Snapshots feel brittle and because they assume that the first run is the correct run. I've run into issues before where I hadn't fully implemented a feature and

Snapshots feel brittle and because they assume that the first run is the correct run. I've run into issues before where I hadn't fully implemented a feature and the test then said the snapshot failed because it had evolved (in a good way). It doesn't really work well with the TDD philosophy either