Inscription is the term generally used in English by UNESCO - it is all over their website about the World Heritage List. So in this context it seems like an understandable choice.

I struggle with the order of namespaces/qualifiers so I hesitate to answer. If anyone has a link to a discussion of these issues (not for heritage but in general) maybe that would help. The “namespace” page on the wiki doesn’t really help. For the moment I would read it as “criteria=” being the basic tag and the cadw prefix qualifying/limiting/restricting that tag. Rather as disused:building qualifies the basic building= tag. But as I say, I am very much open to correction or further guidance on that.

I assume these grades indicate a kind of ranking of importance or level of protection. Not all heritage registration schemes have this kind of ranking. For example my understanding of the Spanish “BIC” registry is that inclusion in itself implies the highest grade of protection, so such a tag would be redundant. So maybe it is not documented simply because few mappers worldwide have seen the need.

I agree that “criteria” may not be the ideal choice of word as it seems to be used more to indicate a category or type than as a reason for listing. Again though it may reflect UNESCO influence - looking at the criteria for listing at http://whc.unesco.org/en/criteria, they also seem to blur this line - many of them are in effect saying that the criterion is to be an especially important example of X, where X is a type or category such as landscape, building etc.

I would say it is more that a multinational group of mappers, most of whom are tagging or documenting in a language that is not their first language, are trying to find a balance between the specifics of their local schemes and a form of tagging that is broadly comparable worldwide. In that context I think the page as a whole is reasonably coherent even if there are points of detail that could be improved. It certainly seems reasonable for you to add UK-specific details if they are not currently documented.