MIP.9 - Blocklisting malicious validators from Stake Auction Marketplace to combat sandwich attacks

kron · February 7, 2025, 9:17pm

Introduction

Based on widespread community concerns and transparent research findings, this DAO proposal seeks to implement a blocklist mechanism within the Stake Auction Marketplace (SAM). Sandwich attacks—where validators manipulate transaction ordering to profit from front-running and back-running—have been widely recognized by the community as harmful practices that undermine fair trade execution and decentralization. Marinade is committed to fostering a staker-aligned, market-driven redistribution system. This proposal seeks to ban validators who have been repetitively found to engage in sandwiching and other malicious behavior from access to SAM-delegated stake.

1. Rationale

Some validators facilitate sandwich attacks and other malicious activities, which leads to unfairly high yields for malicious actors gained predominantly at the expense of retail users. This not only degrades user experience and leads to financial loss but also damages the overall Solana reputation and tends to concentrate stakes with malicious actors, undermining decentralization.

As Marinade’s core mission is to create a system where staking rewards are redistributed fairly through a market-driven mechanism, it only fits the exclusion of malicious actors from access to SAM’s stake. The proposal reinforces SAM’s integrity and ensures its delegations align well with community values.

This proposal focuses solely on banning validators repeatedly identified as malicious—those who meet objective criteria for sandwich attacks or other malicious behavior. No validator will be blocklisted for reasons other than clear, repeated, and harmful behavior that disrupts fair and open market dynamics.

2. How are Malicious Validators flagged?

Public tools were used to flag sandwiching transactions. The collected data spans multiple epochs, ensuring a comprehensive overview of validator behavior.
Validators were put on the blocklist if they were detected facilitating more than a decent number of sandwich attacks concerning their stake across multiple epochs.
Validators have the opportunity to appeal if they believe they were unjustly blocked. Those demonstrating reform can be reinstated after a one-month observation period, while repeat offenders will face permanent exclusion.

3. Blocklist

The proposal blocks the following 73 validators from receiving further Marinade stake in the Stake Auction Marketplace:

