Competition in academia, epitomised by the pervasive aphorism “publish or perish”, affects researcher behaviour: it increases productivity in terms of publication output, but also leads to biases in the scientific literature. This is because academics are not being assessed on whether their research leads to scientific truth, but on their putative ‘impact’. Unfortunately, this ‘impact’ is measured by various proxies that ostensibly reflect the importance of academics’ contributions to the scientific literature, but which are fundamentally flawed. Here, I will synthesise research that shows that, as long as selection in academia rewards impact rather than integrity, academic selection has and will continue to result in: bias in the academic literature; irreproducibility of results and over-inflation of effect sizes; weakening of the public trust in science; the attrition of women in academia; and, in extreme cases, public and/or personal harm. From this evidence, I will ask whether, as an early-career researcher, there is any incentive to develop new systems or whether the ‘safer bet’ results in the continued use of existing models and the potential loss of creativity in approaching research questions.