An alternative way to deal with an unreliable prototype is to make a reliable wrapper around it. Such as paying time in exchange for reliability (e.g. TCP/IP,

An alternative way to deal with an unreliable prototype is to make a reliable wrapper around it. Such as paying time in exchange for reliability (e.g. TCP/IP, retry and backoff), disclaimer for unreliable behavior and law-related issues, version gating for file formats/protocols, redundant storages, and killswitches for going rogue-prone machineries.