I've been thinking more about
[draft-strad-trans-redaction-00](https://tools.ietf.org/html/draft-strad-trans-redaction-00)
and I am now leaning against adoption by the WG.  This working group
is about transparency not translucency.  The goals stated in [What is
Certificate Transparency?](http://www.certificate-transparency.org/what-is-ct)
have not changed since first published in 2013 (according to the
WayBackMachine).  They are:

* Make it impossible (or at least very difficult) for a CA to issue a
SSL certificate for a domain without the certificate being visible to
the owner of that domain.
* Provide an open auditing and monitoring system that lets any domain
owner or CA determine whether certificates have been mistakenly or
maliciously issued.
* Protect users (as much as possible) from being duped by certificates
that were mistakenly or maliciously issued.

I believe that a log inclusion proof should provide assurance that
these goals are met.  It should mean that the certificate was included
in the log intact and that it is available from the log for review.

This is important when looking at things like
NSRequiresCertificateTransparency in Apple App Transport Security and
the [Expect-CT draft](https://tools.ietf.org/html/draft-stark-expect-ct).
These allow a server or application owner to specify that CT is
required for connections.
One can also imagine a future draft to update the [TLSA
RFC](https://tools.ietf.org/html/rfc6698) to add CT as a flag in
certificate usage so that a site could declare via DNS to only trust a
certificate that has an inclusion proof (or SCT).  Adding some
"translucent" option means that these become far less valuable policy
directives.

I think the primary driver for the redaction draft has been the fact
that one major web browser vendor has suggested that they may set a
policy to require CT by default. The IETF has been very successful by
avoiding policy and focusing on technical protocols.  For example, the
DNS RFCs discuss how DNS works not how it is deployed.  Even within
the ICANN/PTI rooted DNS, there are policy differences.  The dk TLD
requires that all DNS servers for registered domains under dk be
publicly accessible and meet certain technical requirements while the
xyz domain accepts name servers that have IP addresses in address
blocks that are not global (as defined in [RFC
6890](https://tools.ietf.org/html/rfc6890), e.g. 10.0.0.0/8).  It is
also possible to use the DNS protocol on standalone networks not
interconnected to other networks.  The same should be true for CT --
the protocol and expectations should be well defined and usable in a
variety of ways and policy should be separate from the RFCs.

To be clear, I do not believe that lack of support for adoption of the
redaction draft should imply support for a policy of requiring CT by
default.  That is a different discussion for a different list.  The
question at hand is watering down CT and that is something I do not
think should happen.

Thanks,
Peter

_______________________________________________
Trans mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/trans

Reply via email to