So, their methodology is to search Google, MSN, Yahoo!, etc. for "XXX programming". In case it's not obvious why this methodology is fatally flawed, here is som

So, their methodology is to search Google, MSN, Yahoo!, etc. for "XXX programming". In case it's not obvious why this methodology is fatally flawed, here is some evidence: See that dip in Java in 2004? They claim that is because of a change in Google's indexing. That such a change can affect one language in vastly different ways than the others indicates a methodology problem. For example, suppose that in 2007, we saw the number of programming job postings on the web go up 2X. That would inflate the numbers for languages that are commonly mentioned in job postings.