In many environments, tournaments can elicit more effort from workers, except perhaps when workers can sabotage each other. Because it is hard to separate effort, ability and output in many real workplace settings, the empirical evidence on the incentive effect of tournaments is thin. There is even less evidence on the impact of sabotage because real world acts of sabotage are often subtle manifestations of subjective peer evaluation or "office politics." We discuss a real effort experiment in which effort, quality adjusted output and office politics are compared under piece rates and tournaments. Our results suggest that tournaments increase effort only in the absence of office politics. Competitors are more likely to sabotage each other in tournaments and, as a result, workers actually provide less effort simply because they expect to be the victims of sabotage. Adjusting output for quality with the rating of an independent auditor shrinks the incentive effect of the tournament even further since output tends to become more slipshod.