DataVector, WitterFTSO, Chainbase Staking, Comfy Nodes and SenseiNode (Collusion)

ftso group test
1

Proposal

That each of the five providers below be chilled under FIP.02 for collusion.

Each provider gets its own vote. These are five separate management proposals, one address each, voted individually, exactly as every sanction this group has ever passed has been voted. This thread is the shared discussion document for all five, which is also established practice: the O1FTSO and Flare Ocean case ran off one thread naming five providers, the FTSO UK and SolidiFi Nexus case off one thread naming two, and the four October proposals off one thread naming five, with the votes kept separate in each.

Provider Registration weight Closest peer, byte-exact Peers at or above 60% Measurement runs at or above 60%
DataVector 0.927% 87.9% 13 326 of 326
WitterFTSO 0.578% 87.8% 13 326 of 326
Chainbase Staking 1.071% 87.1% 13 326 of 326
Comfy Nodes 0.499% 86.5% 13 292 of 292
SenseiNode 0.505% 84.0% 13 315 of 315

Together they hold 3.58% of registration weight (reward epoch 443).

What FIP.02 says

FIP.02 defines collusion as:

“Collusion, defined as multiple FTSO data providers showing a strong statistical correlation in their submissions or clearly submitting through the same node.”

The first limb asks one question: do the providers show a strong statistical correlation in their submissions? It contains no requirement of agreement, coordination, intent, or common control. The adjacent definition of duplication does contain a common-control element, so the drafters knew how to write one and did not write one here.

No coordination is alleged. None is required for the definition to be met, and nothing below asserts any.

The measurement

The infraction is the byte-exact agreement, and it is the only thing measured below.

Over 120 consecutive voting rounds (1,491,368 to 1,491,487), every pair of the 94 providers revealing values was compared cell by cell. A cell counts only where both providers priced the same feed in the same round, and counts as agreement only where the encoded integer is identical to the last digit. It is a count of reveals already on chain. There is no model in it and no reference instance.

Across all 4,371 pairs of providers on the network, median agreement is 9.7%, and 94.5% of pairs agree on fewer than 20% of cells. Two independent providers reading different venues essentially never produce the same integer to the last digit on a volatile feed, which is why the field rate is the natural null and why the figures below are not a coincidence of market conditions.

The five providers

1. DataVector

Peer Byte-exact agreement Cells compared
0x882ee9ce... 87.9% 7,378
Cottage Nodes 87.3% 7,440
WitterFTSO 87.1% 7,440
Chainbase Staking 86.1% 7,440
Comfy Nodes 84.9% 7,440
Encode Club 84.2% 7,440

And 7 further providers at or above 60%, 13 in total, averaging 80.6%. The next provider below that group agrees with it on 38.8% of cells.

Registration weight 0.927%. It has been at or above 60% with at least one peer in every one of the 326 measurement runs in which it has revealed values, the first on 2026-08-02.

Across the 38 daily readings since 2026-09-14, its agreement with its closest peer has ranged from 69.6% to 90.8%, and it has not fallen below 60% on any of them.

Role Address
Identity 0x6053267bd317fe8f2d51bf063271cd1596d12783
Submit 0x71aec2bc126654d116386864b5f3babce0ca87e0
Delegation 0xcaa49c97318b6bb62b7f9241891d70f87fc05d35
Node 1 0x49e8efa8079afded00f9d2aedcb2c712630950a7
Node 2 0x1f2bc874afd332fd2494174d33c464fe9b9c542b

2. WitterFTSO

Peer Byte-exact agreement Cells compared
Cottage Nodes 87.8% 7,440
DataVector 87.1% 7,440
0x882ee9ce... 86.4% 7,378
Chainbase Staking 85.8% 7,440
Comfy Nodes 83.5% 7,440
Encode Club 83.4% 7,440

And 7 further providers at or above 60%, 13 in total, averaging 79.7%. The next provider below that group agrees with it on 38.5% of cells.

Registration weight 0.578%. It has been at or above 60% with at least one peer in every one of the 326 measurement runs in which it has revealed values, the first on 2026-08-02.

Across the 38 daily readings since 2026-09-14, its agreement with its closest peer has ranged from 66.2% to 90.8%, and it has not fallen below 60% on any of them.

