RFC: give_way:type

You could, but you’d probably have to come up with some tag for distinguishing explicit and implicit yielding positions. After all, a visual data consumer would most likely mark the location with a give-way symbol, suggesting that the user should look out for something more explicit. That would be misleading if the yielding is inferred from something more obscure.

My earlier suggestion was to use something like marked=no and/or signed=no, reinforcing the global norm that yielding and stopping positions are normally signposted and/or marked, while acknowledging that these aren’t quite implicit=yes. I think you weren’t in favor of the Boolean tag approach because it requires multiple keys, but on the other hand, we should learn the lesson of crossing=* that classification systems eventually become constraining.