Here's a pretty good guide on what Data Smells are vis-à-vis journalism, and it's loosely analogous to the concept of Code Smells.
When you've worked with data enough, you see the same problems pop up with certain types of data again and again. For instance, geocoding that places points in the centroid of a state, making it seem like there is a hotspot where there is none. Or a column that seemed like a category with a few options but is instead a nightmare of freeform text and misspellings.
We'd like to formulate a repo for collecting these data smells in a ready place and using this assembled knowledge as a guide for new explorers not to get waylaid or even as a means of building automated tools for checking the most common data problems.