Kristine Nowak and Christian Rauh of the University of Connecticut wrote an interesting paper where they describe an experiment to test how people perceive the trustworthiness of avatars. Volunteers where chatting and they were represented as avatars with different levels of gender specific representation. The androgynous-looking avatars were perceived as being less credible. The authors make a leap to trustworthiness and argue that human-like avatars are important for representing trustworthy avatars. Doing experiments with just text chat and observing an amazing level of cooperation, I wonder if the authors make too much of a leap. It might be different if people can represent themselves by choosing a certain type of avatar. But here the volunteers might understand that avatars were not necessarily relate to the participants own choices.