repro-in-cs

repro-in-cs copied to clipboard

Welcome to repo-repe-repro: the repository to repeat an experiment in ``reproducibility''! Recently, the results from

http://reproducibility.cs.arizona.edu/

were noticed by several people. This resulted in significant discussion on social media sites:

https://plus.google.com/+ShriramKrishnamurthi/posts/Tf4D2HKQRKi https://www.facebook.com/shriram.krishnamurthi/posts/10153931516105581?stream_ref=10

A quick reading would suggest that a large number of systems papers have unusable artifacts. However, several authors reported that their work had not been represented fairly. In addition, some disinterested people found problems with the evaluation of work by others.

As the claims multiplied, it became clear that the evaluation done here warranted further investigation. This repository is meant to enable and publicize the findings of that investigation.

The data from this repository is visualized here:

http://cs.brown.edu/~sk/Memos/Examining-Reproducibility/

You can focus on the data/ subdirectory and ignore everything else.

If you really want to build the Web site:

• install DrRacket 6.0 (http://download.racket-lang.org/)
• make sure racket is in your path
• run "make"

This will generate data/index.html.

Note: metadata/summary.csv and most of the data files in data/ are from Collberg, et al. We thank them for making their data public, which has enabled this project.