Scholars have long measured the impact of a paper by counting the number of times other scientific articles cite it. Researchers have just detected a new kind of citation fraud.