Random KV eviction matches learned eviction policies at higher throughput

'Random Attention' (arXiv 2609.03430, Fri 4 Sep digest; UIUC Heng Wang & Jiawei Han with Salesforce AI Research) shows that during reasoning, the choice signal in KV-cache eviction contributes almost nothing: keeping the prompt and evicting uniformly at random within each attention head — with no scoring at all — matches the previous strongest eviction policies across four models and six reasoning tasks, while delivering 32-43% higher throughput when deployed in vLLM. The explanation: the prompt is the fragile part, and reasoning traces protect themselves through redundancy (textual restatement plus per-head copies), so random sampling retains enough copies. The same digest carries an independent convergence: InertiaKV (2609.03515, EMNLP 2026 main) shows that under aggressive compression, temporal aggregation makes near-rank-preserving scorer modifications indistinguishable at the eviction-set level, and proposes Score-Free decoding (score the full context once, freeze the ranking: +0.07 average quality change, no further scoring).

A large slice of KV-compression research is scoring functions, and two independent teams now report that under aggressive budgets the scoring matters far less than assumed — reasoning traces are self-redundant, prompts are not. Practical read: protect the prompt, randomize the rest, and treat scorer-based methods' gains skeptically unless they beat this much cheaper baseline; the vLLM-deployable implementation is public.