I see this argument posted a lot. Even if every one of the underutilized /8 or non publicly announced (US DoD) /8s was given back and allocated in small /24 to /20 chunks to small to medium sized ISPs, it would not have a greatly measurable long term impact. It would postpone the absolute need to migrate to ipv6 by maybe... 3-4 years, globally.

The argument that we should expend a lot of effort (technical, legal, and otherwise) to reclaim a dozen /8s and start making them part of the global routing table has been pretty thoroughly debunked by ARIN, RIPE and APNIC. They are instead focused on getting people to use v6.