Legitimate decreases need to be considered, too.

In addition to the "gold rush" and "low hanging fruit" principles, I would like to add that a major problem on Wikipedia is that articles on small self-contained topics are much more well-developed than core encyclopedic content. While almost all basic topics have articles with sources, many of them are still in a miserable state, because editors need to bring together a wide range of sources and expertise to improve them: it is much easier to work on quirky topics, where there are only a few reliable sources, so that selection bias is not a problem and relevant knowledge can be acquired easily.