A cardinal role of selective visual attention is to serve our action by selecting all the relevant information. One task that has been applied extensively to explore attention is visual search for a target among distractors. Given the extensive practice of human observers with visual search, and its ecological relevance, one could expect a high level of efficiency when performing visual search tasks. Indeed, a prominent in the literature Ideal Observer model (Najemnik & Geisler, 2005, 2008) suggests, that human visual system is extremely efficient, in that every eye movement during visual search is executed to the locations that could be expected to yield maximum information; similarly to the Ideal Observer, humans require the minimum number of eye moments possible to find the target (Najemnik & Geisler, 2005, 2008). The present programme of research tests the prediction of the Ideal Observer model (Najemnik & Geisler, 2005, 2008) against a simpler, but similarly effective stochastic selection model (Clarke, Green, Chantler, & Hunt, 2016), by examining human visual search strategies under normal viewing conditions, and conditions that simulate visual field deficit. Across nine experiments, I observed strikingly inefficient search behaviour that speaks against the assumptions of the Ideal Observer model. Although on the surface these cumulative results appear to be in line with the random process of fixation selection (Clarke et al., 2016), such conclusion appears to be valid only for the observed group results. The individual observer's data that was carefully documented in each of the experiments did not allow such a conclusion. The individual observers' data showed full spectrum of search strategies, with some observers being extremely efficient, some being average searchers, and others applying a very inefficient strategy. The large individual differences between participants suggest that the fixation selection process is neither optimal nor random, but rather idiosyncratic.