Even though I’ve been in the game less time than you, the respect accorded to combinatorics has increased noticeably even since I started out. I think that because many combinatorial problems are so easily stated (if not solved) that it is very easy for them to be dismissed by researchers in subjects of such technicality that you need a decade of study merely in order to be able to understand the STATEMENT of the problem, let alone tackle it.

Your article mentions a few of the factors – in particular, the central role of combinatorial problems in computational complexity and the development of really deep structural theory by Robertson and Seymour.

But I also credit another factor – a vastly increased focus on the concept of an abstract network, driven by the study of real networks in particular the Internet and social networks, and popularised by the “small world” craze of a few years ago (which of course includes genuine research in addition to pop-science).