Plagiarism detectors are a crutch, and a problem

It is one of several firms offering software systems that apply ‘black box’ algorithms to produce a score that purports to show how well one text matches others. A system might fail to find plagiarism if the source of the plagiarized text has not been digitized, contains spelling errors or is otherwise not available to the software system. If the software reports a low number, the person assessing the paper might skip over obvious indicators of plagiarism such as style shifts, misspellings, font changes or underlined words that suggest the text has been pasted from Wikipedia.

Source: www.nature.com