---
name: operator-graph-2
description: Operator graph part 2 (from 4 Sep 2026): method, infrastructure and further links. (frontmatter rebuilt during handover copy)
sources: [cowork]
---
# Operator graph, part 2 (from 4 Sep 2026)

Method, infrastructure addresses, the degree filter and the runs of 31 Aug to 3 Sep 2026 are in [[operator-graph]]. Standing Dune query id 8562361.

---

## Run of 4 Sep 2026, v4.fun factory, 3 day window. **THE PAD IS NOT DEAD. THE 3 SEP CALL WAS WRONG.**

15 first-funding rows (down from 20, the window rolled). **Two brand-new deployers, both launching in the last 24 hours.**

### CORRECTION TO THE 3 SEP RUN, AND IT IS THE HEADLINE
- [stated] 3 Sep recorded `v4fun_24h = 0`, "v4.fun has had NO launches for 29 hours", and concluded "that axis is now dead and the graph will keep returning nothing new forever". **That conclusion is overturned today.**
- **`v4fun_24h` = 2.** Launches resumed after a roughly 41 hour gap: last pre-gap launch 2026-09-01 20:20:41 by `0x040db6261db1961a47014c9e6dd40ca48863a70e`, then **2026-09-03 13:16:23 by `0x39aec46f5ec95169dae94c0f2d3c9f6c97893488`** and **2026-09-03 20:47:12 by `0x1fb3bfabf44e511546499b7a581a4b4684adbd76`**
- **The number that misled: a 24 hour zero on a pad that averages one to eleven launches a day is one quiet day, not a death.** A pad's death needs a multi-day zero, not a single-window zero. **Do not retire the v4.fun launch set.** The 3 Sep action item still stands on its own merits, but as an ADDITION not a replacement: add PonsV2LaunchFactory `0x7eD598BcEf8bd9Edd8C97A195C6d13f40801EC7e` as a second launch set, because Pons V2 took $6,085,253 of fees in 24h against v4.fun's two launches
- freshness re-checked as required: `max(block_time)` on `robinhood.transactions` = **2026-09-03 22:40:23 UTC** with **10,637,486 transactions in 24h**. Chain live, Dune about 2.6 hours behind wall clock

### NEW DEPLOYERS, first seen 4 Sep 2026
| Deployer | Funder | Launch (UTC) | Relay direct (ETH) | Relay one-hop (ETH) | Funder status |
|---|---|---|---|---|---|
| `0x39aec46f5ec95169dae94c0f2d3c9f6c97893488` | `0xada5bb90d0de0bd1b6f3938708f49295a8d1f7cb` | 2026-09-03 13:16:23 | 0.057530784573215 | 0 | **infrastructure, killed today, no attribution** |
| `0x1fb3bfabf44e511546499b7a581a4b4684adbd76` | `0xe7e5702786897f32a5c0012e2d340ade5d0b1b2a` | 2026-09-03 20:47:12 | 0 | 0 | single, no cluster |

**NO REPEAT OPERATOR ON EITHER NEW LAUNCH.** Neither deployer was in the table, neither funder survives the filter as an operator. Both launches are unattributed at hour zero.

### DEGREE TESTS RUN 4 SEP 2026
| Address | Sent to | First-funded | Verdict |
|---|---|---|---|
| `0xada5bb90d0de0bd1b6f3938708f49295a8d1f7cb` | **90,726** | **43,148** | infrastructure, killed 4 Sep |
| `0xe7e5702786897f32a5c0012e2d340ade5d0b1b2a` | 2 | 1 | single, no cluster |
| `0x943f9ff0c22149f6e4732b7a5a7205b66f37e52b` | 7 | **0** | not a funder of anyone. Not a cluster head. Owed since 3 Sep, now closed |
| `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c` | 21 | **9** | **OPERATOR. Owed since 2 Sep, deferred twice, now closed** |

