It’s a classic tradeoff between scalability and some notion of precision. It sounds like your idea of a “decent” presentation relies on the user being able to make certain inferences about the service’s operation based on a certain line treatment. I think this is still possible as long as the map communicates a certain fuzziness to the user.
When I developed a legend for OSM Americana, I had to come up with a label for expressway=yes. Expressway is an infamously poorly defined term, yet transportation maps conventionally distinguish it as a separate class on the legend, with a curt explanation of what to expect as a motorist. That’s what we did too:
| Tag | Label |
|---|---|
highway=motorway |
Freeway (controlled access, divided) |
expressway=yes |
Expressway (limited access, divided) |
Not every stretch of freeway has fully controlled access and not every stretch is divided. Not every stretch of expressway has limited access and many are undivided. We tag highway=motorway and expressway=yes based on weighing various factors that require a long wiki page to explain. Yet Americana includes these curt generalizations, figuring that they’re good enough for most users. This is only possible because the tags’ definitions are not arbitrary decisions by a mapping committee, but rather a distillation of common traffic engineering definitions.
So I don’t think subjectivity or inconsistency is fatal to your idea. It just means we need to manage expectations about consistency and decide how much leeway to give mappers.
It’s up to you, but I don’t think this is totally hopeless. We could embrace the lack of objectivity and just tag whether the operator identifies the route as BRT. A global renderer may not be able to make a confident statement about the specific reasons why a particular route is highlighted more prominently, but it will be consistent with other map depictions of that route.