> Engineers are also sandbagged with "we need more data" until a risk-taking idea gets snuffed out of existence.

Not at Google, but what I saw was worse than that. It was "we need more data" over and over and, on each round, I could watch other people's riskier changes just fly by without a shred of evidence that they even did what they were claimed to (let alone that they were beneficial). "Need more data" is not always just inertia. It's often a way of defending territory or enforcing hierarchy. Being "data driven" only really works when it's done consistently, and abuses called out.