### THE FINDING: `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c` IS A CONFIRMED OPERATOR
- This is the wallet the DEBT off-chain covenant paid **0.4 ETH** to on 2 Sep, recorded in [[operator-graph]] as "EOA, 12.43 ETH, active sniper via `0x4A86009A36FceC5Aa341ffCEB3205a911FcF6f60`, degree test still to run"
- **It passes the degree filter at 21 sent to / 9 first-funded.** Nine first-funded addresses is a cluster, not a hot wallet
- **It is therefore both a payee of a live operator (`0xc7926ca2Ba6aF6FcdF3a627687a9363885c9048F`, the DEBT Obligor) and a funder of nine wallets of its own.** That is the same shape as cluster A, where `0xba545fbc19b5d141a0166836b47d8432d844a7ae` was an 8.765 ETH extractor bankrolling three children
- **ACTION: treat any token whose deployer or creator-fee recipient traces to `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c` or to `0xc7926ca2Ba6aF6FcdF3a627687a9363885c9048F` as pre-flagged at hour zero.** The nine children of `0xaf924bbc...` are not yet enumerated. **Enumerate them next run; that list is the highest-value unrun query in this file**

### SECOND-ORDER USE, THE THIRD INSTANCE
`0xada5bb90d0de0bd1b6f3938708f49295a8d1f7cb` was the FIRST name on the ARC screen's top-15 buyer list, called "a spray bot, early into 231 tokens". At 90,726 out-degree it is a router. **ARC's spray-bot count drops from 10 of 15 to 9 of 15.** This is the third wallet the degree filter has pulled out of a buyer list (`0xa67d7eb4...` 31 Aug, `0xf70da978...` 1 Sep, this one today) and the second from ARC's list specifically. **Run the degree test on the top 5 buyers of every screen AT SCREEN TIME.** Correction also written into [[screen-log-outcomes-2]].

### THE THREE CLUSTERS, unchanged
A: `0xba545fbc19b5d141a0166836b47d8432d844a7ae` + `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f` · B: `0x2f075cbdce9c2e776bb720f7d5300e9c26915983` + `0xbc2aaca7cedfc9c07d66d19444f4f87282a8dd22` · C: `0x18c2ff18dcec0d3ca5d338ed68bab88a3dbebc5c` + `0x758d3b0c79a032cfd83c4315d071d333bebf203e`. `0xba545fbc19b5d141a0166836b47d8432d844a7ae` appears again this run in both columns, deployer and funder of `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f`. None of the three clusters launched in the last 24 hours.

### Launch counts in the 3 day window
`0x31511477f9d0a45a65f7e0a211749573fb5e7b7f` 11 (last 1 Sep 19:02:55) · `0x943f9ff0c22149f6e4732b7a5a7205b66f37e52b` 7 (last 1 Sep 19:34:50) · `0x758d3b0c79a032cfd83c4315d071d333bebf203e` 2 · `0xba545fbc19b5d141a0166836b47d8432d844a7ae` 2 · eleven wallets at 1 each, of which only `0x39aec46f5ec95169dae94c0f2d3c9f6c97893488` and `0x1fb3bfabf44e511546499b7a581a4b4684adbd76` are inside 24 hours.

### Extraction re-read 4 Sep, direct AND one-hop
- `0xba545fbc19b5d141a0166836b47d8432d844a7ae` **8.11388354270176 direct**, unchanged for a third run
- `0xcdbb76d1d731b268c090f18603c888b3f75c8a58` 1.9099 · `0xb451b975731bc04302b06e68ddace5e04b947b89` 1.6 · `0x943f9ff0c22149f6e4732b7a5a7205b66f37e52b` 0.59 · `0x39aec46f5ec95169dae94c0f2d3c9f6c97893488` 0.057530784573215 · `0x040db6261db1961a47014c9e6dd40ca48863a70e` 0.0465706088783805 · `0x0bb3af56aca11237a5e7e8cf1c2ed395b0095721` 0.016049101884779798, all unchanged
- **one-hop relay measured on the four wallets that mattered and it is ZERO for all four**: `0x1fb3bfabf44e511546499b7a581a4b4684adbd76`, `0x39aec46f5ec95169dae94c0f2d3c9f6c97893488`, `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f`, `0x943f9ff0c22149f6e4732b7a5a7205b66f37e52b`
- **`0x943f9ff0c22149f6e4732b7a5a7205b66f37e52b`: 7 launches, 0.59 ETH direct, 0 one-hop, 0 addresses first-funded. Fully characterised now and it is a lone launcher, not a cluster head.** The 3 Sep note called it "the second-most active launcher on the pad AND a live extractor" and flagged it for a degree test. Closed: extractor yes, operator network no
- **`0x31511477f9d0a45a65f7e0a211749573fb5e7b7f`: 11 launches, 0 direct, 0 one-hop, still nothing extracted after five days.** Either every launch failed to raise, or the extraction is queued behind a claim it has not made. **Resolve which by reading its claimCreator history on the v4.fun fee router `0xCa3ea6d93aE7Ec71edC74AbaaCCd8bA736373045` next run**

