[Cryptography] Two questions of security

Natanael natanael.l at gmail.com
Tue Mar 8 04:29:41 EST 2016


Den 8 mar 2016 06:14 skrev "Phillip Hallam-Baker" <phill at hallambaker.com>:
>
> I just released a new set of Mesh specs, I am now working on getting
> the site set up with a decent explanation of what is going on. In
> particular, I am trying to compress the following argument to a couple
> of pithy one liners. Was wondering if people might have ideas:

I've got plenty of ideas  :)

It seems like the more I read about your idea of the Mesh, the more they
overlap with my ideas of coordinating personal devices and creating digital
"entity declarations", digital representations of identity (of all kinds),
and hooking in permissioned cryptographically authenticated API:s to it. I
haven't written down most of it yet, but you can look at my blog. The
currently most relevant post is the latest one that's covering my key
courier idea.
https://roamingaroundatrandom.wordpress.com

I should also write down my case for why everybody should have a home
server (one part of it is to have it be an IoT API firewall). And the
Raspberry Pi 3 with its integrated WiFi and Bluetooth 4.1 is very well
timed, it would be an excellent prototyping tool, using it as an
always-online Mesh node could be useful (including for keeping an updated
cache of all your permissions and Mesh configurations, as a proxy between
networks, etc).

Now, for my way of re-expressing the intent of the Mesh;

> Traditional access control approaches were created in the era when the
> problem was how to divide the computing resources of one machine
> between many people. Today our typical security problem is the exact
> opposite - one user has many, computing devices. And as every consumer
> good 'becomes intelligent' or at least adds networking capabilities,
> the problem gets worse.

We used build secure personal sandboxes out of big shared sandboxes on
shared machines.

Now we have so many networked machines with unique capabilities that
sandboxes are just not enough.

> As a result, we end up authenticating the wrong thing. Instead of
> authenticating the users, we need to authenticate machines.
>
> The Mesh reduces this problem to two questions:
>
> 1) Do I want to control this device? If so, for what purposes?
>
> 2) Is this device under my control?
>
> I buy a new device, it might be a laptop, a mobile phone, a new car
> but in this case it is a garage door opener.
>
> I tell the garage door opener it is under my control, the only purpose
> it supports is opening the door. The door opener confirms that it is
> under my control. Now I can press the open door button in my car and a
> cryptographically secure challenge response protocol using real
> cryptography rather than junk that was broken 20 years ago takes
> place.

The Mesh can let us securely define what each device can do (their
capabilities), who or what they can do it for (their permissions) and then
tell each device what they can ask of others (making it *your* mesh
network).

This way we can make our devices communicate securely in the exact way we
want them to.

We could for example securely tell our garage doors to only accept commands
to open up from our cell phones and car computers. Or a friend's phone, its
your choice. And this extends to all our devices.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20160308/e4581028/attachment.html>


More information about the cryptography mailing list