user-relative names - was:[Re: Yahoo and IPv6]

Scott Weeks surfer at mauigateway.com
Tue May 17 20:26:10 CDT 2011


--- Valdis.Kletnieks at vt.edu wrote:
From: Valdis.Kletnieks at vt.edu

> Why give the corpment (corporate/government contraction) an easy time at it?
> Just like the early days, security and privacy do not seem to be in folk's mind
> when things are being designed.

But more importantly, who has more/better lobbyists, you or the people who
want things like COICA and ACTA?

You're going to have to fix *that* problem before trying to address it at the
protocol level will do any real, lasting good.  Either that or we need a *lot* more TOR
relays (while those are still legal).
-----------------------------------------------------------

It's a multi-layered problem and designers at all layers need to keep privacy in mind.  You can't solve the multi-layered privacy problem with a design at one layer.



--------------------------------------------
Oh, and an article that coincidentally popped up since I hit 'send' on the
previous mail:

http://radar.oreilly.com/2011/05/anonymize-data-limits.html

Designing things to evade good data mining is a *lot* harder than it looks.
--------------------------------------------

This article doesn't really address what we're discussing.  It looks at the 'upper' layer only.  I'm just saying that we don't need an ID that follows us everywhere like, I believe, LOC/ID split and "Unmanaged Internet Architecture" (from the "Persistent Personal Names for Globally Connected Mobile Devices" paper) apparently does (I haven't read their paper thoroughly enough to comment in an authoritative manner, though).  There has got to be another way.  RINA (http://www.cs.bu.edu/fac/matta/Papers/rina-security.pdf) addresses privacy/security, but the nanog show-me-the-code folks were unimpressed with the existing code when I asked the list about it in the past.

scott







More information about the NANOG mailing list