### WHAT CHANGED IN THE METHOD, 4 Sep 2026
1. **A one-day zero is not a dead pad.** The 3 Sep run drew a structural conclusion from a single 24 hour window and it inverted in one day. **Require a zero across at least three consecutive daily checks before declaring any launch set dead**, and keep querying the set in the meantime; it costs nothing.
2. **The degree test, the launch-count query and the one-hop relay column all fit in ONE execution** using a `union all` with a `kind` column and `double`-cast value columns. Two executions covered the whole run at 100 to 130 seconds each, against three on 2 and 3 Sep. Keep this shape.
3. **A degree test owed from a prior run is worth more than a fresh one.** Both deferred tests closed today, and the one deferred twice (`0xaf924bbc...`) is the only operator the run produced. **Clear the owed list first, before the new-funder set.**
4. Still owed into the next run: enumerate the nine children of `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c`; add PonsV2LaunchFactory `0x7eD598BcEf8bd9Edd8C97A195C6d13f40801EC7e` as a second launch set; degree-test `0x88766f7d78004dd5b2556b6d88bab685859dde5b` (funder of `0xb451b975731bc04302b06e68ddace5e04b947b89`, untested since 1 Sep).

---

## Run of 5 Sep 2026, v4.fun factory, 3 day window. **THE OWED QUERIES ARE CLOSED AND THE LAUNCH FLOW IS MEASURED FOR THE FIRST TIME.**

**2 first-funding rows only, down from 15. Both deployers are already in this table (added 4 Sep), both funders already degree-tested. ZERO new deployers, ZERO new funders, ZERO repeat operators.**

| Deployer | Funder | Relay direct (ETH) | Status |
|---|---|---|---|
| `0x39aec46f5ec95169dae94c0f2d3c9f6c97893488` | `0xada5bb90d0de0bd1b6f3938708f49295a8d1f7cb` | 0.057530784573215 | both known, funder is infrastructure (90,726 / 43,148), no attribution |
| `0x1fb3bfabf44e511546499b7a581a4b4684adbd76` | `0xe7e5702786897f32a5c0012e2d340ade5d0b1b2a` | 0 | both known, funder is a single (2 / 1), no cluster |

The standing intersect (new funder set against the existing deployer column) returns nothing this run.

### v4.fun: zero launches again, and this is the SECOND zero in three days
- **`v4fun_24h` = 0.** `v4fun_3d` = 2, and both of those are the 3 Sep launches already recorded. **No v4.fun launch since 2026-09-03 20:47:12 UTC**, about 29 hours at the time of this run
- freshness checked as required: `max(block_time)` on `robinhood.transactions` = **2026-09-04 19:44:51 UTC** with **10,605,022 transactions in 24h**. Chain live, Dune about 5.5 hours behind wall clock
- **The 4 Sep rule holds and is being followed: three consecutive daily zeros are required before declaring the set dead. The sequence so far is 3 Sep zero, 4 Sep two launches, 5 Sep zero. NOT consecutive. Do not retire the v4.fun set**

