Use case class 1 -- User Story

Use case class 1:

The interactive paper review process. 

Inspired by general considerations of what open science could be like. 

Present Marie and Antoine:

Marie and Antoine are tasked to review a paper. They are being sent a pdf file. They read the paper and have to infer the workflow, implementation, data, simulation and statistical analysis and interpretation from the text and the plots. It is the explicit goal of this review to ensure a very high scientific quality of the publication, but Marie and Antoine have typically no access to how the described workflow is implemented in reality. They may ask the authors for additional information, but even then they may not possess the computing infrastructure to cross-check any parts of the workflow in a running example. All potential unintended bugs or statistical inconsistencies can only be inferred indirectly from potentially visible inconsistencies in the results. Any result without an obvious inconsistency -- either internally or relative to other work -- can basically not be cross-checked thoroughly when reading the paper. Marie and Antoine are doing the community a great favour, but they are significantly hampered by not having access to the whole workflow.

Future Marie and Antoine:

Using the PUNCH Science Data Platform, a digital Dynamic Research Product DRP will be associated with the paper under review, either publicly since the paper was submitted to arXiv.org, or under access control limitation under the discretion of the journal. Journals could formally cooperate with the PUNCH-SDP and be incorporated in the development process. Marie and Antoine can now cross-check the complete implementation of the workflow, or in case of very data intensive analysis, at least the unique final steps which distinguish the analysis from others. They could run the DRP on the PUNCH-SDP with a single click and look at all the intermediate results, if they want. This would qualitatively improve the thoroughness of scientific review considerably, and bring it from indirect inference to an interactive learning experience. This would also be a strong profit for the authors: Any helpful suggestion by Marie and Antoine on the actual analysis workflow or the extraction of results could directly be included in improved results.