This is a question that comes up once in awhile, and it has a fairly nuanced answer. One of the key factors we consider when building the Identity Graph is how to balance the coverage vs accuracy of our data. For example, we could limit our data to only that data we know to be 100% accurate, but the coverage would drop significantly and we would lose a lot of great data that is still accurate and useful. On the flip side, if we wanted to append a name to every phone number in the US we could do that, but the accuracy of that data would plummet.
At the end of the day, the key charter of our Data Services team is to strike the right balance of coverage and accuracy. We do this by understanding our customer use cases and what data we realistically have available, and by testing various versions of our entity resolution algorithm. The end result is that our accuracy will actually vary over time based on what needs our customers have and what data we are using.