Every proof system I have read the docs for is thorough about how verification succeeds and vague about what happens when it fails. Suppose I hold a response and a proof, and the proof does not verify. Now what. Who do I show it to, on what timeline, and what do I get back. If the answer is that I open a support ticket and someone reads it on Tuesday, then the cryptography has not replaced the dispute process, it has only given me better evidence to bring to the same dispute process. That is not nothing. But it means the practical strength of the guarantee is set by the slashing rules and the arbitration path, not by the soundness of the proof, and those are the parts nobody writes about.