>
One author exposed how becoming an exit node for Tor allowed all the traffic on the Tor network to pass right through your machine. Becoming an exit node was the same as performing a Man-In-The-Middle attack.This is de-contextualised scaremongering.
What the poster is referring to is that when you leave the Tor network, the connection is as it would have been before. This is by design. So if you were not using TLS, then your traffic could be read, as is the case with ALL http etc. Simple. Use SSL/TLS/SSH/ etc
Do a trace route to google. All those intermediate parties are capable of the same thing.
But if you use TLS the attack is useless.
Tor provides one property, Anonymity, and it does this incredibly well. Anonymity and Privacy are related by distinct properties. Obviously if you send traffic to a site and sign off with your name, Tor can't help you be anonymous there. Tor cannot prevent misuse & ignorance.
As for this 'MITM' attack, Tor's design is such that you do not have to trust the exit nodes for it to work.
As for the comparison with I2P, I don't know much about it, but I support any FOSS project that aims to provide new types of anonymity. As I understand it the problem with I2p at the moment is that there is 1 exit node facing the regular net. It's not entirely clear how I2P evades what you consider 'the problem with Tor' when connecting with the regular net.
EDIT: Yep, did a bit of research and i2p is subject to the same sort of "attack" you describe "Like Tor, I2P does not magically encrypt the Internet. You are vulnerable to snooping by the outproxy operators." http://www.i2p2.de/faq.html#outproxy