How to interpret COBOL statistics

Dear Bob ...Consider the following items allegedly gathered from various Gartner reports:* There are 310 billion lines of legacy code operating in the world (65% of all software).* Five billion lines of new COBOL code are being written every year. * Fifteen percent of new application development is written in COBOL.* Thirty-four percent of coding activities are in COBOL.* Seventy-five percent of the world's busi

Dear Bob ...

Consider the following items allegedly gathered from various Gartner reports:

* There are 310 billion lines of legacy code operating in the world (65% of all software).
* Five billion lines of new COBOL code are being written every year.
* Fifteen percent of new application development is written in COBOL.
* Thirty-four percent of coding activities are in COBOL.
* Seventy-five percent of the world's business data resides on mainframe systems.
* There are 38,000 legacy systems at more than 10,000 mainframe sites worldwide.
* COBOL developers are retiring much faster than they are being replaced as educational institutions emphasize more cutting edge technologies.

I am a 20 year software developer (IBM midrange, RDBS, OLAP, Windows) who happens to love what he does and has no aspirations to move into management. However the impact of globalization on the field seems to be driving more and more of the development work overseas. Would I be crazy to consider moving into COBOL development to keep doing what I love best? What recommendations or alternatives would you suggest?

- Planning my career

Dear Planning ...

Like so many published statistics, what's there can serve to conceal what isn't there. For example:

* Five billion lines of new COBOL code are being written every year.
* Fifteen percent of new application development is written in COBOL.

What's the trend line - is the number of new lines of COBOL code declining? The percent of all application development? And assuming it's declining, how steep is the decline? If Gartner were to subtract the COBOL work that's been taken offshore, how much is and will be left for developers located in the U.S.?

* COBOL developers are retiring much faster than they are being replaced as educational institutions emphasize more cutting edge technologies.

Are they retiring faster than the rate of decline of new COBOL development, and especially the rate of decline within the U.S.?

* Fifteen percent of new application development is written in COBOL.
* Thirty-four percent of coding activities are in COBOL.

Put these two statistics together and you reach a fascinating inference - it appears that COBOL coding is almost exactly half as efficient as coding in other languages. That is, the amount of coding is double the amount of development.

Which leads to another unknown - how much of the COBOL work is adaptive rather than truly new - that is, how much of it is going into interface maintenance and minor enhancements, to support new functionality being created using Java, Visual whatever, SOA development tools and so on?

My opinion: COBOL is in decline. If you're looking for job security, you can probably achieve it in the COBOL domain for at least another decade. My expectation is that wages will be stagnant, and most of the new, challenging, enjoyable work will be in more modern development environments.

I'll also say this: More than the technology, the future lies with versatile developers who can interact productively with the end-user community at all levels to understand what they need to do and can translate that into working software without creating volumes of documentation artifacts.

Just my opinion, of course.

- Bob