@eLifeInnovation: Thanks for sharing this here. It's great to see the results of this survey and well done on the write up!
There's plenty of food for thought in there and I'm sure it will be a valuable resource when we're thinking about development priorities for Stencila. For me, a key paragraph was:
Several issues and limitations with implementing more computationally reproducible articles were highlighted. For example, it may not be possible to share research involving large or complex datasets and computational scripts in a format other than the traditional manuscript. Any infrastructure would need to manage dependencies, accommodate versioning, and deliver computational outputs with sufficient speed to be more useful than computing locally. Finally, the imbalance between author burden - including the time taken to produce well-documented assets and the effort to learn these new methods - and third party benefit is a concern, and one which may prevent some researchers from engaging with these technologies without appropriate incentivisation.