Re: [Coin] challenges for COIN

["Aaron Falk" <[e-mail]>]

On 1 Mar 2019, at 11:02, Carsten Bormann wrote:

> On Mar 1, 2019, at 13:25, Dirk Kutscher <[e-mail]> wrote:
>>
>> Let’s please forget the notion of “managed networks where privacy 
>> and security is not a big problem”.
>
> +1 (with a very big ONE).
>
> The usual approach here, however, is
>
> (1) Somebody dreams up something under that flawed assumption
> (2) This something gets deployed
> (3) Security breaches start to get noticed
> (4) Some very weird usage restrictions are dreamt up to “manage” 
> these security breaches
> (5) Insecure something plus the straightjacket from (4) becomes the 
> new normal
> (6) This becomes the way “real networks are run” by “real 
> engineers”
> (7) All this cannot be questioned any longer and becomes a drag on any 
> further development.
>
> Examples are galore.  I’d offer RFC 6092…
>
> The flip side of the coin [sic] is that, for research, it is usually 
> quite useful to ignore some requirements, and that may include 
> security.
>


I’m happy to see attention onto this topic.  Perhaps what is needed 
for compute to work ‘in the network’ is to have an architecture 
where (at least) one of the endpoints can delegate trust to elements in 
‘the network’ to act on their behalf and audit their actions.   IMO, 
addressing this challenge will make this work (much more) valuable. 
YMMV, of course.

--aaron