I was trying to help someone copy settings (email, web bookmarks, etc) from one Windows machine to another and they asked "Why is it so complicated to copy my settings?" The true answer, at least in my mind, is some combination of: each program handles its settings information differently, Windows doesn't provide a standard mechanism for transferring settings, and Windows tends to store pieces of a program in a zillion undocumented places making it difficult to do a post hoc gathering of the data. Each of these pieces would require some back explanation ("applications need a place to store your preferences so that it can remember them between usages"), and so there is a natural tendency to summarize and compress.

This compressed answer, something like "The information is stored all over the place", is basically correct, but it isn't the whole story. It sufficient in this case, but often a choice has to be made between giving a long, complicated answer and giving a terse but somewhat incorrect answer. I call this line beneath which any compression would result in incorrect information the "Irreducible Correctness Boundary". The problem is, the boundary is not fixed, but it varies depending on the sophistication and skill of the recipient.

I was reading "Fortune's Formula" recently and he posed the example of two different people asking if the world was round. The first was someone from the middle ages where the general conception was that the world was flat (I believe the theory of widespread belief in the earth's flatness has come into dispute recently, but bear with me), versus someone from modern times studying the shape of the earth. To the first person, an answer of 'yes' is a signficant amount of information and additional content regarding the non-perfectly-spherical nature of the earth is more harmful than helpful. The modern person, who knows that the earth is at least roughly round, might benefit from additional information that the earth is not really spherical, but an unqualified 'no' is a much less correct answer than an unqualified 'yes'. The real problem is that the irreducible correctness boundary is much higher.

Substitute a non-technical and a somewhat-technical questioner for the two people in the previous example, and you have a situation that technical people are encountering all the time. If you guess too high on someone's level of technical sophistication, you will swamp them with unnecessary or even misleading details. On the other hand, if you guess too low, you can quickly drop below the correctness threshold by omitting important details. Estimating a particular person's level of technical sophistication can take time, but it's critically important to know when passing data back and forth, especially when using someone as a trusted debugging contact.