search

IAB-PrivSec-program / draft-iab-privsec-confidentiality-mitigations

The Internet Draft recording the program's draft on mitigating confidentiality threats
0 stars 3 forks source link

Don't mislead about hop-by-hop #28

Open martinthomson opened 8 years ago

martinthomson commented 8 years ago

This statement is misleading:

Hop-to-hop security mechanisms may be susceptible to downgrade attacks (e.g., STARTTLS- secured SMTP has been downgraded by intermediate network nodes [WaPo- STARTTLS]) in which case end-to-end mechanisms are advised.

It appears in a discussion about the role of hop-by-hop security. A valuable discussion.

However, it appears to make a statement about hop-by-hop security that is inherently false. This is a failing with the actors in the system not having an expectation that security is used. They are willing to accept unsecured communication. As such, it's a statement about opportunistic security.

This is a statement that more rightly belongs in the section on unauthenticated encryption (which might be better repurposed to discussion opportunistic security in that case, see #23).

If there is a statement to be made here, it is that downgrade in a hop-by-hop system is inherently harder to a) retrofit, and b) detect. Since the status of hops that an actor doesn't participate in is unknowable, only systems that insist on end-to-end security actually get any verifiable guarantees.

martinthomson commented 8 years ago

Three things to capture:

  1. it's hard to retrofit security on a hop-by-hop system
  2. hop-by-hop makes security posture of remote hops difficult
  3. use end-to-end

Remove or move the oppsec into a separate treatment.