FBKFWadXZJahGtFitAsBvbqh5968gLY7dMBBJUoUjeNi
KBoNKoxPjdEmR4TrV9wH9wi96x9vSAs9NET94h91SGJ
4ZUdaCPz4t1gJK6bFN2YdG6BDfxZ3ApvGMiQUmKPPtny
4o2TxN5RNxjiLvEbK56ZyaZ1bg3ZTgaTLpPRqi8vEkRS
GvihePhu26EfKNsR8BPNthQE5Le9AGDVUoT1Co4EZvpD
2x3pmdwex3s71i8c4VvnHCnYTwXtqPU1LgdX5BLDtN9L
ANTmwMfoBzEk31Rw8e11nLRvDgzWEVEMGUUy1v9fdRxV
Fire6ZGPLaqBBGWXC8PgweVjREVXRhwzgRNkdGs1wfQM
Frog1Fks1AVN8ywFH3HTFeYojq6LQqoEPzgQFx2Kz5Ch
8ZvC5d39VKnz4UUJV928s76Mv7FdSbX5WwHCuCFAanj3
3D31JtXceje3h2QQynhDnZ1b54oY38rMoJBnFtGdBz59
Ez4heEN9SzxrjgYA7wBWK8FwxAuYXnmwALADGfpB32E6
7Q79sw9Sb625PFzxTvRzhgYme8Jb6mzuTR1CaU3UfV1w
Bs1AYgU6v6MiKivhpNpHnU9VePJAfdeC1yC3FuRaBWNa
71M936kzQRe7eWrABba6yKqPsmTMVhijQqDNQP9qM9pP
4SKy65C9373k9WZnq2ViR7nq8eCu32TkLhoXq45MYQm6
b8ThsbsARWyjqWrBdU9JNFhXg8ZAjDaJtaqXXzy1sRS
1znL3zFHi3znoaz6T6rnnEnRj8Ar3fohDq7ZNk37sUL
CZLU7uNs1SNvmV1vodR29hwJFjQ2eVUXsQJwGt2SKVKk
CyUEJ2KvRCDZqPM5PZ7RVznyjsto8zKc5or6CJZJDPda
Atom7LRkdXj6MBoWJPgjaetrCMrgB9nnkQBYXTWE8Z3S
BJu6CLyEP2M5Fvj88DF7ZcJYhs9qb2FYBBrVKwrFYoQk
8JyX6MV8KshAo8ajUtXBHws5uhtZnH4nzBU2x9LhEBvo
CbyfiyQAy9pyWKrAv3KViqHgcDPQ9ECmYx3eaQoV5hBw
Bnqie7FYWudbSuBjyRHzoKQrz7eGFxmY9wFAMQKKQjEe
8bDP7mZsx6Z1pZbRoMtzj5AXaqoyBLqEfgAi157AnKJX
AYNvSS5XV23ezdUBUzMjrZoGpmDibADE7P3edS4wNKAv
HS8DF8wP3A8qSHPnzbE6pSvXub4FeZQ6xMNSJuqArAHS
5HXxjDZwm7MAZAm2aCgGcGRr3SKiwugcQymoByyd7pfv
27CnXybL6bvwgw869z2JmA6WtGypryEVJRYX2Wg3WM4F
3n7mZaehg8vskkSQY7VuF7yxpG1tYiqeVVHAwgagWjAo
5iSuYrTc2k9SLod5p2iKx62GyvvfAfvkwKMMZu4AmDhe
5cQ27F39MuLu3ZHSyuTBxu9xaarnxezAAiLuyf1FCL2W
EJc4QLsiiX7ap78EqcKBhxZEMUTHZVterxnw9zqaH5rD
BNEYUi39wjxeK6p7NuCS6AGiWhcmft1sVNNzfp9yreEd
6w2vQK2ZZkj3unHMCQjoXLqUwCAxB648NR1yBerUbFNP
4ZntDHEBsPycaJPm4JxPwBc6ZhQHHCQorvuWaS116Wn4
BPX9WVx1MjVpSWY9fo3hGXfE74yKXcreS2XYKjpiyJfZ
GBuP6xK2zcUHbQuUWM4gbBjom46AomsG8JzSp1bzJyn8
FPDzex8iJBMg9APpimUuhNcfaQJnUFWxSSJmdbiyCtZw
ESX5Q9powgfUL8cssZmLMwaJCjfRDn8kYJ5ZVZqPU95P
hDwKSYB3cG1oL68zTPpNZsiAmVe2W6ptCFdeb7z8wFc
82LRaauvKeJV5YxKaXhCL7DsyGWTNFGneCkQ2Ej671M8
8KWzT7D9AuZShQEKdSUNYJXMXKugRvWDK5ctWBT1yE7y
9e8SzLBT4wgCLgQwifj8fSGd6CdefbmJSMboesrETmG
HG3siAZ8sSHbejLNGTrCUN3MRHu3Ngn52Tud9wwAbLi5
woJhJEATD57sEhxCSQri5FvVAv4qgzobNspqXyfSLBV
FidrVxdx4bXj4Myse6dtcmeGXDT1iuwG5YCdSV7jkwcp
9sgALTPB3UPRxkikfSb5w3jXcHkZ6M9Mg4SRBqQaR2Si
HzNwDXpszprUUVvxPtiQMV44C9Has24HEBpPRoG4gg1T
HjE3Aj5Z72U7k7p22qNqQnBr8VwtBj9hWkefZFFqmGNo
AGsrW2sp9mqNp7g9Nqw6g1A9wMyDYysqTN6d8oaw2W4s
7v5fMKBqC9PuwjSdS9k9JU7efEXmq3bHTMF5fuSHnqrm
Xrh9LuJJSNWfN686vGbj7kYT9gvrchufFo64KHNqxA8
6yecu2K5BN1kYCYf6wLEYECg1Dmqf2CyJuD4uZ6iDAQH
9xoQZuDuMLT8TLHWcEUx3QAC4Rr2wNkgriN4KFo6FTgS
Garx8JMpBAhJhLyci6KySLbqCFCHLuxUNqDXs96Xgw2b
7tegjkVvZmYSpGJQVpiRqKuehxtiAEEXyWk1NEbXriGC
CcHWz3tpDVr1gVXLaSj4xxh915Ksd9ze5i1FFFKFJGRq
HVqbYJgHWR8X5WaAyUebtff8LB3mvToe3CQRQMjbLjJ6
Bkucd9XTD2geqNsgcbMcqsSnLhqfHUuDvAUSGCXsJBdK
J1XPKDSbMgByUURYLQW65S2z94g6dHJdXDs2F1jhPLG9
ECULsxDc7pqadpQSQxFaNWAsemLW3wQq1vDKXg5P7GjW
9qxAWvMWzPLcPmPsj2SXnZ5HUWvgFjFVnQCAYLoLtSV5
3LKjD9Cb8RKKbmwM3LphHEvfZdjEU4rAFGDDUiVnuXhJ
HyV9NJ5oQXtZb4Apk6t7ZqyFQpJNFARRutcbP6NpdU69
FxhKTay63tcYR8WuXwRvWJCmAV239fXizqRXNMMMqhjB
664Lpj7PgLLd8SnG8oWWaCJ821P92TL2maMfwvqUYn9T
xenMPWy6UGzSffNmR77roHnrndzjLDf2zedw5pFcSnk
GWeCYsThGTLzsozCiwDUHoDHPwMoij91Vd5Cn7NDoWjP
6aCna9ZopJJUuTijkuKLmd57tnMco8KQBH7J8ydCjT2r
ERCKpwRXracenvc4jF9or7STjME9bmiE6iMArCPvxpaF
9g6RHEoHaBYJkH3ukDg4bhfgKYxXXHkAebqUpWWvA1cN

