The strength of your community is the best predictor of your project’s long-term viability, but you can't improve what you can't measure. It's particularly hard to convince developers that focusing on community health even matters without quantitative evidence proving its importance. The huge problems that occur if you fail to deal with issues quickly are nearly unbelievable, but over the past 10 years we've gradually begun to realize the need to think about the people involved in FLOSS projects and not just their output, whether its code, docs, or something else. The most important part of metrics is context -- you need to compare them to something, whether it's historical trends or other open-source projects. Furthermore, by understanding the qualities of data, including statistically expected variation versus unusual outliers, you can gain a true understanding of, and even predict, community health and growth.