[cryptography] wont CA hackers CA pin also? and other musings (Re: PKI "fixes" that don't fix PKI (part III))

Andy Steingruebl andy at steingruebl.com
Mon Sep 12 13:32:35 EDT 2011


On Sun, Sep 11, 2011 at 8:37 AM, Douglas Huff <dhuff at jrbobdobbs.org> wrote:
>
> On Sep 11, 2011, at 9:25 AM, Thierry Moreau wrote:
>>
>> E.g. http://datatracker.ietf.org/wg/dane/ (DNS-based Authentication of Named Entities (dane))
>
> Which makes a huge assumption about DNS SEC that is just not realistic. Namely, the one I just mentioned, that end clients would actually be validating. Meaning that the MITM I mentioned becomes hilariously effective in the vast majority of scenarios where the clients themselves are not doing the validating. Giving a nice illusion of additional verification with no substance.

It doesn't make that assumption at all, and way too many cycles were
spent going over this problem repeatedly.  All of the discussion has
essentially required the client to end-to-end verify the answer, which
given the amount of network breakage in the world that makes this
difficult, is a serious wrinkle in attempts to deploy solutions like
this.



More information about the cryptography mailing list