I’m trying to understand this better, but what I don’t get is when/how cache invalidation happens.

The article states that `apollo-client` will cache e.g. any query trees that match `author(id: 5)`, but then it says:

what if you fetch some other query and realize that the author with an ID of 5 has changed his/her name

How would this query know to do a real fetch and not simply retrieve (stale data) from cache? Are we talking about `forceFetch`? If so, what guidelines can we follow to determine when to force and when not to?