Obviously "like Cobol" doesn't mean "useless". A hell of a lot of vital applications are still running in Cobol. Indeed, I'm sure that one way to locate a vital application is to look for one that is still running in Cobol.
And I seriously doubt that "the way of Cobol" really means "completely stagnant". Just because I don't notice the work that the world's Cobol hackers do doesn't mean that it isn't real. If nothing else, you would think that the Cobol folks have had to quietly and heroically build many, many things over the years just to make sure that available modern hardware and infrastructure can support their ancient codebase. For example, we all know that they had to reinvent all the date-formatting code a few years ago, right? That one even I have heard of.
What "the way of Cobol" means is "things happen, but nobody cares". The remaining Cobol code is very important, but folks don't go out of their way to create much more of it. There are people still inventing things in the Cobol universe, but they go quietly about their jobs with a minimum of fuss and marketing and absolutely no mentions in the popular press. It is intensely unfashionable, kids don't learn about it, and language designers make jokes about it.
Java's not there yet, but you can see the handwriting on the wall. It won't last another twenty years as, e.g., the de facto standard language for college students.