Harald,
MSDP deployment is as much an experiment as IP multicast deployment
in general. In other words, it's more than just an experiment, but
it's definitely not very successful, which is why we don't want to
put the spec on the STD track. Hope this is not confusing...
it makes sense, actually. The other possible designation would be
"informational", which could in this case mean "this exists, is broken
enough that we don't like it, but we can't change it" - see ICAP, the
pre-OPES HTTP callout protocol for an example.