I think the failure mode that Ted T'so was talking about was data loss, not filesystem corruption. The assumption is that you will be able to replay the journal to get the filesystem back into a known state without fsck.
The only wildcard is the behavior of the flash firmware. If it lies to you and says things are committed to flash when they really aren't, really anything could happen. However, it's hard to see how any filesystem can get around this problem, if it exists. Keep in mind you can't rely on ordering either, since the flash firmware needs to reorder your requests to do wear leveling and so forth.