We have been testing second view (visiting one page and then measuring the next page) for a couple of years to better simulate a hot cache.
I think we should focus on two longer user journeys instead, that way we can get more data on page views with more or less hot caches and will better measure the Wikipedia rabbit hole.