Morris et al. give an example in which a system that outputs two wrong words for each input word gets a WER of 200 percent, although both systems in their example communicate exactly zero information.

Checked against its source on October 17, 2026

In the source's own words

“A similar system which outputs two wrong words for each input word has S=N1, H=D=0, I = N 1, and WER = (N1+N1)/N1 = 200%”

From WER and RIL to MER and WIL: Improved Evaluation Measures for Connected Speech Recognition, Andrew Cameron Morris et al., Proceedings of Interspeech 2004, ISCA, 2004, Section 2, worked examples.

What it measures: The authors compare a system that gets every word wrong, which scores 100 percent, with a system that inserts two wrong words per input word.
What it does not show: The example is a theoretical one, written to show that WER can exceed 100 percent and that it does not measure information. It is not a measurement of a real system.