Role Address
Identity 0xd6122bb3d05454b4e4cc9883064a96607eb835bf
Submit 0xfa84cf492ea6e648c09ed1048182ce6f2058bd9f
Delegation 0x4a74e79738009711a6ddb6653614d43ba9f431b9
Node 1 0xecdedf9f40c7e39474bbb64c7a7a258c1bb0e593

3. Chainbase Staking

Peer Byte-exact agreement Cells compared
Cottage Nodes 87.1% 7,440
0x882ee9ce... 86.8% 7,378
DataVector 86.1% 7,440
WitterFTSO 85.8% 7,440
Comfy Nodes 84.7% 7,440
Scintilla 80.2% 7,440

And 7 further providers at or above 60%, 13 in total, averaging 79.7%. The next provider below that group agrees with it on 38.5% of cells.

Registration weight 1.071%. It has been at or above 60% with at least one peer in every one of the 326 measurement runs in which it has revealed values, the first on 2026-08-02.

Across the 38 daily readings since 2026-09-14, its agreement with its closest peer has ranged from 70.0% to 91.7%, and it has not fallen below 60% on any of them.

Role Address
Identity 0x568a27b3770b6ea514c2008d8f6a3fddf61ff017
Submit 0xdc12432b3d1b8f0a7c9c77d80f20f54132970c73
Delegation 0x6434b1ed626585d3e58e995ad3c2cc0d6718755c
Node 1 0x44e7c98506685dbfc828ac8df574ac9183d350b9
Node 2 0x49b4744b6b9115aacb3455bec7237f72135154d6
Node 3 0x4ce31f62d51dd2e527af2d0c6b8a3a76ed38bee9

4. Comfy Nodes

Peer Byte-exact agreement Cells compared
0x882ee9ce... 86.5% 7,378
DataVector 84.9% 7,440
Chainbase Staking 84.7% 7,440
Cottage Nodes 83.9% 7,440
WitterFTSO 83.5% 7,440
Luganodes 82.4% 7,440

And 7 further providers at or above 60%, 13 in total, averaging 79.3%. The next provider below that group agrees with it on 39.1% of cells.

Registration weight 0.499%. It has been at or above 60% with at least one peer in every one of the 292 measurement runs in which it has revealed values, the first on 2026-08-02.

Across the 38 daily readings since 2026-09-14, its agreement with its closest peer has ranged from 69.4% to 89.6%, and it has not fallen below 60% on any of them.

Role Address
Identity 0xc058d02fdd08294d29fd4db631e6edc789741ea6
Submit 0x7b6c3bf0da61353a195fb2accc5c2c6d0b72dcae
Delegation 0x9c44c70180642cf4cfd06d77b84bb8f864ea5697
Node 1 0x9c1bf8866dd532ccf1760c4817deb1f857b38e38

5. SenseiNode

Peer Byte-exact agreement Cells compared
Encode Club 84.0% 7,440
DataVector 83.2% 7,440
WitterFTSO 81.3% 7,440
0x882ee9ce... 81.3% 7,378
Comfy Nodes 80.7% 7,440
Cottage Nodes 80.6% 7,440

And 7 further providers at or above 60%, 13 in total, averaging 77.5%. The next provider below that group agrees with it on 37.8% of cells.

Registration weight 0.505%. It has been at or above 60% with at least one peer in every one of the 315 measurement runs in which it has revealed values, the first on 2026-08-02.

Across the 38 daily readings since 2026-09-14, its agreement with its closest peer has ranged from 62.8% to 91.0%, and it has not fallen below 60% on any of them.

Role Address
Identity 0x12649c1937fbdad9bebc1246110fcafea490b8a2
Submit 0x797a667434454e54fd1e66c25d2c73353143f43f
Delegation 0xe08898b7b8b18dbcddcc6339c8b9c19effa81413
Node 1 0x9c55d0c3d52f3f05fd14d5c012cf7d674a417116

How to verify this before voting

Do not take it on trust. Every input is public.

Each provider is named here by its identity address, which is the address the entity registry, the block explorer and the provider lists all key on. The submit address, which is what appears as tx.from on a reveal, is in each table above.

What a chill has to cover

FIP.16 registration weight is computed from two sources: mirrored stake across a provider’s node ids, and the WNat delegated to its delegation address. VoterRegistry.chill takes a list of beneficiaries, so a chill must name the delegation address and every node id listed above. Chilling the delegation address alone leaves roughly 80% of an average provider’s registration weight untouched, which is the appearance of a sanction rather than a sanction. With every leg named, registration weight is zero and require(_weight > 0) prevents registration for the epochs concerned.

WitterFTSO, Comfy Nodes and SenseiNode each have one registered node id, so each chill names two beneficiaries. DataVector has two registered node ids, so its chill names three beneficiaries. Chainbase Staking has three registered node ids, so its chill names four beneficiaries. All are listed in the tables above.

