Somewhere in another thread I retracted my issue with the ICMP handling code. But you've nailed my #1 gripe, the per-hop loss indicator is testing control path for diagnostic packets at each hop when the tool is meant to diagnose end to end latency and loss. How do you square a router or many routers showing packet loss when there's 0% end to end loss; it doesn't make sense, it's unintuitive, and that's how misleading inferences manifest.