What we see in these models is that even perfectly rational—albeit simple—agents who learn from others in their social network can fail to form true beliefs about the world, even when more than adequate evidence is available. In other words, individually rational agents can form groups that are not rational at all.
