What's your argument exactly? That they should have used narrower terms until they found an article that was already written?
Here the guy is just saying "whoa great dude and next I'll use regexps oh my god woaw". We have word2vec, doc2vec, many cores, … why aren't journalists using the tools they should be using?
If you write the post maybe someone will be inspired to build a tool for journalists using these algos.
Hopefully out of the many tech heads who read HN, one will have the same thought as you and build the tool.