I disagree with the title, that IPv6 will kill telecom - and I am not sure this was supported by the video??
A couple nits: I agree that NAT may still be present, but disagree that it will be widespread enough to be a serious concern (talking v6-to-v6 NAT, or NPTv6). I truly believe, and hope!, that most IPv6 deployments will follow the default scenario - global addresses, everywhere.
Multiple addresses should not be a problem in the general sense (Source Address Selection handles this properly in almost all cases).
DHCPv6 is certainly still present & usable where needed, even MacOSX now supports it!
A normal host doing normal things will not request AAAA resolution if it is IPv4 only. It *is* a problem if the host *thinks* it has IPv6 connectivity when it doesn't really; but even then - if ICMPv6 error message work correctly - it should fail rather gracefully. (Using Dig as an example of this not working is not quite accurate, that is not a normal user-land traffic behavior - that is troubleshooting and diagnostics type of work.)
I do agree that SIP has some challenges, specifically when one side is IPv4 only and the other side is IPv6 only - and realizing that this is the situation you are in :). Absolutely a problem, try to avoid that where possible (and DNS can help a bit there) ... maybe you should move towards dual-stacking everything, and ensure you can properly decide which protocol to use, and when to use it.
Great point WRT the _lack_ of a broad forklift effort being required - every modern OS ships out as IPv6 capable _and most are also enabled by default. OTTOMH, some corner cases (thinking IPS/IDS and Load Balancers primarily) are lagging ... push you vendor(s)!
