blockchain / LEVEL 4

Ten excellent peers from one subnet

EST.45 MIN

THEORY / RETRIEVAL

What to restore

Combine reputation ranking with topology diversity
Deduplicate peer identity before applying subnet quotas
Explain limits of scoring against Sybil and eclipse attacks

Peer score is an input, not trust

Latency, valid message rate, protocol violations and resource cost can inform peer scoring. Attackers can farm reputation or create many identities, so selection also needs diversity across network groups, providers, discovery sources and connection direction.

Gossip needs admission and backpressure

Nodes should validate cheap framing before expensive signatures, bound per-peer queues, deduplicate messages and penalize abuse without letting one peer force global work. Eclipse resistance is a layered topology property, not one magic score threshold.

CHECKPOINT

Why is selecting only the globally highest-scoring peers dangerous even when every score is accurate historically?

Confidence: 3/5

02 / IMPLEMENTATION

Implement the contract

Select up to limit peers by descending score then ascending id. Skip banned peers, keep only the highest-ranked record for a duplicate id, and admit at most max_per_subnet peers from one subnet. Zero limits return an empty set.

Initializing editor…

CLOUD SANDBOXnetwork off · 256 MB · 12 s

1 / 64 KB

OUTPUT

Runner is waiting for a submission.