### PONS V2 LAUNCH SET ADDED, owed since 3 Sep. **This is where the flow is.**
- PonsV2LaunchFactory `0x7eD598BcEf8bd9Edd8C97A195C6d13f40801EC7e`: **7,529 successful transactions in 24h, 29,423 in the 3 day window**, against v4.fun's 0 and 2
- **METHOD CAVEAT, and it must be fixed before these numbers are quoted as launches.** The v4.fun launch set uses `"to" = <factory> and success`, which on that factory is a launch. On PonsV2LaunchFactory the same predicate counts EVERY interaction, buys and sells included, so 7,529 is transactions, not launches. **Filter by method selector (`bytearray_substring(data, 1, 4)`) next run and record which selector is the launch.** Until then treat the creator counts below as interaction counts
- top interacting addresses in 24h, none of which appear anywhere in this table or in [[operator-graph]]: `0x5be0405dc84593fddbeccb80abb9d8cb0df75519` 195 · `0xa60e892ab5fbf4754e3052b7643e48a1bd6b3065` 193 · `0x7b444d22f099fd238210161791de26d16c3cedf2` 183 · `0x49bbf2b70955fb3a106e084d4bfda92d334573d2` 159 · `0x6704043f2c37e112c65a06c1794e52c31d8dc6a8` 152 · `0xf38ce55fe4e517e77e1056a54e64f22d43d88c3a` 124 · `0x7de5b9c86d2b47607a2962043bb165f7befeb06b` 124 · `0x38e5353393c82a1e22ff0c2ea32f978e2b74cb74` 90 · `0x4c27c9f67169e90fd03f9baa45c29acb888c052d` 88 · `0xeeefff8ce2710fa490e0fcb794235e873c252d2e` 72 · `0x6baaaf2082816d59c712b7061ec2209f337cecba` 71 · `0x51e005b49b86c4becc67ba88cfe507921741536e` 69 · `0x61e697835cb1bc6f0d3e7fccb83e6b819c066f1e` 67 · `0x6cf0b007326dbbadbe6fe35b05a533118b0b88a9` 65 · `0x9938ee2c028dddb73c45b3fc249a46b00f54a896` 59 · `0x29e9bc56d4d48c383177690b8bebc5cdd8bd298a` 54 · `0x64f3f1691101cd41b9908fc10a6c7a292d5defd3` 50 · `0x196da7b62b807e76a00496ff0a4e7e3c30dde4cb` 43 · `0x20e9be8b1b0419c9298feb775a97b90d92980560` 42 · `0x63084294c4d9cc40f8f1542a8a9f9ab8f8246fa1` 41 · `0x8e6528dd780bcf5448a7714d5f545576398ba882` 38 · `0x3e2037f693b1a8463a11da80790b5d775a5cb145` 38 · `0xca227657b0ffc755cccf9f063cc430d1d663ae8c` 37 · `0xde1ed485022549d87aec85fe08107682322f1dab` 37 · `0x767bdf5e00f84b4868ab4342366c561fc049c0db` 37
- **None of these is a repeat operator against the existing table. Once the selector filter separates launches from trades, degree-test the top five that survive**

### THE CHILDREN OF `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c` ARE ENUMERATED. Owed since 4 Sep, closed.
The 4 Sep degree test read 21 sent to / 9 first-funded. **The enumeration returns TEN first-funded children, so the cluster has grown by one since that test.**

| Child | v4.fun launches (lifetime) | Relay direct (ETH) |
|---|---|---|
| `0xa32103f8a3139c24c07fcc0875664e7df178c4f5` | 0 | 0.1237888453964608 |
| `0x25dda99d080625e47fd347d857b3bcabaed257da` | 0 | 0.1190378963007188 |
| `0x441f1efb06603cc2e864f73e0747f8f9d8d18876` | 0 | 0.09866473417483 |
| `0x163c00c479cdb5acf5ab57c11afb4cd4b42a847b` | 0 | 0.08708925571609 |
| `0x901ec318d53d9a3b84edec9cbefcf998995e5e56` | 0 | 0.0799986836007008 |
| `0x4e53d9354510f5ff9e286d1fa6271652588493fb` | 0 | 0.07717209163598 |
| `0xbd00876ec4abf12db06afd1a50b17527a6577229` | 0 | 0.0299822596417056 |
| `0xe4c2bfbf0323bccbfc728fe9d7797b1e1e9285d8` | 0 | 0.021453000093648 |
| `0xf6da779a479ae2384e1a5a27cc1c236d1968a440` | 0 | 0 |
| `0xd5b84a358083f9122b1fe8726f56c56b95b00ccb` | 0 | 0 |

