Some people say COBOL is kind of like a Wall Street bank: It's too big to fail. Not long ago, our blogger David Strom argued that as a stand-alone skill, COBOL wasn't all that valuable anymore despite its prevalence in legacy systems. Along with Fortran and to some extent Visual Basic, he said COBOL was pretty much "dead and forgotten." A number of users disagreed, pointing out that the federal government and corporations from banks to manufacturers rely on COBOL and are always on the lookout for people who can maintain their legacy systems. Last week, IT World's Eric Bloom wrote that prospects are good for COBOL talent. Big COBOL applications are "too big and work too well to consider replacement." A fair point, but he goes on to say that there are complementary skills it would be smart to develop:

Regarding expanding your knowledge and skill set, there are versions of Micro Focus COBOL running on Windows, UNIX, and Linux. I’m not suggesting that entirely new applications on these platforms are being written in COBOL, but instead COBOL shops are using it to create off-host interfaces for existing applications. They are trying to move these applications as-written to other platforms. The advantage is that this provides the opportunity to work on and learn these other operating systems as a way to expand your marketability.

Indeed, the demand for talent is there, it just doesn't get much buzz. Experienced mainframe programmers are beginning to retire -- one observer estimated that half of the Social Security Administration’s 1,000 COBOL-skilled programmers will be eligible to depart by 2015. Meanwhile, the number of those trained to replace them is inadequate. That dynamic, Bloom suggests, means COBOL-centric skills will be marketable for a good while to come. Should new graduates think about getting into legacy work? That's not clear. If they do, they should be prepared to undertake a lot of training in order to keep up with large-system developments as platforms continue to evolve. But for experienced techs, it seems a whole lot of people still believe mainframes remain a viable specialty. Should mainframe specialists stick to what they know, or is it time to move into another area? Let us know what you think by posting a comment below.