https://medium.com/bamtechmedia/protocol-chaos-4f8391a532e7

Chaos engineering is a way to introduce unexpected failures in order to discover a system’s failure modes. The chaos engineering tools usually introduce faults that focus on the connections between the system’s components by disrupting the expected operation of the network (by introducing latency, packet loss, all the way to complete network partitions) or by disrupting the operation of the nodes hosting the components (by reducing the available CPU or memory resources). Our work on chaos testing showed that at the start more “damage” can be achieved by focusing on how a system handles unexpected input–trivially structurally invalid messages; or structurally valid messages that contain invalid values; or structurally valid messages with valid, yet damaging or malicious values. Once the system can reliably handle invalid and malicious inputs, the infrastructure chaos engineering can be used to tease out further failure scenarios.

If you simply cannot wait to read the paper, head to https://raw.githubusercontent.com/janm399/papers/master/protocol-chaos/txt/root.pdf otherwise, keep reading for a DR version.