Some people prefer Dear Abby, others Ann Landers. I’m an Ann kind of guy — too often, Abby’s advice lacks specifics and substance, whereas Ann’s advice is always practical and specific.

One of Ann’s better pieces of advice goes to “the other woman”. “If he cheated on his wife with you,” Ann regularly points out, “he’ll cheat on you, too.”

Unfamiliarity with Ann’s analysis is just one of the many flaws in a new Gartner Group report titled, “The Cost of Migrating COBOL Developers to Java,” by J. Feiman and R. Flatau-Reynoso (mentioned to me by IS Survivalist Joe Kruger who also pointed out some of its flaws). If you’re the kind of person who likes to look at the aftermath of train wrecks and traffic accidents, you might want to read the report yourself — you’ll find it at http://gartner5.gartnerweb.com/public/static/hotc/00092702.html.

The report begins, “Gartner has warned IS organizations that migrating their mainframe COBOL developers to Internet and Java development would be a painful experience. Here, we measure in dollars just how painful this process would be.” What’s painful is reading the report. It totes up every imaginable cost associated with retraining COBOL programmers in Java, compares it with just one of the costs of new Java programmers (their compensation), and concludes the Java recruits cost about $20,000 less.

It’s pretty hard to take seriously an analysis that ignores recruiting costs and hiring bonuses. These obvious omissions are less astounding, though, than the report’s assumption that Java developers will be available and productive the day the company starts recruiting.

Last I heard, the average lag time for hiring a skilled Java developer was about three months, but there’s no estimate of the business impact from a three month recruiting-generated project delay.

Nor is there any recognition of the ramp-up time any new developer needs to get accustomed to a new company, corporate culture, and IT environment. If the standard tools used by their new employer are different from those they already know, that will cause additional delays. Usually, new developers also find it pretty handy to understand the legacy environment they’re going to be dealing with. That learning isn’t instantaneous either.

New developers are never productive their first day of employment, and in fact are rarely productive the first month. Significantly, most of the process they go through to become productive requires the involvement of currently productive programmers, whose productivity suffers a bit from the time needed to get the newbies up to speed.

And then there’s the Ann Landers effect. A company that replaces long-time employees rather than retraining them will do the same thing again for the next bunch. I’m guessing most Java recruits will be familiar with Ann’s advice, which means the chance they’ll be loyal to their new employer is exactly bupkis. The next good offer they get and they’re gone, and why not? The report ignores this factor, but ironically it does comment on the risk of retrained COBOL programmers either demanding a raise or leaving. Maybe it’s assuming the Java recruits will all be here on H1b visas, where they can’t change employers.

When Ann Landers gives bad advice, she accepts 50 lashes with a wet noodle. Perhaps we should all mail some noodles to the Gartner Group and see what it does with them.