In another thread about GBIF dataset evaluation @datafixer gives us some succinct examples of data issues seen. These clearly offer some insights into skills needed (see question 7). This discovery process is happening at the level of the data publisher. Thanks @datafixer for that post. I wonder at what point or points in our research data pipelines are the most critical ones to take on these a) skills / knowledge needs, b) infrastructure (tools and methods) needs, and c) sustainable community development beyond workshops.