Based on my, admittedly quite limited, experience as an interviewer I think the author is mistaken. The type 5 questions, which he praises, are not very useful

Based on my, admittedly quite limited, experience as an interviewer I think the author is mistaken. The type 5 questions, which he praises, are not very useful in practice. "Tell me about your research" is only good as a warm up question, to make the candidate feel a bit more comfortable before getting to the real questions. You can have very bad candidates that will nevertheless be able to reasonably discuss their research. It's the topcoder-type algorithm/coding questions that will give you a much better insight into the candidate's coding skills and analytical abilities. So, if a company put emphasis on type 5 questions, it doesn't mean that they value "thinkers". It simply means that they're bad at interviewing.