Re: [Uta] opportunistic keying / encryption considered of dubious value

[Daniel Kahn Gillmor <[e-mail]>]

On 03/15/2014 03:45 PM, Keith Moore wrote:
> On 03/15/2014 03:07 PM, Paul Hoffman wrote:
>> On Mar 15, 2014, at 11:43 AM, Watson Ladd <[e-mail]> wrote:
>>
>>> All DANE/TOFU/latching imply a commitment by both sides to maintain
>>> and manage identity.
>> A big +1 to Watson's simple explanation. DANE/TOFU/latching all
>> specify that if the initiator gets a key that is wrong, they will
>> prevent communication in order to protect the initiator.
 [...]
> I also think that if you're running a server that uses encryption, you
> should be regularly and frequently monitoring your certs, DNS records,
> etc. for validity and correctness.

Just to clarify, the work you're describing *is* the "commitment ... to
maintain and manage identity" that Watson and Paul are identifying as a
major barrier that has kept systems administrators from deploying strong
crypto.

This maintenance work is extra work that does not have to be done by a
server operator who only runs cleartext services.

Opportunistic keying (OK) also doesn't require this extra work.

OK doesn't replace or supplant encryption with properly authenticated
endpoints.  It replaces and supplants cleartext communications, while
placing no additional demands on system administrators.

Those administrators still really should move to proper strong crypto,
and software developers should produce tools with a sane UI/UX that can
minimize the hassle involved with ongoing cryptographic credential
maintenance work, to reduce the barrier that still remains.

	--dkg