http://techcrunch.com/2012/07/29/surprisingly-good-evidence-...
I could know someone is named John Smith for example, and that still wouldn't help me pierce his anonymity unless there were other distinguishing features (picture, phone number, etc.).
So majority people are actually nice even when anonymous. And penny arcade is a comic site, not a dispenser of truth.
I expect it was about Paul Feldman -- it was a very interesting story -- the guy was an MIT-trained economist who, in frustration at getting nowhere at a certain point in his career, gave it up to deliver bagels, etc. to companies every day and he collected data on people's behavior when it came to leaving payments.
The following NY Times Magazine article is by the Freakonomics duo about him (the story made it into the Freakonomics book):
http://www.nytimes.com/2004/06/06/magazine/what-the-bagel-ma...
PDF of the same article from one of the authors (i.e. a non-nytimes.com link):
http://pricetheory.uchicago.edu/levitt/Papers/WhatTheBagelMa...
Although your anecdote is just as anecdoty as the comic author's anecdote...
This has actually been studied academically (the "online disinhibition effect"). Anonymity definitely seems to correlate with poor behavior, even if it's not a causal relationship.