CEP 4 - Attribution of Code Products in the Cyclus Ecosystem

CEP:

4

Title:

Attribution of Code Products in the Cyclus Ecosystem

Last-Modified:

2014-06-27

Author:

Kathryn Huff

Status:

Accepted

Type:

Process

Created:

2014-06-01

Abstract

The purpose of this CEP is to define the method for attribution of code products within the Cyclus ecosystem. In particular, along with any software release a unique, publicly citable object should be created which appropriately attributes its authors. This CEP therefore introduces a method for creating a Digital Object Identifier for each code release, with alphabetically ordered authorship, and encapsulating all of and only the authors responsible for that effort.

Concerns

The concerns driving this CEP are threefold :

  • Unambiguous Citation

  • Author Identification

  • Contribution Equality

First, public and unambiguous citation of scientific and code products like those in the Cyclus ecosystem is of paramount importance to scientific reproducibility. Due to the algorithmic and authorship differences between versions of the codebase, a unique identifier is required for accurate citation of that version. Therefore, to enable reproducibility in the use of the Cyclus ecosystem, each version must be unambiguously citable with a distinct DOI.

Second, to provide an appropriately transparent public link between the identities and efforts of the authors, the true names of the authors should be listed. As with other forms of scientific publication, public attribution of authorship is of importance for appropriately enabling direct praise and blame related to that scientific work [SR03]. This feature of scientific publication holds the scientist accountable for their work and enables them to establish a record of their scientific pursuits. It is for this reason that clear attribution of true names is superior to other options, such as using a pseudonym for the group (Cyclus developers).

Finally, since ordering of authorship has conventional meaning and value in the scientific community, the ordering of authorship for code products in the Cyclus ecosystem should similarly reflect the community consensus concerning relative contribution importance. Specifically, due to the inherently apples-and-oranges nature of code contributions, no metric for code importance was deemed suitable for comparing contribution importance within the Cyclus ecosystem. That is, a lack of hierarchy was preferred. Thus, alphabetical ordering, which is perceived by convention as non-hierarchical, shall be used.

What releases should have a DOI?

Cyclus versions are indicated by numbers of the form : vMAJ.MIN.MICRO-patch. At the very least, major, and minor versions should have associated DOIs. For micro versions, which do not undergo a release process, a DOI is optional.

How should a DOI be created?

Creation of a DOI should be conducted or delegated by the code release manager. Accordingly, CEP3 has been updated to reflect this responsibility. That person is responsible for identifying the code contributors since the last release at that level.

Among online services that have the capability of creating a DOI for a code product, figshare is preferred. It was initially chosen due to both its nascent adoption by the scientific computing community and its native ability to capture github-hosted codebases.

Once the DOI is created, it should be recorded on the cyclus website and should be shared with the development team.

Who is an author?

An author has contributed code contributing to the release. For a major release, all authors that contributed code between version m.0.0 and version n.0.0 should be listed in the n.0.0 code release DOI.

In the same way, for a minor release, all authors that contributed code between 1.m.0 and 1.n.0 should be listed in the 1.n.0 code release DOI. Note that additional authors should not be included if they have not contributed to this minor release. For example, if someone has contributed to version 1.0.0, but not since then, that person should not be listed as an author on the DOI for version 1.1.0.

How are authors listed?

All authors should be listed in (english) alphabetical order. This serves both to (1) convey a lack of hierarchy in code contribution importance and to (2) reflect the true identities of the authors.

Summary

In the absence of a rigid convention in the scientific computing community concerning contribuion metrics, author attribution, and code citation, this CEP proposes that uniquely and publicly citable DOIs as well as comprehensive but flat authorship reflect the needs and desires of the Cyclus development community. Finally, this document provides resources for implementing the requirements it introduces.

Document History

This document is released under the CC-BY 3.0 license.

References and Footnotes

References

[SR03]

Adil E. Shamoo and David B. Resnik. Responsible conduct of research. Oxford University Press, 2003. URL: http://books.google.com/books?hl=en&lr=&id=RLZin-f9eooC&oi=fnd&pg=PA3&dq=responsible+conduct+of+research&ots=RGE6h_FEt5&sig=qaiickHjp9r-qhJYU5TXo5WvoA4 (visited on 2014-06-19).