I agree with Roger about the technical side. LSIDs are largely for insiders with good IT support, preferably building their software themselves. In a way, there is a social business model associated with LSIDs: keep it a small and thus perhaps better manageable community. However this comes at the cost of excluding many initiatives, especially those relying on unmodified or only slightly modified standard software.
From biologists-social standpoint I would argue that:
urn:lsid:zoobank.org:act:8BDC0735-FEA4-4298-83FA-D04F67C3FBEC
is far harder than this:
http://purl.zoobank.org/8BDC0735-FEA4-4298-83FA-D04F67C3FBEC
which, socially, and for people who might want to cite identifiers in paper or PDF form, is harder than this:
http://persistent-id.zoobank.org/8BDC0735-FEA4-4298-83FA-D04F67C3FBEC or http://persistent-identifier.zoobank.org/8BDC0735-FEA4-4298-83FA-D04F67C3FBE...
I just try argue to speak to the communities outside of the inner technological circle, rather than saving a few characters.
These people may also be the managers, making it clear that this class of URLs need to be managed over longer time periods.
Gregor