[saag] Re: On path Active Attackers, and Meddlers in the M iddle

On Sun, Jan 5, 2025 at 2:06 PM Michael Richardson <[email protected]>
wrote:

>
> Carsten Bormann <[email protected]> wrote:
>     > I remember doing a quick informal survey of the usage of related
> terms
>     > in research papers at some time in the late 2010s, and there was a
>     > clear consensus for MITM, and, incredibly, still for the
>     > “man-in-the-middle” expansion.
>
>     > For what RFC 4949 describes as MITM, we should stick with MITM, but
>     > maybe update with the better expansion.  (For other active on-path
>     > attacks, and specifically for describing the capability instead of
> the
>     > attack, we can use the more general term.)
>
> Yes, so MITM properly is Active On-Path Attacker.
> The idea is that MITM is a CNAME for Active On-Path Attacker.
>

While I agree with you that we should mostly just stop saying "MITM",
I don't think that this is strictly correct, for two reasons:

1. Classically, a MITM attacker impersonates Alice to Bob and Bob to Alice,
(hence the DH example), though I agree that 4949 is kind of vague on this
point. However, not all on-path impersonation attacks involve impersonation
in both directions; many involve impersonating  (for instance) a server to
the
client without connecting to the server at all.

2. There are active on-path attacks that don't involve any impersonation
at all. For example, consider the case of TLS 1.3 0-RTT replay, where
the attacker is just retransmitting valid data from the client to the
server.

With that said, I like the term "Active On-Path Attacker" for capabilities.
I just think we ought to use more precise names for the attacks themselves
(e.g., "server impersonation") if we do that, I doubt we will need to use
the
term MITM much if it all.

-Ekr

_______________________________________________
saag mailing list -- [email protected]
To unsubscribe send an email to [email protected]