Wednesday, April 20, 2016

Federal government waking up to "agile systems methodology" for health care, FBI, burned by SLDC debacles in the past


Timothy B. Lee has an interesting article on Vox explaining how the federal government and Obama administration is reforming its information technology management at critical agencies, most of all the FBI and HHS.  The 2013 “Healthcare.gov” rollout catastrophe may have been the last straw, and it came rather late.

Lee discusses a project management technique called “agile software development”, whereby users deliver definite capabilities to users live and then reiterate to fine tune the products based on user feedback.  This seems to be a departure from the old SDLC, or “systems development life cycle”, with which familiarity was so essential to job interviews in the past, most of all the 1980s and 1990s, as Y2K approached and as mainframe had to integrate with Internet and other end user GUI platforms.

Source Seek has a discussion of agile methodology and refers whimsically to the “manifestos” out there on software development, and everything else.  “Agile” means “iterative” (and “streamlined” and “timeboxed”) rather than the “waterfall” or “blueprint” technique.  Ever heard the mantrum, “no more changes”?  I have (below), but agile loves change.



The older ideas were certainly in full boil when I worked for a Blue Cross and Blue Shield Consortium, “CABCO” in Dallas from 1979-1981. We depended on various “phases” with walkthroughs (or prototype simulations, as Lee describes), pseudocode, steering committee meetings, CEO meetings, in a repetitive cycle.  The project eventually failed because of squabbling over turf among the plans, when the right concept was to let the end users (affected by conditions and local laws in different states) input many of their requirements, an idea already getting traction in the mainframe world (with CICS interfaces) by the mid and late 1970s.

The skill of the people hired is also maybe an issue.  The transition from old school procedural (often mainframe in verbose languages like COBOL) programming to OOP is not simple for older workers, who need the opportunity to build things from the ground up, not jump in and maintain what other people did.  But there was also a tendency for many “mature” mainframe professionals to leave and go do something else (as I did) after Y2K and then the shock of 9/11.  The recruiting industry and the way contracting firms did projects – with temporary W2 gigs – did not encourage retaining the level of maturity needed to put in a system like “Obamacare” with all its complexities, many of which were oversights in policy as well as technology.

I could have contributed materially to the development of “Obamacare”.  But if I had anything to do with it at a senior level, it would not have gone in with all the problems it had, and I would have squawked about the idea of people paying more under the new system than old, partly because of being forced to buy coverages they personally didn’t need. Maybe that wouldn’t have been my call – but I think nobody could see (or simulate) how this could happen technically.

Of course, talk is cheap, and free.

No comments: