Leaving aside what they're willing to pay or not pay for COBOL developers, does anyone seriously believe that a bunch of developers can waltz into an old code base and speed it up by a large amount in a relevant timeframe? Which seems to be what these articles are at least implying.