Transparency

To assess the energy efficiency of a product, the study considered as a further criterion the downward compatibility of the software packages as a function of the operating system used. This criterion required a very different approach to evaluation in the study. While many Linux programs do not have fixed minimum requirements for the age of the operating system, there are usually such specifications under Windows. Some Windows applications would still run on a 15-year-old OS as a platform. According to the measurements, operating system versions that were no older than 10 years could be considered as potential platforms under Linux.

The researchers also examined the transparency of the software with regard to its licensing and possibilities for further development by the user. Here the open source software packages released under a free license had an obvious advantage. At the same time, the researchers made it clear that the impact of open source code on energy demand cannot be assessed as a "general rule," but is instead a "strongly context-dependent" indicator.

Transparency also includes the handling of vulnerabilities by vendors and developers and the number of vulnerabilities found using CVE data collection [4]. This is where the free software impressed; it can be used securely for a longer period.

In a further step, the project teams investigated the extent to which ballast-free deinstallation of the respective software is possible; this is part of the "autonomy of use" category. Here, ballast remained on the local SUTs of all candidates. With server-side applications, users had to remove the data manually throughout due to the lack of automatic routines. When restoring data, the local applications offered more flexibility than the server-side applications due to the autosave options.

Conclusions

Altogether, the authors of the study argue in favor of introducing an eco label for resource-efficient software. The study demonstrates that differences in resource consumption can be measured using standardized criteria and indicators for each software package.

The project teams led by Öko-Institut have defined standard criteria for determining the resource efficiency of software. This also includes typical usage scenarios that considerably simplify the certification process for an eco label.

The authors do not claim to be able to judge the quality of software applications they tested on the basis of the measurement results. The study has its justification as proof of concept, but due to the narrow database with only one test system at a time, it cannot make any statement on the general energy efficiency of the tested software packages.

One point of criticism concerns the general experimental setup: The study does mention that Ubuntu Linux 16.04.1 LTS is used on the workstation system. However, it does not provide any measured values for the free operating system. It would also be helpful in terms of traceability if the tested products were clearly named.

At the request of this publication, the supervising head of the department informed us that the measurement results on the client SUT cited in the study had only been obtained under Windows 7 Professionsl SP1 64-bit. As a prerequisite for certification with an eco-label, however, the interoperability of software packages should be the subject of the catalog of criteria. Hopefully, in the future, researchers will test the corresponding products under all relevant operating systems available in order to be able to detect any differences.