Scientific publishing has a deep problem right now with papers being published with data, algorithms, and/or code being proprietary. In addition but not always related, there is the problem of reproducing results in papers. From bad data to crafting results to have p be under 5% causes headaches for scientists and in general creates an ecosystem of bad science.

If I had enough time, I'd like to do two things:

1) Create a third party rating system where we attempt to rate the quality of the paper based on openness, reproducibility, who funded the research, data quality, etc. The hope is that researchers will not look for papers in their favorite journals, but refine their search to only quality papers.

2) In a similar vain to the above, have a company that does nothing but tries to reproduce experiments in papers. I think this would be a ton of fun and provide great value to the world. Where funding would come from however is unknown...