Sharing and Preserving Computational Analyses for Posterity with encapsulator
Open data and open source software have been proposed as the primary solutions to the lack of reproducibility observed in a number of disciplines. However, simply releasing data and code does not automatically make results reproducible. There are numerous details essential to re-running computation and analysis, for example, minor changes in the computational environment can significantly affect the outcome, yet computational context is rarely provided. We present encapsulator, a system for producing a time capsule that preserves this context and provides end users with a self-contained environment. One of our primary goals is to make capsule creation accessible to a wide range of scientists with minimal technical expertise. Our solution is close to the vision of an executable paper in that it provides end users with a fully featured desktop environment for validating findings or conducting new research that builds on prior work.
READ FULL TEXT