I think his point was that querying a structured data is better than grepping unstructured text. SQL vs Regex, for example. I get the impression that he didn't state what solution to use but simply that binary/structured > text/unstructured. He even says that Journal isn't his ideal solution and never will be.