- **Not one of the ten has ever launched on v4.fun. Eight of ten have bridged to RelayDepository, total 0.6372 ETH.** This is a pure exit-wallet fan-out, the same shape as cluster A's two non-launching children, not a production cluster like `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f`
- **THE READING THAT MATTERS: `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c` is a cash-out network, not a launch network.** It was paid 0.4 ETH by the DEBT Obligor `0xc7926ca2Ba6aF6FcdF3a627687a9363885c9048F` on 2 Sep and it splits funds across ten fresh wallets that bridge out in small amounts. The hour-zero pre-flag on any token whose deployer or fee recipient traces to it **stands and is strengthened**: this wallet's function is laundering the exit, so a project connected to it is a project whose operator already has an exit pipeline built
- **The ten children have zero launches TODAY, but that is the number to re-check each run.** The moment one of them appears in a launch set, this cluster converts from an exit network to an operator network

### The three clusters, unchanged for a fourth run
A: `0xba545fbc19b5d141a0166836b47d8432d844a7ae` + `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f` · B: `0x2f075cbdce9c2e776bb720f7d5300e9c26915983` + `0xbc2aaca7cedfc9c07d66d19444f4f87282a8dd22` · C: `0x18c2ff18dcec0d3ca5d338ed68bab88a3dbebc5c` + `0x758d3b0c79a032cfd83c4315d071d333bebf203e`. None launched in the last 24 hours; none launched since 1 Sep.

### WHAT CHANGED IN THE METHOD, 5 Sep 2026
1. **The whole run fit in TWO executions, 48 and 44 seconds each, using the `union all` with a `kind` column and `double`-cast value columns.** Freshness, both launch sets, the PonsV2 creator board and the ten-child enumeration came back in one query. Keep this shape; it is now the standard.
2. **A collapsing row count is itself the finding.** The funding query went 24 rows (2 Sep) to 20 to 15 to **2** as the 3-day window rolled past the last active day. **Read the row count as an activity series, not as a query failure.**
3. **Adding a second launch set immediately exposed that the launch predicate does not transfer between factories.** `"to" = <factory>` is a launch on v4.fun and an interaction on PonsV2. **Never reuse a launch predicate across launchpads without checking the selector.**
4. Still owed into the next run: the PonsV2 launch-selector filter; degree tests on the top five real PonsV2 launchers once that filter exists; degree-test `0x88766f7d78004dd5b2556b6d88bab685859dde5b` (funder of `0xb451b975731bc04302b06e68ddace5e04b947b89`, untested since 1 Sep); the `claimCreator` history of `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f` on the v4.fun fee router `0xCa3ea6d93aE7Ec71edC74AbaaCCd8bA736373045` (11 launches, still 0 extracted, owed since 4 Sep); and a `robinhood.logs` query that would close the SCL, DICE and holder-count legs in [[screen-log-outcomes-3]] at once.

---

## Run of 6 Sep 2026. **DID NOT RUN. DUNE IS OUT OF QUOTA AT THE ACCOUNT LEVEL.**

- [stated] PATCH of standing query 8562361 succeeded (the API accepts SQL writes regardless of quota). Execution `01M1T48VRDC7MY01JZZFN9HMDS` submitted 01:10:37 UTC at `performance:"medium"`, ran 37 seconds, returned **`FAILED_TYPE_RESOURCES_CAP_REACHED`** — "Query execution has exceeded the user defined maximum amount of resources" — after consuming 1.682517304 credits
- The SQL was then narrowed and resubmitted twice: funding and relay scans bounded to `now() - interval '90' day` (all known first-funding dates are 4 Jul 2026 or later, so the semantics are preserved), with the freshness check and the v4.fun 24h launch counter folded in as a `union all` per the 5 Sep standard shape. **Both attempts were refused before execution: `This api request would exceed your configured datapoint limit per billing cycle. Please visit your subscription settings on dune.com and adjust your limits.`**
- `GET /query/8562361/results` returns `not found: No execution found for the latest version of the given query`, so **there is no cached row set to fall back on**. Zero rows this run

