The Chief Scientist has set down seven rules to lift “the rigour and reproducibility” of research practice. Writing in Nature, Alan Finkel reports on ideas, arising from a research leader roundtable.

* “Accredited, practical and respected,” research training as part of PhD programmes. “Institutions must provide explicit instruction in research integrity, data management and professional expectations.”

* Training for PhD supervisors in mentoring and fostering “healthy research culture.”

* Mandatory reporting by supervisors of PhD students’ research impact and career progression

* Alternatives to “simple metrics” for assessing research performance. For example, “the rule of five,” best five papers over past five years, supported by an impact statement

* Completing accredited training programmes as preconditions for public funding. Applicants records in mentoring and peer-review included in assessment

* Journals to be “knowledge custodians” rather than “knowledge distributors.” Custodians do not publish and forget. They ensure that data remain accessible and prioritize concerns about research quality.”

* Grants and promotions are contingent on practice