Do Insurance Systems Suffer from a Legacy of Bad Decisions?

My post last week on the continuing value of “legacy” systems generated a number of responses, and I thought I would surface the discussion as a follow-up this week.

In my post, I said that legacy systems, such as IBM mainframe or midrange systems, still hold quite a bit of value for insurance companies, and that perhaps vendors are a bit too hasty casting existing systems to obsolescence while concurrently rushing us into buying their latest and greatest offerings. Modernization strategies such as service-oriented architectures, in fact, make platform and language (no matter how “legacy”) irrelevant, thereby extending the life of these investments.

However, some observers say there’s a good reason for putting many systems out to pasture. Matthew Josefowicz, formerly with Celent and director of the insurance practice at Novarica, wrote in a responding post that “the problem with Mainframe/COBOL systems is not that they’re bad at what they do. The problem is that for most applications built on this stack more than 10 years ago, it’s really hard to make them do something new.”

Josefowicz emphasizes that it’s not the platforms themselves that are at issue—they are well proven to be stable and scalable, but the legacy applications sitting on these platforms lack the flexibility needed to move forward into this new, networked age in which we live. They tend to be “poorly documented and incompletely understood by the people responsible for maintaining them,” he says, in addition to being “poorly structured, so that business logic, business rules, data management, and presentation layers are all mixed together in linear, non-reusable code.”

Still, readers appeared to be divided on the question of how to handle legacy technologies. One reader points out that a lot of good technology is being cast aside in favor of less-robust commodity server-based solutions for Windows and Linux environments. As a result, a talent drain is hindering the industry.

“It's unfortunate that the vendors of mainframe software have abandoned insurance software developments and, for the most part, colleges and universities have stopped any mainframe-related curriculums,” the reader says. “Those two things are, from my perspective, what is really killing the industry.”

Another reader feels that many companies are saddled with IBM legacy hardware, middleware and software, and the vendor has been perpetuating these systems with various modernization solutions versus going to the more flexible and agile systems available on the market.

“Clients loose track of what they could be doing to modernize their applications,” the reader writes. “They retire some of their legacy applications along with some of the IBM hardware and middleware they’ve purchased over the years, and becoming more agile at lower operating costs. Some of the insurance application vendors have figured this out, and started to leapfrog IBM methodologies and approaches.”

Still, another reader said the legacy discussion “was a real eye opener,” since those not steeped in technical knowledge are constantly bombarded by vendors armed with an “out-with-the-old, in-with-the-new” message. However, even newer, more agile systems are only as good as the people designing them.

“As a retail insurance broker, my experience with insurance companies is that they communicate extremely poorly, and they allow their systems to be designed by people who don't know squat about what the end user needs to deal with a customer,” he writes. “Somehow, common sense seems to get killed off in the corporate IT world.”

The runaway costs of IT projects have convinced this reader that perhaps “legacy equals bad” is a myth, especially “when it's far more likely a matter of garbage in equals garbage out.”