### What this means, and it is not a query bug
1. **No launch set was read.** v4.fun and PonsV2 both unqueried, so the v4.fun zero-streak counter (3 Sep zero, 4 Sep two launches, 5 Sep zero) has a **gap at 6 Sep, not a third zero**. The three-consecutive-zeros rule set on 4 Sep cannot advance on a missing reading. **Do not treat this gap as a zero**
2. **No repeat-operator check ran.** The standing intersect of the new funder set against the existing deployer column, the single line that found cluster A, did not execute. Any launch in the last 24 hours is unattributed at hour zero for a reason that has nothing to do with the chain
3. **The three clusters are unchanged by default, not by measurement.** A: `0xba545fbc19b5d141a0166836b47d8432d844a7ae` + `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f` · B: `0x2f075cbdce9c2e776bb720f7d5300e9c26915983` + `0xbc2aaca7cedfc9c07d66d19444f4f87282a8dd22` · C: `0x18c2ff18dcec0d3ca5d338ed68bab88a3dbebc5c` + `0x758d3b0c79a032cfd83c4315d071d333bebf203e`. **Unread is not unchanged; say so rather than carrying the line forward as a finding**
4. **The ten children of `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c` were not re-checked for launches.** The 5 Sep note made that the number to re-read each run, because the moment one of them appears in a launch set the cluster converts from an exit network to an operator network. Owed again

### ACTION REQUIRED FROM BEN, this is the only thing that unblocks the file
**Raise the datapoint limit (and the per-query resource cap) in Dune subscription settings.** Until then this file cannot advance, and three legs in [[screen-log-outcomes-4]] stay frozen with it: SCL falsifier 2 (holder count), DICE falsifier 3 (the fee-EOA transfer-log read), and the chain-wide holder-count gap — all three of which the 5 Sep run identified as closable by one `robinhood.logs` query.

### Method note added 6 Sep 2026
**A quota failure and a "nothing new" result look identical in a summary line and must never be written the same way.** The 3 Sep run drew a structural conclusion ("the pad is dead") from a real zero and was wrong; today there is no reading at all. **Record the failure mode explicitly in the run header, and never let a missing run count toward a streak, a base rate or a cluster's "unchanged" status.**

### Still owed into the next run, unchanged from 5 Sep and now one day older
the PonsV2 launch-selector filter · degree tests on the top five real PonsV2 launchers once that filter exists · degree-test `0x88766f7d78004dd5b2556b6d88bab685859dde5b` (funder of `0xb451b975731bc04302b06e68ddace5e04b947b89`, untested since 1 Sep) · the `claimCreator` history of `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f` on the v4.fun fee router `0xCa3ea6d93aE7Ec71edC74AbaaCCd8bA736373045` (11 launches, still 0 extracted, owed since 4 Sep) · the `robinhood.logs` query that closes the SCL, DICE and holder-count legs at once · re-check the ten children of `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c` for any launch.

---

## Run of 7 Sep 2026. **DID NOT RUN. DUNE STILL OUT OF QUOTA, SECOND CONSECUTIVE DAY.**

- PATCH of standing query 8562361 **succeeded**, `{"query_id":8562361}`. The v4.fun 3-day funding SQL was written in, with the funding and relay scans bounded to `now() - interval '90' day` per the 6 Sep narrowing
- **POST /execute refused before execution: HTTP 402, `This api request would exceed your configured datapoint limit per billing cycle. Please visit your subscription settings on dune.com and adjust your limits.`** Identical refusal to 6 Sep, so nothing was consumed and nothing was learned
- `GET /query/8562361/results` returns `not found: No execution found for the latest version of the given query`. **No cached row set to fall back on. Zero rows for a second day**

### FINDING OF THIS RUN, and it is about the file's own plumbing, not the chain
**The standing query 8562361 had been repointed at an unrelated Solana `coinbench holder backfill` SQL by another session before this run.** Its stored `query_sql` read `select trader_id as w, sum(...) from ... token_bought_mint_address='HcRLc9VDgjLeK154xDawfb1dmVJ98DoSqcwTHGqiDeJR'`, which has nothing to do with the operator graph. It has been PATCHed back to the v4.fun operator SQL.
**STANDING RULE ADDED: query 8562361 is shared across sessions and gets overwritten. Always GET `/query/8562361` and read the stored `query_sql` BEFORE trusting any result set, and never read `/results` without confirming the body first.** A cached result from someone else's SQL would have been reported here as an operator finding.

