Why would a unidirectional link failure cause a switching loop?

Question

Based on what I've read so far about UDLD and LoopGuard, it seems that in the case of a unidirectional link failure, a switching loop would form because the blocking port would transition into a forwarding state. However, I don't understand why the other switch wouldn't transition from a designated port to a blocking port, as this would mitigate the loop. Let me know if I'm missing some information here. To illustrate my question, here's a sample topology I pulled from the internet:

In the above topology, once SW2 stops receiving BPDU's from SW3, Gi0/1 on SW2 would supposedly transition into a forwarding state. So why wouldn't Gi0/1 on SW3 transition into a blocking state once it starts receiving traffic from SW2?

Answer 1

So why wouldn't Gi0/1 on SW3 transition into a blocking state once it starts receiving traffic from SW2?

The spanning tree calculation does not depend on "receiving traffic," but on receiving BPDUs.

SW3 still hears the same BPDUs from SW2 with a higher priority/cost than itself, so SW3 decides it's the designated switch, and sets gi0/1 to forwarding. That creates the loop..