But not in the back office of credit unions where even today, almost 20 years later, old cores, 20 to 40 years old and older, continue to log transactions.

"The core is the essential organs of the credit union," said David Gibbard, senior vice president at core company EPL in Birmingham, Ala. 

And yet, admitted Gibbard, "Most cores are not doing what their credit unions want."

What credit unions want, he elaborated, is a management information system, computers that tell the institution all it needs to know about its members, its lines of business, its opportunities and more. 

"That's not what most cores are today," said Gibbard, who said that the opportunity for core providers is to create systems "that let credit unions better understand and better serve their members."

Other experts envision a revolutionary shift away from cores that record transactions to cores that track relationships, that is, "social cores" that let an institution see exactly where it slots in its community and in the lives of its members.

Quite probably something along that line will happen as institutions shift to an off-premise, so-called service bureau or hosted cores, very possibly one that serves multiple institutions simultaneously, suggested Gibbard.  

A curious fact about cores is that most provide more power than their institutions need, but they also, paradoxically, do much less than their institutions need, that is, they have severely constrained abilities and flexibility. Shifting to a hosted and shared environment just might solve both issues at a more favorable price point for institutions.

But most credit unions still cling to their old cores and resist shifting to service bureaus, no matter how dazzling the possible computing benefits. Hodgins' advice is to stick with an existing core, regardless of age,  until it simply no longer provides what the credit union needs.  

And that usually happens only when the institution wants to extend into a wholly new service (perhaps mobile banking) or offer a new line of business (business lending is a for instance of a product line that challenges many legacy cores, said Hodgins).  Either way, until the core is broken, don't throw it out, is the advice from Cornerstone.

Core vendors, incidentally, also have few reasons to kill off aged cores.  "They are cash cows," said Brad Smith, CEO of consulting firm Abound Resources in Austin, Texas. The vendors are not investing in upgrades, so the service fees that come in mainly fall straight to the bottom line, said Smith.

That leaves neither side motivated to kill off an old core.

Also troubling is that one recent attempt to build a core from the ground up, Fiserv's Acumen, has seemed to stumble to a premature death, with Fiserv opting to buy the aging Open Solutions Inc. DNA core and use it, with some upgrades borrowed from Acumen, as its new flagship core.

A reason for the lagging core development is that shifts off legacy cores are very few in number. There's a handful every year, no more, and that sluggish market has not fueled substantial investment in developing new core architecture.

There also, increasingly, are clever strategies for prolonging core life, mainly by moving functionality off the legacy core and into newer computers.  

Mary Nugent, a vice president at systems modernization firm Micro Focus, said that the first thing her company does when it is brought in to tweak a sagging core is hunt for processes to shift off the core. "That is low-hanging fruit with little impact on [members]," said Nugent. The prime worry with any core update is that members will be inconvenienced or worse. Selective off-loading of functions that do not need to be on the core is a generally safe strategy, said Nugent.