There are so many types of caching too. Most developers do at least one type without paying any attention to it: memoization. An example of this is to get a value outside of a loop that is used within it as opposed to within it. Jeff mentioned the caching of static (or at least relatively static) data which is another extremely common use.

Caching should always be considered when developing applications. In this forum many of us might be considering maintaining consistency with the current state of the database i.e. what if it changes and we don't have the latest. Of course, getting the latest might be exactly the wrong thing to do. This is where the classic "it depends" applies: sometimes you want to ensure that the state of the data is the same for the whole of a particular context. That context might be the whole of the duration of a system is running (process), a service remaining active, an application domain's lifetime (AppDomain in .NET), an operation, a thread or, perhaps, a transaction.

This is an example of why experience counts. Also, it is an example of why some people will always be better at software development than others. There will always be some developers who will never consider these types of questions. It may matter. Or it may not.

Those of us who do consider these facets of software development have a responsibility to the projects we work on to highlight when and where we need considerations such as caching to be made.