### What did NOT run, and the state that must not be carried forward as a finding
1. **No launch set was read.** v4.fun and PonsV2 both unqueried. The v4.fun zero-streak counter now has gaps at **both 6 and 7 Sep**, not zeros. Sequence on record: 3 Sep zero, 4 Sep two launches, 5 Sep zero, 6 Sep GAP, 7 Sep GAP. The three-consecutive-zeros rule set on 4 Sep cannot advance across a gap. **Do not retire the v4.fun set**
2. **No repeat-operator check ran.** The standing intersect of the new funder set against the existing deployer column, the one line that found cluster A, did not execute for a second day. **Any launch in the last 48 hours is unattributed at hour zero for a billing reason, not a chain reason**
3. **The three clusters are unchanged by default, not by measurement, for a second day.** A: `0xba545fbc19b5d141a0166836b47d8432d844a7ae` + `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f` · B: `0x2f075cbdce9c2e776bb720f7d5300e9c26915983` + `0xbc2aaca7cedfc9c07d66d19444f4f87282a8dd22` · C: `0x18c2ff18dcec0d3ca5d338ed68bab88a3dbebc5c` + `0x758d3b0c79a032cfd83c4315d071d333bebf203e`. **Unread is not unchanged**
4. **The ten children of `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c` were not re-checked for launches** for a second day. That is the number the 5 Sep run said to re-read every run, because the moment one of them appears in a launch set the cluster converts from an exit network to an operator network

### ACTION REQUIRED FROM BEN, second consecutive day of asking, and it is the only thing that unblocks this file
**Raise the datapoint limit and the per-query resource cap in Dune subscription settings.** Until then the operator graph cannot advance, and four falsifier legs stay frozen with it: SCL falsifier 2 (holder count), DICE falsifier 3 (the fee-EOA transfer-log read), COMMOTITTES falsifier 3 (holder count, which has never had a first read), and the chain-wide holder-count gap. All four were identified on 5 Sep as closable by one `robinhood.logs` query.

### Still owed into the next run, unchanged from 5 Sep and now two days older
the PonsV2 launch-selector filter · degree tests on the top five real PonsV2 launchers once that filter exists · degree-test `0x88766f7d78004dd5b2556b6d88bab685859dde5b` (funder of `0xb451b975731bc04302b06e68ddace5e04b947b89`, untested since 1 Sep) · the `claimCreator` history of `0x31511477f9d0a45a65f7e0a211749573fb5e7b7f` on the v4.fun fee router `0xCa3ea6d93aE7Ec71edC74AbaaCCd8bA736373045` (11 launches, still 0 extracted, owed since 4 Sep) · the `robinhood.logs` query that closes the SCL, DICE, COMMOTITTES and holder-count legs at once · re-check the ten children of `0xaf924bbc01686d0e89e00cb69c079daa8a3c595c` for any launch.

---

## Operator added 13 Sep 2026 from the 中国制造 screen (see [[screen-log-5]])

**`0x87ae15bcbf080df92bda15bf6884c9db9c023808`** — deployer of 中国制造 `0xE339A891D14FafD37AF44A3C9F227856268A1620`, Robinhood Chain, via Pons V2 (`PonsV2LaunchDeployer` `0x3711ceA4feaDE896C913C68F01Eda97Cb06D1A42`), 2026-09-13 10:05:10 UTC. X identity **@Trader4apes ("Bezerk", 77 followers, account opened 30 Jul 2026)**.

- **Pattern: SELF-SNIPE AND INSTANT FLIP.** Bought 57,292,641 tokens (**5.73% of supply**) inside the deploy tx `0x0ae1f0611174f2117a1f903233ddd281c75369a48fdce40cdc42acc1313f46e0` for 0.0935 ETH (~$232), sold 100% of it **216 blocks (~22 seconds) later** in `0x53590ffbaeaab7523e219ea1253607a57c4a6f1a091b828c5f4d47bf33b4ab79` for 0.1704 ETH (~$422). Profit ~$190.
- Wallet nonce 17. Residual balances at screen time: 0 tokens, 0 USDG, **0.000198 ETH (~$0.49)**. Funding source and prior launches UNRESOLVED — Robinhood Blockscout was Cloudflare-walled from the container and Dune remains out of quota.
- Fees claimed: none observed. He has not touched the project since 11:10 UTC.
- **Pre-flag shape to watch for, not a rug shape:** he left no bag behind, so the risk he creates is abandonment, not dumping. **Any future launch from this wallet should be treated as a flip-the-snipe launch by a meme hunter, with the operator expected to be gone within hours.**

