The key problem here is that difficulty has two elements:
- The question
- The person answering it
Now lets assume you can perfectly gauge the question part, you can't account for the person answering it. To take an extreme example, I conside myself a reasonable Java developer, I am however not a poet. So I would consider the first question much easier than the second
How can I create a persistent file that represents objects in a binary form?
What is iambic pentameter?
A different person (in fact most people) would probably feel the opposite.
This is clearly an extreme example, but one persons easy question is annother persons impossible question. As such such a measure isn't very helpful