Hand Off a Research Packet Someone Can Recheck
Package queries, source decisions, conflicts, and a revision rule so an independent reviewer can reproduce the recommendation.
Today's Goals
A useful research process must survive the original researcher leaving the room. Today you will assemble the fictional Harborfield question map, dated search log, source register, conflict matrix, brief, and claim ledger into a handoff packet. Sam will try to reproduce three consequential claims without asking Maya for an explanation. You will specify what happens when a source is updated, corrected, becomes inaccessible, or is withdrawn. The final deliverable is a reviewable trail and a response procedure, not just a polished recommendation.
The Harborfield case is entirely invented and offline. No actual transit department, resident, or Perplexity result is represented here. You can use a live search product to practice on another public question if you record it separately. The required packet uses the fictional source excerpts so that every learner can complete and inspect the same exercise without product access.
Plain-Language Walkthrough
Make the work reproducible by another person
Imagine Maya is unavailable the morning Jo must discuss the proposal. Sam receives only a final paragraph saying that a night-bus trial looks promising. What could Sam verify? Very little. A footnote to a search assistant answer does not reveal why Maya chose certain questions, which sources she rejected, whether two citations came from one origin, or why the recommendation is conditional. A handoff packet gives Sam the route from decision to claim to original passage and back again. It should also reveal where the route ends because no evidence exists.
Think of a kitchen recipe with ingredients, temperatures, and timing. A photograph of the finished dish is not a recipe. Similarly, a research brief shows the result, while the packet records enough inputs and choices for someone else to reconstruct it. Reproducibility does not mean Sam will automatically agree with Maya's judgment. It means Sam can identify the point of disagreement: a source admission, a unit definition, a sampling limitation, or a decision trade-off. The packet turns “I don't trust this” into a specific review question.
The minimum packet has six connected parts. The D1 question map states Jo's decision, options, scope, as-of date, and unresolved questions. A query log records exact search phrases, dates, and no-result searches, including attempts to find weekend data. The D2 source register records producer, original location, dates, method, allowed and excluded claims, and whether apparently separate pages share an origin. The D3 conflict matrix explains HF-05's correction to HF-01. The D4 brief gives Jo a bounded recommendation. The claim ledger connects each important sentence to a passage and a verdict. A version index tells Sam which editions belong together.
Interview questions
How can Sam reproduce your research without your memory?Sam 怎样不靠你的记忆复现研究?
Common in ChinaCommon overseasIntermediate#research#evidence🔒 Unlock this chapter to see the answer keys.
What happens when new evidence changes the recommendation?新证据改变建议后怎么办?
Common in ChinaCommon overseasIntermediate#research#evidence🔒 Unlock this chapter to see the answer keys.
How should you respond if a cited source is withdrawn?被引用来源撤回时如何处理?
Common in ChinaCommon overseasDeep dive#research#evidence🔒 Unlock this chapter to see the answer keys.
