In addition to the reasons given by Steve, Paul, and others, the barrier to entry to the standard library has grown as the domains to which Python is applied have increased. To justify the maintenance effort, when considering a module or package for inclusion in the standard library, you want it to be as broadly useful as possible. Is YAML widely enough used in domains as varied as web application development, bioinformatics, and machine learning to justify its inclusion in the standard library? Maybe not. In addition, I suspect more and more people are using virtual environments of one sort or another. When constructing such environments it's pretty trivial to tailor them to contain just those modules and packages appropriate to a particular task. I use Conda environments almost exclusively these days. It frees me from the glacial pace of updates to the default Python installation at work (stuck on 2.7.2).
Skip -- https://mail.python.org/mailman/listinfo/python-list
