A synthesis agent reports a "major contradiction" between a survey finding 22% adoption and another finding 61%. On inspection, the surveys were conducted in 2020 and 2025. Subagent outputs include the figures and sources but no dates. What is the most effective fix?

Answer: A — Require subagents to include publication and data-collection dates in their structured outputs

Without dates, a change over time looks like a conflict. Requiring publication and collection dates lets synthesis interpret temporal differences correctly. (B) is arbitrary selection. (C) discards useful historical data. (D) is unrelated, and content types should be rendered in fitting formats anyway.