Fernando, >>>> I will observe that Alissa's term "random per-network" isn't in any of the >>>> possibilities >>>> below and the reasons given wouldn't apply if that term were used. >>>> Perhaps that >>>> could be used in a title? >>> >>> Nah. Too complex for a title, and "random" is a bad word - you should always >>> say pseudo-random or (more pedantic) uniformly distributed. >>> >>> IMHO the current title is clearest. >> >> these addresses have the following properties: >> - stable per link >> - randomly generated as opposed to based on a MAC address. >> (making scanning attacks harder, makes tracking across links harder) >> - intended to replace existing EUI-64 identifiers >> >> truth in advertising; given that these addresses are meant to be used 'in >> public', what is >> "privacy enhanced" about them? > > One thing is some node using an address to communicate with you. Another > thing is that other node being able to learn other information by means > of such address. > > -- e.g, since the IID changes from one network to another, you cannot > track the device.
this brings us to the larger issue of how an application should choose an address type. if I wanted "privacy" (assuming some suspension of disbelief (that choice of a particular address results in better privacy)), I would use a temporary address. this draft is about stable addresses. a use case for a stable address/public address, is to provide a service. that would typically mean it is published in DNS or available via some form of service discovery. that is, an address I explicitly don't want to be private. >> the use of "privacy" confuses it with RFC4941 addresses, which these >> addresses do not replace. > > Actually, RFC4941 are called "privacy extensions" (which isn't > incorrect), since those temporary addresses have interesting privacy > features. Probably, people assumed temporary == privacy because, before > stable-privacy-addresses you didn't have any other privacy-enhanced > addresses. > > Both RFC4941 and stable-privacy-addreses are "privacy addresess" -- > RFC4941 are temporary, while stable-privacy are stable. as soon as an address is used externally it isn't private anymore. >> "Stable per-network Addresses for IPv6 Stateless Address Autoconfiguration >> (SPN-SLAAC)"? > > An address could be "stable per network" without having any interesting > privacy/security features. For instance, traditional slaac addresses are > "stable per network", too. true. any other suggestions that does not contain the word "privacy"? I don't want to explain to my dad (substitute with your favourite family member), why he is still being tracked across web-sites (AppNexus, Bluekai, Brightroll, Facebook Connect, Google Analytics, Google Adsense, OpenX, Taboola...) when he explicitly chose to use a privacy enhanced address for the communication. > Me, I don't care much about the title. However, given that folks have > become used to refer to this scheme as "stable-privacy addresses", and > that so far alternative titles don't seem to do a much better job, I'd > leave the title "as is". "random Network Stable Addresses for IPv6 Stateless Address Autoconfiguration (NSA-SLAAC)" :-) cheers, Ole -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------