The growing popularity of methodologies that turn "to the wild" for real world data creates new ethical issues for the HCI community. For investigations questioning interactions in public or transient spaces, crowd interaction, or natural behaviour, uncontrolled and uninfluenced (by the experimenter) experiences represent the ideal evaluation environment. We argue that covert research can be completed rigorously and ethically to expand our knowledge of ubiquitous technologies. Our approach, which we call Deep Cover HCI, utilises technology-supported observation in public spaces to stage completely undisturbed experiences for evaluation. We complete studies without informed consent and without intervention from an experimenter in order to gain new insights into how people use technology in public settings. We argue there is clear value in this approach, reflect on the ethical issues of such investigations, and describe our ethical guidelines for completing Deep Cover HCI Research.