Batesian mimics are deliciously palatable species that gain protection from a predator by resembling a defended or unpalatable model. Theory predicts that mimics that closely resemble their model should have the greatest advantage, while inaccurate mimics should be recognised and attacked by predators. This is all fine and good, but in reality, we find that mimics are highly variable, some are excellent in mimicking their model and others are rubbish at it. There are many different hypotheses that try to explain the persistence of inaccurate mimics, and we have a research project at Macquarie University that tests these ideas using ant mimicking spiders. In this seminar, I will illustrate the range of mimic fidelity in ant mimicking spiders and present the evaluation of some of the common hypotheses. If time and technology permits we will even play an online game.