The redundancy technique used is called a Reed-Solomon erasure code, which is pretty standard. It allows for space efficient and yet multi-error tolerant storage.
This is roughly the same technique as used in a RAID disk arrays, except we use a way higher level of fault tolerance than in classic RAID systems.
Furthermore, because we provide a backup system, we always keep the original version of the data at the source.
Therefore, the very unlikely condition needed for you to lose some of your precious data in our system is that (i) your own storage at home suffers a fatal failure and, (ii) during this vulnerability window, all the redundant shards needed to recover your lost data are also lost.
Given that we continuously monitor the failures and immediately initiate recovery procdures when this event occurs, this has just NO chance to happen.