I've actually been wondering if there's a pattern to these closures related to some scaling issue on the Google infrastructure. Google tends to build services t

I've actually been wondering if there's a pattern to these closures related to some scaling issue on the Google infrastructure. Google tends to build services tightly integrated with a common fabric. Scaling issues might cause the design of smaller services to take more than their share of resources. Consider that Reader and other small services needing event and data synchronization, were often started by independent "20%" teams, earlier on in google's infrastructure build out. So maybe services at a higher priority are needing those resources. Or maybe they're looking at upgrading the "google fabric" in an incompatible way and if you have to redesign fundamental assumptions through the code of smaller services, you might stop and take a hard look at seeing if a large rewrite is worth it.