To address the lack of standardized benchmarking in speech brain-computer interfaces (BCIs), researchers Jayalath, Ballyk, and Jones propose a new information-theoretic metric called open-vocabulary mutual information (OVMI). Current evaluation methods, such as Word Error Rate (WER), often yield incomparable results because they rely on heterogeneous datasets, recording techniques, and restricted vocabularies. The authors argue these traditional metrics can overestimate a system's utility by only measuring accuracy within a limited, pre-defined word set. In contrast, OVMI quantifies the information conveyed by a decoder relative to a broader reference distribution of potential intended words, allowing for a direct comparison between systems using different vocabularies and communication constraints. By applying this metric, the study exposes fundamental trade-offs between vocabulary breadth and decoding accuracy. The researchers demonstrate that optimizing vocabulary selection specifically to maximize OVMI can lead to a relative accuracy improvement of up to 16.3% across three distinct speech domains. Ultimately, OVMI provides a principled framework for the speech BCI community to evaluate progress, design more effective vocabularies, and compare heterogeneous decoding systems on a unified communication scale.