), 3) AS pct_of_all_txs,
  round(100 * countIf(success = 0) / count(), 3) AS revert_pct,
  uniqExact(from_address) AS senders,
  uniqExact(to_address) AS targets,
  sum(gas_used) AS gas_used
FROM tx
GROUP BY transaction_type
ORDER BY txs DESC;

Type-4 was small by count and weird by shape:

transaction type	deduped txs	share	revert rate	gas used
2	27,200,967	87.054%	1.525%	2,593.4B
0	3,623,052	11.595%	1.363%	208.4B
4	257,388	0.824%	5.974%	223.1B
3	141,686	0.453%	0.006%	15.3B
1	23,040	0.074%	6.862%	3.3B

Then I reused the deduped type-4 rows and bucketed them by the outer call target and selector. This is not an ABI-perfect taxonomy. It is intentionally blunt: the address-label table says one top target is Poisoner, 0xa9059cbb is the ERC-20 transfer(address,uint256) selector, and 0x34fcd5be is executeBatch((address,uint256,bytes)[]) in the signature table.

WITH tx AS (
  SELECT DISTINCT
    transaction_hash,
    success,
    gas_used,
    lower(ifNull(to_address, '')) AS to_address,
    ifNull(substring(input, 1, 10), '') AS selector
  FROM default.canonical_execution_transaction
  WHERE meta_network_name = 'mainnet'
    AND transaction_type = 4
    AND block_number BETWEEN 25355072 AND 25455433
)
SELECT
  if(
    to_address = '0x00fe78205f5f0e63b8ad2b2ae5337f538a610e04', 'Poisoner target',
    if(selector = '0xa9059cbb', 'ERC-20 transfer selector',
      if(selector = '0x34fcd5be', 'other executeBatch selector', 'other type-4')
    )
  ) AS bucket,
  count() AS txs,
  sum(gas_used) AS gas_used,
  countIf(success = 0) AS reverts
FROM tx
GROUP BY bucket
ORDER BY gas_used DESC;

That split is where the mental model breaks:

bucket	txs	tx share	gas used	gas share	reverts
labelled Poisoner target	44,084	17.13%	107.6B	48.24%	0
ERC-20 transfer selector	43,446	16.88%	85.4B	38.27%	41
other executeBatch selector	7,005	2.72%	1.3B	0.59%	97
other type-4	162,853	63.27%	28.8B	12.91%	15,239

The labelled Poisoner row is the ugly one. The target 0x00fe...0e04 received 44,084 deduped type-4 transactions from one sender, reverted zero times, and used 107.6B gas. By transaction count it was a large minority. By gas it was almost half the type-4 surface.

The ERC-20 transfer row is the part I did not expect to be that loud. A type-4 envelope can wrap ordinary-looking calldata, and in this window 43,446 type-4 transactions had the plain transfer(address,uint256) selector. Those calls used 85.4B gas, another 38.3% of type-4 gas. A few were normal labelled tokens like USDT, USDC, WBTC, and USDS, but most of the high-count transfer targets in this slice were unlabeled contracts with one dominant sender. I would not turn that into token-adoption prose without a separate token/value check.

There is a smaller table lesson hiding under the Ethereum one. If I had counted raw transaction rows without deduping, type-4 would have looked like 357,914 rows instead of 257,388 transactions in this block range. That is not a subtle rounding error. It is another reason to reduce to the semantic key before writing anything public.

The safer sentence is boring but honest: type-4 transactions were under one percent of Ethereum transactions in these fourteen days, and the gas was dominated by a labelled Poisoner target plus ERC-20 transfer-shaped calls. That is not a wallet-adoption metric. It is a set-code-envelope metric, and it needs to be split by outer call shape before it says much about how people are using EIP-7702.