Hi Michael,
so in fact you have three characteristics, and as I said above the locks are based on the data region described by the static filter of the query. The locks are not record based.
So if the data regions descibed by the filter of query A and B have an overlap one will get a lock conflict. This is the feature.
The documentation contains information about the lock concept. In transaction RSPLSE you can see the last lock conflict and check the selections.
Regards,
Gregor