Apologies all for the OT; just a few _brief_ replies. If anyone wants a serious discussion about this, the internet-history list would be the place to start it. Yes, but not explicitly; the 'official' IETF position was 'IPv6 will replace IPv4', and they pretty consistently refused to acknowledge that NAT would likely play a major role. I 'sort of' understand the second part - NAT is, architecturaly, very grubby (for a long list of reasons this is not the place to go into) - but it soon got the point of ostrich-like refusal to recognize reality - which meant that instead of an _architected_ approac to using NAT, it mostly got an utterly 'ad hoc' adoption. It's different from IPv4 (i.e. old code can't understand it), but not different enough (i.e. it doesn't have enough new capabilities to make it worth switching to - IPv4 has many architectural issues, but that topic is too complex to go into here). You have to start by realizing that IPv4 addresses serve at least three functions: i) identify the communicating device (in the sense that 'Noel Chiappa' identifies me), ii) says _where_ the thing is in the Internet (like a street address does IRL), and iii) is used by intermediate switching nodes to forward traffic. So the first step is to pull out ii) and iii), which can be done without modifying the hosts, and there are many designs that did so. Alas, a fuller discussion of this complex topic is not really appropriate here... Ask on internet-history, if you want to know more. Noel