How do reviewers have enough context to provide good reviews?

Our reviewers can either provide surface or deep reviews.  

Surface reviews are surprisingly comprehensive in the kinds of bugs and technical issues that they can bring up.  For the majority of projects, either first or second pass surface reviews will catch the vast majority of bugs and provide a significant reduction in technical debt.

For more comprehensive architecture level review, a reviewer team is assigned to a project and engages in a kickoff review, where the codebase (minus any proprietary or sensitive data) is worked through before on-going pullrequests are reviewed.