### THREE PONS SYSTEM ADDRESSES THAT MUST BE HARD-EXCLUDED FROM EVERY HOLDER TABLE
Recorded because misreading one of them produced a wrong "8.16% whale sitting on top of you" line that had to be corrected mid-screen.
- **`0x267444d099b10fB5Ed7c3Cc7B7c767AdcA574952` = PonsV2LaunchLocker.** Takes a standard **8.16%** slot at the graduation second on every Pons V2 launch. Same figure appeared on the B案 screen the same day. **It is locked and cannot sell.**
- **`0x8366a39CC670B4001A1121B8F6A443A643e40951` = Uniswap v4 PoolManager singleton.** Holds the pool's base reserves.
- **`0xbc42404a778e3730d56cfd902a6a6f606ba5fa7a`-class = the Pons bonding curve** for that token (per-token address; read it off the token itself via selector `7165485d`). Holds the full supply pre-graduation, zero after.

**WHY THE USUAL SNIPER TEST FAILS HERE: on Robinhood Chain ordinary user wallets are ALSO contracts with nonce 1**, because the Robinhood app uses smart accounts. So "is this address a contract" carries zero information on this chain and must never be used as a bundle or sniper flag. Match against the known system addresses above instead.

**PONS TOKEN SELF-METADATA, useful on every Pons screen and reachable by `eth_call` with no explorer:** `53cd512a` returns the socials struct (ABI-encoded string array, the project's X link), `7165485d` returns that token's bonding curve, `536dac9b` returns the Pons factory `0x7ed598bcef8bd9edd8c97a195c6d13f40801ec7e`.

**AND THE REPEAT TRAP, third instance now: GoPlus `creator_address` returns the Pons FACTORY, not the human.** It returned `0x3711ceA4feaDE896C913C68F01Eda97Cb06D1A42` here and the B案 screen hit the same thing. **Always read the `from` field of the token's creation transaction.**

---

## Operator added 13 Sep 2026 from the NoteFi screen (see [[screen-log-6]])

**`0xc174c3cb7487e5313f7d0ba455abd018d8ab957e`** — deployer of NoteFi `0xd22efb8cb47b45485d8dfaae036214e2c22e35aa` (ticker $NOTE), Robinhood Chain, Pons V2 curve, pool `0xd4c88fe5358ca0848896ef53652da2d481e513f2` created 2026-09-13 13:57:56 UTC. X identity **@useNoteFi** (created 2026-09-12T07:06:26Z, 406 followers, 8 posts, site notefi.ink).

- **`eth_getTransactionCount` = 2.** Two transactions in the wallet's entire life. **No prior launches on this chain, full stop** — a nonce of 2 cannot conceal a launch history. Balance 0.01905 ETH (~$47). Holds **1.7%** of supply, unsold at screen time.
- Not a cluster member, not a repeat operator, no funder resolved (Blockscout Cloudflare-walled, Dune out of quota).
- **Shape: single-launch anon with a pre-built Replit frontend.** The site (notefi.ink) and the X account both predate the token by one day. The risk here is not a bag overhang (1.7%, no history) but abandonment of a product whose core function already does not work.

### **REUSABLE SHORTCUT, and it replaces a whole Dune query when the graph is blocked**
**`eth_getTransactionCount` on the deployer answers "has this wallet launched before" in ONE RPC call.** A low nonce (under ~5) is proof of no launch history; a high nonce means the question is still open and needs the graph. **Use this first on every screen while Dune is out of quota — it is the cheapest deployer test available and it has no dependency on Blockscout or Dune.**

### **DUNE STATUS, 13 Sep 2026: STILL OUT OF QUOTA. Seventh non-run.**
`POST /query/8562361/execute` returns **`This api request would exceed your configured datapoint limit per billing cycle. Please visit your subscription settings on dune.com and adjust your limits to perform this request.`** — the same HTTP 402 refusal recorded on 6 and 7 Sep, so the limit has been spent for the whole billing cycle, not just a day. **PATCH still succeeds, so SQL writes give a false sense of working; only `/execute` reveals it.**

**ACTION REQUIRED FROM BEN, now asked on three separate days: raise the datapoint limit and the per-query resource cap in Dune subscription settings.** Everything in the "still owed" list below has been frozen since 5 Sep for this one reason.