It's perfectly reasonable for a job interview to include questions that require you to use your skills to solve a problem. Each of this book's chapters contains exercises that might make good interview questions—at least if the candidate is well-versed in algorithms. Many of those questions would be quite difficult if you hadn't recently been reading about the relevant algorithms.
Recently, certain kinds of puzzles have also become popular in interviews at companies such as Microsoft and Google. The puzzles are intended to measure a candidate's creativity and critical-thinking ability. Unfortunately, these sorts of puzzles come with a large set of assumptions that may not be true. Most business situations, even in programming, are not phrased as puzzles involving balance scales, marbles, rickety bridges, and goats. They usually don't involve a clever trick or an amazing insight that is blindingly obvious after you hear it but that is practically impossible to figure out in a 10-minute interview.
It's true that finding the best solution to a real-world problem often requires creativity, but many of these kinds of puzzles don't measure creativity. Instead, they measure whether you've scoured the Internet long enough to find the problem that the interviewer is asking about, or something similar.
For example, consider the following questions: