> We don't have enough resources to run every test for every commit, so they end up batched and we have to rely on humans and heuristics to determine which CL i

> We don't have enough resources to run every test for every commit, so they end up batched and we have to rely on humans and heuristics to determine which CL in the batch to revert. You could automatically bisect with just the failing test to find which commit to revert. At least if it's not a flapping test.