Precedent

The Management Group has chilled for collusion on this ground many times. In each case below the evidence was the submissions themselves: values that matched or moved together in a way the submissions of independent providers do not. Every one of these carried.

Date Provider Ground Vote
May 2023 FLRLabs Common developer and algorithm with LoveFTSO 61 to 0
May 2023 LoveFTSO Common developer and algorithm with FLRLabs 62 to 0
July 2023 HXK Oracle Same algorithm as Sirius FTSO 54 to 0
July 2023 Sirius FTSO Same algorithm as HXK Oracle 52 to 0
July 2023 CoffeeFTSO Same algorithm as LoveFTSO 54 to 0
July 2023 Best FTSO Same algorithm as tSovreign 47 to 1
July 2023 tSovreign Same algorithm as Best FTSO 47 to 1
July 2023 Worlds FTSO Same algorithm as tSovreign and Best FTSO 49 to 0
December 2023 FLR-Labs Same algorithm and price feeds as an anonymous provider 54 to 0
December 2023 Anonymous provider 0xe51d146c... Same algorithm and price feeds as FLR-Labs 54 to 0
December 2023 Best FTSO Same algorithm as AilandFlare 52 to 0
December 2023 AilandFlare Same algorithm as Best FTSO 51 to 0
November 2025 FTSO UK Shared infrastructure: identical price data to SolidiFi Nexus 39 to 8
November 2025 SolidiFi Nexus Shared infrastructure: identical price data to FTSO UK 40 to 7
September 2026 Provider 0x882ee9ce... Byte-identical submissions with 21 other providers 26 to 8
October 2026 Stakeway Byte-identical submissions with 20 other providers 31 to 6
October 2026 Encode Club Byte-identical submissions with 20 other providers 31 to 6
October 2026 Luganodes Byte-identical submissions with 20 other providers 31 to 6
October 2026 Cottage Nodes Byte-identical submissions with 20 other providers 31 to 6

Three of these are worth reading closely.

FTSO UK and SolidiFi Nexus (2025). The discussion thread rested on one observation: the two were “submitting 100% identical price data”, “identical down to the last decimal in every submission”. The providers answered that it was a configuration error and not intended. The thread’s reply, which the group then voted for, was that “the claim of an ‘accident’ is irrelevant”, and that “by mirroring data, the providers acted as a single entity, not as the two independent oracles they are registered as.” That is the same test applied here: what was submitted, not why.

Best FTSO and AilandFlare (2023). The evidence was two providers “submitting near identical pricing data” over a sustained period. Near identical was sufficient. The figures above are identical to the last digit.

Provider 0x882ee9ce... (September 2026) and the October four. Provider 0x882ee9cebe3d7b38eebd640… (Collusion) is the same finding on the same measurement as this proposal. It passed 26 to 8, closing 2026-09-19, and the four October proposals (Stakeway, Encode Club, Luganodes, Cottage Nodes) each passed 31 to 6, closing 2026-10-22. Provider 0x882ee9ce... is still submitting, and it is among the six closest peers of all five providers named above.

Every sanction this group has voted, 33 of 33 across all four polling deployments, named exactly one address. There has never been a bundled sanction vote, and this is not one. Where several providers were implicated in the same case they were voted separately on the same day, four of them at once on 2023-12-19 with 51 to 54 members voting on each, and again in October. Concurrent per-provider proposals are proven practice here, not a bottleneck.

What a chill means here

Under FIP.02 a first chill removes a provider for two reward epochs. A second is a permanent ban. For all five of these, this is a first chill.

This proposal creates no sanction of its own. It applies one that already exists.

The five proposals

Each will be submitted separately to the Management Group, with this thread as its discussion link. Titles and payload addresses:

Proposal title Address named
DataVector (Collusion) 0x6053267bd317fe8f2d51bf063271cd1596d12783
WitterFTSO (Collusion) 0xd6122bb3d05454b4e4cc9883064a96607eb835bf
Chainbase Staking (Collusion) 0x568a27b3770b6ea514c2008d8f6a3fddf61ff017
Comfy Nodes (Collusion) 0xc058d02fdd08294d29fd4db631e6edc789741ea6
SenseiNode (Collusion) 0x12649c1937fbdad9bebc1246110fcafea490b8a2

Five proposals, one per provider. A “for” vote is a vote to chill.

2

The question going forward should be what is the threshold between vanilla EP deployment and “good enough”

Some providers have openly stated all they’ve done is tinker with the source lists and haven’t bothered with much else.

Does a slightly tinkered EP pose the same threat as a vanilla deploy?

In any case at or close to 90% identical is cause for concern.

1 Like