That's a pretty uncharitable characterization of their experiments framework. Have you ever worked on large-scale software with tens of millions of users or end

That's a pretty uncharitable characterization of their experiments framework. Have you ever worked on large-scale software with tens of millions of users or end-points? Then no doubt you'll know the tremendous value incremental deployments and test/control groups have.