This sounds really interesting, but the layer of AI slop makes me not want to read it, with dense inhuman statements such as:

    T3a proves that a correct channel is never disabled; it 
    does not prove that a command is issued when two 
    channels fail with wrong values on the same step (a 2–2 
    split at four active channels gives OutsideHypothesis, 
    no command). “Survives two failures” in this article 
    means only T3b, with its own assumption.

And platitudes like

    Synchronisation matters here because comparing outputs 
    word for word only makes sense if the channels compute 
    the same frame at the same time; a channel a step 
    behind looks exactly like a failed one. 

At this point I can ask Claude myself to "tell me about the four computers in Buran".