Flexible Systems for Synthesis in Roam Research with Professor Joel Chan

A synthesis workflow built around Roam Research is aimed at turning the “black box” of literature review into a learnable, debuggable process—so researchers can trace claims back to evidence, revisit earlier conclusions as new context arrives, and share a reusable structure with others. The core idea is to separate notes into distinct roles—observation notes (grounded, contextualized results), synthesis notes (claims that generalize or interpret across observations), and context snippets (the specific details—tables, quotes, figures, page locations—that justify each observation). That separation creates a disciplined path from reading to insight, while still leaving room for creativity and later correction.

Joel Chan, an assistant professor of information science at the University of Maryland, frames knowledge synthesis as a “hidden dark art” that many researchers struggle to systematize. His broader research focuses on tools and social practices for creative knowledge work, especially how people integrate information across disciplines to form coherent problem frames, theories of change, and actionable recommendations. In that setting, synthesis isn’t just summarizing papers; it’s constructing something new—like Darwin’s natural selection—by reorganizing relationships among observations into an explanation or design argument. He contrasts “assemblage” reviews that list what different authors found with synthesis that produces a new viewpoint, model, or decision-relevant set of conditions.

The workflow’s practical target audience is researchers who feel stuck between reading too many papers and reaching the endpoint where a literature review yields promising angles of attack. Chan’s goal is to publish a conceptual model and practical guide that lowers the barrier for others to replicate the method, adapt it to their constraints, and stress-test it. He emphasizes that the system should not force copy-paste behavior; instead, it should make the rationale and theoretical grounding visible enough that users can modify the process when it doesn’t fit.

A key mechanism is the “evidence ladder” from specific to general: observation notes capture results in past tense and context-specific language (e.g., a contact-tracing study finding lower secondary attack rates for children under 10). Synthesis notes then state a more general claim (e.g., children are about half as likely to contract given equivalent exposure), while also allowing counter-claims when evidence conflicts. This structure supports both rigor and creativity: it slows down premature conclusions and makes it possible to backtrack to the underlying observations when new studies, different contexts, or longer time horizons change the interpretation.

Questions act as the navigational layer. Instead of treating notes as a flat library, the system organizes work around open questions; sources and observations feed those questions, and syntheses accumulate as provisional answers. Speculation is treated as a first-class activity too, but it lives in linked references and question-linked spaces until it earns grounding. Chan also highlights a Roam-specific implementation detail: indentation and block structure can represent argument steps even when it reduces certain query conveniences, so the system balances writing ergonomics with retrieval.

Overall, the approach aims for long-term reuse and collective intelligence: other researchers can start from a structured subset of questions, observations, syntheses, and context rather than extracting everything from scratch. The method is presented as simple in principle but powerful in practice—designed to help ideas “marinate” over time until they crystallize into claims worth sharing and revisiting.