GCP disk snapshots: Would losing the full snapshot render the incremental snapshots useless?

On GCP, when you do periodic snapshots of disks, the first snapshot is a full snapshot of the disk and the following snapshots consists of only the deltas of the previous snapshots. If a snapshot is deleted and it has other snapshots dependent on it, the changes required for snapshot to be usable for restoring are moved to the following snapshot. This process is detailed here.

What would happen if a snapshot would fail in such a way that this process didn't happen? If the first snapshot would somehow get deleted or corrupted and the data would not be passed to the second snapshot, would that render all the following snapshots unusable or would GCP now consult the original disk and construct the first snapshot anew or just do a new full snapshot?

If both the original disk and the full snapshot would somehow become corrupted, would you be just terribly out of luck with no chance to recover the disk at all?

Unless Google were to release the actual software source code and algorithms for their snapshot software ...

Snapshots are basically copy-on-write systems with each snapshot being a list of pointers to copies of the changed data (disk blocks). If the snapshot software failed, the type of bug would determine if recovery is possible. Assume all would be lost and plan for that event.

Real-world backup designs never trust only one backup mechanism or one backup storage location or one copy of a backup. For high-value data, I use three forms of backup at all times plus a cold standby.