From what I'm reading on Bruce Schneier's blog post, man in the middle attacks (MITM) are commonplace for https connections. The agents of the attack are your ISP and corporate IT folks.
How can I configure chrome such that when I talk to some website over https connections, I can know that the encryption that is occurring is secure against MITM attacks coming from nodes in the network or at least have chrome warn me when someone is trying to do a MITM attack?
To avoid the type of "attack" described in that blog post, just don't use "cloud"-based browsers. A cloud-based browser doesn't directly communicate with websites; instead, it relies on a sort of proxy server (operated by Nokia in this case) to access websites on its behalf. The proxy establishes the SSL connection to the site, and receives and decrypts the pages before sending them back to you.
To protect against other kinds of SSL MitM attacks, you'll need to verify that the site's certificate is genuine: that it really belongs to the site, and not to an attacker masquerading as the site. In principle, the fact that the certificate is signed by a trusted CA should provide enough assurance that they've verified the certificate's authenticity. In practice, CAs make mistakes sometimes, and attackers have been able to obtain signed, "authentic" certificates for high-profile domains.
Perspectives can alert you when a site is presenting you a different certificate than it presents to others, which can be a sign that you're connected to a MitM attacker instead of to the real server. However, this works by asking the Perspectives notary servers for information about each domain that you visit using HTTPS, so those servers can learn which domains you're visiting.
Certificate Patrol can alert you when a site is presenting you a different certificate now than it did in the past, which can also indicate a MitM attack. This avoids revealing what domains you're visiting to a third party, but it's also more prone to false positives since sites do change their certificates from time to time.
In a corporate environment, your IT department may be acting as a private CA and (internally) creating certificates that appear to be for outside domains. In this situation, your browser is configured (by the IT department) to trust certificates from the corporate CA. When you try to connect to an outside site, your connection is routed to a corporate proxy, which presents one of those internal fake certificates that claims to be the site you're trying to visit. Your browser thinks it's communicating with the real site, but it's actually communicating with the proxy.
This sort of attack is only possible if the attacker is able to configure your browser to trust a CA that it wouldn't by default. The usual example is a corporate network with the IT department as the "attacker", but it could also happen on a personal computer infected by malware. (Of course, if there's malware on your computer, it may be able to eavesdrop on your browsing more directly, with no need for MitM.)
Perspectives would detect this sort of attack, since you're seeing an internal fake corporate-issued certificate while everyone outside your corporate network sees the real one. Certificate Patrol wouldn't, unless your computer moves back and forth between the corporate network and an outside network, and you try to visit the same site from both.
