[Cryptography] On the deployment of client-side certs

ianG iang at iang.org
Mon Nov 14 22:07:42 EST 2016


On 14/11/2016 17:45, Ray Dillinger wrote:
> [client-certs rule but aren't used...]

> How can we change that?

That is the question.

> What can we do to make it easier to do, provide
> a transition path toward, and get pinning, certificate checking, and
> revocation list checking integrated on hosts and cert generation/use
> integrated into clients?  And make it simple and easy for consumers to
> share their certs from multiple devices, reliably remove them from
> devices before loan or sale, and revoke-and-replace whenever one of
> their devices with a copy of their cert gets stolen?

I think the headline problem is basically down to the poor GUI, the poor 
management of client certs.  Which is down to the lack of interest in 
the browser vendors developing it, or experimenting with alternates such 
as self-signed certs.  Which is down to the CAs being terrified of 
anyone futzing with their business model.

It's actually quite easy.  I programmed a fully client cert site in PHP 
about 6 years back, for the experience.  In the end I concluded it was 
as easy as programming up with passwords, and a lot more friendly and 
usable for the users.  It is of course much more secure at the 
cryptographic level.

No more lost passwords!  Which seems to be about half the user support 
costs and a big security risk eliminated.  Of course what this does is 
shift the burden across to the users needing to get/keep certs, so the 
work becomes much more beneficial if one has several sites to program 
up, and share the bounty.

Hence the strategy that emerged was to pull the users on to one key site 
that they all had to go through with client certs to get access.  Then 
the rest came for free.

iang


More information about the cryptography mailing list