This document discusses different approaches to evaluating information retrieval (IR) systems, including user studies, in-situ evaluation, A/B testing, interleaving, and collection-based evaluation using test collections like Cranfield. It describes the TREC Session Track, which aimed to improve search over an entire user session, and the TREC Tasks Track, which focuses on understanding and assisting users with their underlying tasks. The CLEF Dynamic Search for Complex Tasks task also examines simulating users to dynamically build test collections and understand good document rankings and sessions for complex tasks.