4. Expected Effects on SAM APY

Applying this methodology to recent epochs, from 680 to 735, the Marinade Labs team found that:

SAM wanted to delegate between 2.0% and 20.0% of the total stake to blocklisted validators.
In epoch 710, up to 1.27M SOL (~15.3% of the total stake) was allocated to these validators.
SAM APY decreased by 0.9 percentage points on average, at most by 4.5 percentage points.

Conclusion

The proposal wants Marinade DAO to remain steadfast in its commitment to a transparent, decentralized, and staker-first ecosystem. This proposal addresses the widespread and well-documented issue of sandwich attacks—recognized by the community and validated through public, reproducible research—and reinforces the protocol’s mission to maintain a market-driven redistribution system that benefits all stakers and fosters fair and open markets across the Solana ecosystem.

Marinade DAO members, please review this proposal, provide feedback, and cast your vote to exclude malicious validators from SAM.

mint · February 12, 2025, 5:25am

This is a good start despite being very late. IMO there should be continuous effort to keep identifying new malicious validators as it’s very easy to switch keypairs. We at Hanabi Staking is committed to doing that and have identified a lot more malicious validators recently, taking up >50% of all Marinade stake. We’ll publish our methodology in due course pending further verification but the provisional results can be viewed on hanabi.so :: Marinade Stake Selling.

dave_sf · February 14, 2025, 1:37pm

We at Staking Facilities support MIP-9 and the effort to blocklist malicious validators from the SAM. Sandwiching attacks undermine trust in the network, harm users, and degrade the Solana ecosystem.

As a validator, we believe it’s essential to foster an environment where users can stake with confidence, knowing that manipulative tactics will not go unchecked. MIP-9 is a proactive measure that sends a clear signal: protecting stakeholders and maintaining a reliable network is a top priority.

Solus · February 16, 2025, 2:06pm

Better late than never.