"Usefulness" does not have enough content to be operationally useful. After all, from an academic perspective, a comprehensive record of all the attestable meanings, pronunciations, usages of all words is "useful". Such a thing is not necessarily terribly "useful" to most users or potential users. Perhaps we can use customization techniques to allow many (all?) kinds of users to have a "useful" experience at Wiktionary, but not today. Our entry design and our inclusion criteria necessarily involve considerations that favor "usefulness" for one class of user over another. Given the low priority we give to non-contributing users, I suppose we should just do what is easy and make decisions based solely on what we want. That should work as long as the resources are here for us, which may be for much of my lifetime.
"Usefulness" could be better operationalized if we actually had a few model users in mind. Language learners (many classes), native-speaker English writers, native-speaker English readers, native-speakers hearing different pronunciations of words they may already know. New users (with and without web familiarity). Infrequent repeat users. Would-be contributors. Impulsive teens. Local boosters. Advocates. Members of subcultures.
For almost all of these users, it is hard to see why they would need anything other than to be able to find "broken English" in a usage example under "broken" or "English" to know that it was acceptable English. If someone is starting from another language and looking for the right English expression
