Note how your statement does not hold in the real world.
Email being transmitted by 3rd parties is not different from voice calls being transmitted by 3rd parties. Yes, you are trusting a provider, with the expectation that your provider will send the data where you've asked it to send the data, and nowhere else. This is still true whether you're talking postal service, landline voice calls, SMS, cellular voice calls, Skype, etc.
From the parent of my original comment:
Almost every SMTP server out there today requires authentication ... That is the very definition of trying to keep things private.
Do you disagree with me? Do you believe that SMTP authentication contributes to privacy and not authenticity?
If I am using them to communicate with a 3rd party, I have a reasonable expectation of privacy between myself and that 3rd party. You would most certainly need a warrant to turn around and try to get access to a message stored on their servers.
On top of that, quite a bit of email today doesn't even touch SMTP. If I'm sending an email from one GMail user to another GMail user, I'm pretty sure it is just shuffled around on Google's internal servers. And, of course, I'm connecting to Google using SSL, an encrypted connection.
How does that not scream private?
I cut out the SSL and TLS because I don't disagree; SSL and TLS certainly says "privacy". I wanted to specifically address the notion that "authentication =~ privacy"
Why can't it be both?
Why should storing something on a third party's system obviate the need for a warrant? If I rent real property and use it to store my papers, does that mean the government should be entitled to seize them without a warrant because I'm storing them on the premises of a third party? If not, what's the difference?
In other news: you can't invoke the 4th amendment if you stash boxes of weed at a friend's house and he hands them over to the government when asked.
There are privacy policies and terms of service that cover what a service provider will share with whom and under what conditions. This represents an agreement between the user and the service.
However, that is wholly different from the rights held by the government (IRS) to compel the service provider to provide information at the government's demand.
Your analogy with the "boxes of weed" are similarly misguided and completely unrelated. The more relevant analogy would be that the government forced your friend's landlord to unlock his door without a warrant, entered, discovered the boxes of weed with your name on it, then arrested you.
And, to put your analogy back into the subject e-mail context, if you e-mailed something incriminating to your friend and your friend opted to turn you in to the authorities, that is entirely different from the government gaining access to your e-mail of its own volition, and without a warrant.
Mail theft was a pretty serious crime in USA I am led to believe, regardless of whether it is sent in encrypted form or not...
To prove to yourself that SMTP is primarily concerned about authenticity over privacy, just try setting up your own mail server. Ensuring your mail is not blocked and/or marked as spam is an involved process of establishing multiple checkpoints re-confirming your identity. DKIM, SPF, domainkeys- each intended to establish that you sent this (unencrypted) message, and not someone pretending to be you.
Google won't hand you a packet capture from eth0 on smtp.gmail.com because you asked nicely. You would have to coerce an insider (or exploit your way in).
The only situation in which your statement applies is if you're abusing a position of trust as a network administrator. While it's true that this is possible, it's also possible for someone to break into your (postal) mailbox. You still have an expectation of privacy.