L7-09 · A research protocol from question to result

Learn¶

Choose one narrow question with an identifiable target. State the proposed contribution relative to a named baseline. A useful first project can compare experiment selection strategies for two or more channel families. Avoid claiming a universal quantum causal learner from one small example.

Write a protocol before tuning. Include hypotheses, process family, allowed instruments, parameter range, resource budget, metric, uncertainty method, and rejection criteria. Check the protocol with an analytic toy case. Then run a controlled pilot and preserve all attempts, including failures.

The final report should separate established results, numerical evidence, and interpretation. A negative result can reveal an identification failure or a cost omitted from earlier comparisons. Publish enough details for replication. Label any proposal beyond the measured regime as a future research question.

Practice¶

Check your understanding¶

What makes the project falsifiable?

• A. A stated observation that would reject its model or claim

• B. Only a positive expected outcome

• C. An unspecified future experiment

Sources¶

This lesson draws on Wang et al., 2021Costa & Shrapnel, 2016Watrous, 2018.

Next: L7-10 · Open problems and critical synthesis