Legal RAG systems often retrieve passages by semantic relevance, lexical score, or citation match. In versioned legal corpora, however, a passage can be topically relevant and citation-matched while being legally invalid for the operative date. We study law-in-force retrieval: retrieving evidence that matches the authority, is valid at the operative date, and supports the answer. We build LawInForce-eCFR from 805 human-validated eCFR/CFR amendment pairs and 3,220 controlled answer-flipping yes/no questions. We also present AuthorityFirst-RAG, a transparent diagnostic retriever that decomposes retrieval into authority identification, temporal scoping, and rule-support ranking. AuthorityFirst-RAG improves Recall@1 from 2.80% for BM25-date-filter to 30.90%, but generation remains difficult even with GoldOracle evidence. These results show a dual bottleneck in versioned legal RAG: retrieving the law in force and applying numeric legal rules.

The quality of a conversation goes beyond the individual quality of each reply, and instead emerges from how these combine into interactional dynamics that give the conversation its distinctive overall “shape”. However, there is no robust automated method for comparing conversations in terms of their overall dynamics. Such methods could enhance the analysis of conversational data and help evaluate conversational agents more holistically.In this work, we introduce a similarity measure for comparing conversations with respect to their dynamics. We design a validation procedure for testing the robustness of the metric in capturing differences in conversation dynamics and for assessing its sensitivity to the topic of the conversations. To illustrate the measure’s utility, we use it to analyze conversational dynamics in a large online community, bringing new insights into the role of situational power in conversations.