Article on transparency, reproducibility, and open development


#1

A pre-preprint for an article is available for download. It should not be circulated or quoted. If accepted, it could be published before November 2017 under a CC BY 4.0 license. The title and abstract are given below.


S T A T U S  :    C L O S E - T O - S U B M I S S I O N

Energy system modeling: public transparency, scientific reproducibility, and open development

Robbie Morrison

Release: 06
Date: 11 July 2017

Caution

  • all legal statements are currently provisional
  • comments about the machine use of copyrighted datasets may need to be reclassified as unresolved
  • I am currently seeking more information on the above

Highlights

  • growing calls for policy-based energy system models to be “opened up”
  • energy system modeling projects are adopting open source development methods
  • energy system database projects are co-evolving to serve datasets to open models
  • source code distributed under standard copyright cannot be legally used, built, or run
  • datasets distributed under standard copyright cannot be legally machine processed

Abstract

A switch to open energy system models and the establishment of open energy system databases to support these models began in earnest in 2010. As of mid-2017, such projects number about 45, up from five in 2010 and none in 2000. Three distinct yet overlapping drivers can explain this shift in paradigm: a desire for improved public transparency, the need for genuine scientific reproducibility, and a nascent experiment to see whether open source development methods can improve academic productivity and quality. This article examines these drivers, their tensions, and the central role that open software and dataset licensing plays. It also provides an audit of open energy system projects by type, license, and country.

The key message is that while public transparency can often be served by the publication of code and data under standard copyright, scientific reproducibility and open development can only be achieved through the open licensing of the associated software and datasets. These requirements necessarily extend to the public datasets from official and semi-official sources that are normally indispensable when building public policy models. Open development has the potential to improve public trust while publishing behind paywalls can hinder participation.


morrison-code-data-licensing-commentary-2017.06.pdf (1.5 MB)