Context Space exposes citation lag blind spot

Hot take: this is the right kind of annoying research, because it turns a vague indexing complaint into a modelable failure mode. The labels are still heuristic, but the operational warning for lit-review stacks is real.

• –The repo's own logs define a lag edge of roughly 0-6 months after publication, so frontier papers can look isolated even while they are already influential.
• –That biases citation embeddings and graph-proximity ranking toward index latency, not field structure, which is exactly where automated review systems get fooled.
• –The cold-node functional modes are the deeper insight: gateway, protocol, and foundation nodes capture structural roles that raw citation counts and centrality routinely miss.
• –The practical fix is hybrid retrieval: graph signals plus direct PDF/reference parsing, freshness-aware scoring, and an explicit not-yet-indexed bucket.
• –Treat the taxonomy as corpus-specific until it's validated across domains; otherwise you risk turning a useful diagnostic into a universal rule.