Maybe you”ve heard this one before, but … OPM has decided it”s time to improve its processing of federal retirement applications and its solution is to put more information technology behind it.

OK, there”s no maybe about it. You have heard that one before. Time and time again. So many times, in fact, that not so long ago OPM declared that approach a failure and tried a different course.

The retirement application processing system is a mess and has been for years, decades even. It results in some new retirees not getting a final adjudication of their benefits for many, many, months. In the meantime they get only partial benefits that often are well below their entitlement. Many times the calculation is wrong even after all that, and throughout the whole process, customer service is iffy to poor.

There are many reasons for this, including errors by individuals as they fill out application forms and by employing agencies that receive those forms before sending them along with supporting information to OPM. But the delays happen at OPM itself.

On one level, calculating benefits is a simple task. Civil service benefits are based on a formula with only two variables—length of service and high-3 salary level. However, there are numerous special rules regarding the former that must be taken into account—military service crediting, needed payments for crediting of certain civilian service, questions about whether certain types of work count, and so on. Overlaying Social Security onto the system, which happened when FERS was created, added further complexities, with various types of offsets or added benefits possible. There are rules, and exceptions to rules, and exceptions to exceptions.

This is where IT problems come in. Over many years, OPM has struggled to bring federal employee records out of the 19th Century; it still maintains vast rows of filing cabinets inside an old mine in Pennsylvania. Recently a major newspaper “discovered” this well-known and never disguised fact and published a lengthy article about it, causing OPM to hunch into a defensive crouch with a posting on its own site, and several members of Congress to express outrage.

To only outline just the recent history of attempts to solve this problem, the Bush administration launched a project called retirement systems modernization, which involved not only moving records to electronic format but automating the processing of applications. The public-facing aspect of that was to be an online portal called RetireEZ.

As OPM prepared to go live with RetireEZ, GAO warned again and again that the system was not ready and that it was vulnerable to making mistakes. OPM launched it anyway and guess what, the problems GAO warned of soon happened. After months of nasty finger-pointing between OPM and the main contractor and futile attempts to salvage something usable, RetireEZ was abandoned.

Early in the Obama administration, OPM”s new director John Berry had a mea culpa moment on behalf of the agency and called in reporters to announce the end of the long chain of IT-based initiatives culminating with retirement systems modernization. He specifically deemed them all failures that suffered from changes in direction, cost overruns and other common woes of big IT projects. Worse, they actually were counter-productive, he said, because in anticipation of automation improvements, OPM had reduced the number of people on the job. Rarely has an OPM director been so upfront—adamant, actually—about any shortcoming of the agency.

Instead, OPM ditched the IT-heavy approach and diverted the money into people. It increased hiring, added overtime, and shifted some people in from related occupations to become application processers.

Later, OPM eliminated any references to retirement systems modernization from its rules and documents, a somewhat Orwellian move: We have no references to retirement systems modernization; therefore retirement systems modernization never existed. It meanwhile appeared to try to shift blame by saying that individuals and agencies need to do a better job of preparing the application for adjudication in the first place, and said that certain delays were out of its control because SSA was slow in providing needed information. And there went much of the good will OPM generated by confessing to its failures.

That aside, the personnel-intensive effort did start to show results. It produced a steady decline in the backlog until the middle of last year when budgetary cutbacks caused OPM to pull back a bit. Despite that, OPM is on track to reach during this year its goals of processing 90 percent of applications within 60 days and eliminate the backlog while processing as many applications each month as it takes in moving forward.

That was why so many people scratched their heads last year when now-OPM director Katherine Archuleta said during her confirmation process that boosting OPM”s IT systems would be her top priority. At the time, OPM had much more pressing responsibilities, including dealing with then-ongoing sequestration furloughs, bracing the workforce for a potential government shutdown, trying to end a three-year freeze on salary rates, fighting proposals to denigrate retirement benefits, finalizing phased retirement authority and on and on.

Some also found it puzzling that a nominee with no particular background in managing big IT projects would decide in advance to take that approach, without having had time to really understand what had gone before.

Recently she carried through with her intention, launching a new “strategic IT plan” that focuses on “case management” that will “track and report on cases at a more granular level than we are currently capable of doing.”

“Granularity” is fine and good. But if the long, sorry history of OPM”s retirement processing improvement efforts has proven one thing, it”s that adjudicating a federal employee”s retirement benefits is a labor intensive task. There simply is no substitute for putting someone”s brain on the case. OPM itself learned that lesson once, or at least it had seemed to.