At some point in the future I'd like to see PLoS ONE allow authors to include data, but not in the style of supplementary materials in other journals. I think it's important that the data be linked to the specific results in the paper so that we can understand the complex analysis that makes up a paper.

I like to think of reproducible research as the scientific research analogue of the open source idea in software. Simply giving someone raw data is like giving someone the source code to a large project without giving them the Makefile. Yes, it's possible to figure it all out eventually, but we can surely do better to achieve the goal of reproducibility.