Re: [TLS] [therightkey] Fwd: Improving EV Certificate Security

["Ryan Sleevi" <[e-mail]>]

On Wed, September 25, 2013 10:15 pm, Joseph Bonneau wrote:
>  I'd like some elaboration on the plan for step 6, creating a whitelist of
>  valid EV certificates without an SCT. How is this going to be achieved?
>  Also, if we could do this, why not do it for all certificates and
>  bootstrap
>  CT that way? Are the parameters of EV special for this (fewer certs,
>  better
>  records, etc.)?

Fewer certs, by many orders of magnitude, such that whitelisting is a
reasonable approach.

>
>  An alternate approach to a whitelist is to require SCTs for certs with a
>  "not before" validity period after time T (presumably this requirement
>  kicksn in around time T). With a stolen/compromised EV CA key you could
>  still issue a fraudulent cert and backdate it, so you'd have to more
>  strictly enforce the limits on validity periods for EV certs which I
>  believe are 27 months in the CA/Browser forum guidelines and 39 months in
>  the EV code-signing cert proposal. Of course this isn't attractive in that
>  it means years before you really have protection against fraudulent EV
>  certs. Has this approach been considered?
>
>  Joe

As you note, with a stolen/compromised EV CA key, you can still backdate
certs. The whitelist approach balances the tradeoffs and brings the delta
for practical checking on the client side down to a reasonable value and
with acceptable (and negligible) performance overheads, which decrease
over time.