I understand well what happens when SMTP server, say Postfix contacts remote SMTP server to send or receive mail. The connection could (and should) be encrypted, certain ports must be used and the certificate must be valid and accepted by all parties. But when it comes to the local e-mail delivery for me it's a grey area. Ok, let's say if we're talking about receiving mail and Postfix already got it from the remote server. Then we have MDA (like Courier) that takes the mail from Postfix and gives it to a MUA (like Thunderbird Mail, for example). I'm talking about one physical machine and one user session. What the purpose of the encryption if it all happens inside and no third party could be present? And if it's the case of the same computer and the same user, then does the validity of the cert play any role for e-mail delivery process from Postfix via Courier to Thunderbird Mail?
... I am not an expert on this so any correction or clarification is welcome.
Most often Postfix and Thunderbird(s) are not running on the same machine, because rarely a mail user runs its own mail server. The traffic between the Posfitx-machine and the Thunderbird-machine(s) will be readable if you don't encrypt it.
I did an experiment right now that you may replicate on your (virtual) machines.
1) On machine 172.16.10.10 run $> nc -l -p 12312 2) On machine 172.16.10.11 run wireshark 3) On machine 172.16.10.12 run $> nc 172.16.10.10 12312 [type] Hello [Enter]
Now go to Wireshark, you should be able to find the comunication between the two machines 172.16.10.10 and 172.16.10.12 even if that comunication was not intended to be for 172.16.10.11 ! But 172.16.10.11 was able to sniff it and read it... The same, in principle can happen to unencrypted mail communications.