The recent report on the state of the Government Printing Office’s Federal Digital System (FDsys) should raise important questions for GPO and should be a wake up call for FDLP libraries. The report documents that the project is over budget, behind schedule, and lacks sufficient resources and planning to move forward successfully.
The report (Federal Digital System (FDsys) Independent Verification and Validation (IV&V) – Tenth Quarter Report on Risk Management, Issues, and Traceability Report Number 10-05, “IV&V Risk Management, Issues, And Traceability Report,” January 14, 2010) was written by American Systems, under contract to the GPO Office of Inspector General and is attached to the following memo available from GPO:
- Federal Digital System (FDsys) Independent Verification And Validation – Tenth Quarter Report On Risk Management, Issues, And Traceability. Assessment Report 10-05, March 24, 2010.
The consultants were charged with assessing the state of FDsys implementation. Some of the findings of the report that will be of most interest to FDLP libraries are:
- FDsys as it exists today “bears only partial resemblance of the system that was envisioned.”
- The program is significantly over budget. The original cost estimate for the first phase of FDsys implementation was $16 million. Through August 2009, GPO had spent more than twice that (approximately $33.6 million) and, by the end of FY 2010, the total costs for FDsys contractor support will be approximately $42 million.
- Even though the cost has more than doubled, the project is significantly behind schedule. “Release 1” of FDsys was slated to be accomplished in three phases ending in the Fall of 2009, but only the first phase has been deployed — and even that phase is incomplete.
- GPO now says that only 42 Collections will be migrated to FDsys instead of the originally-planned 55.
- There is an on-going indexing problem within the FAST search product. The FDsys database was due to approach a critical threshold of 2.5 million records in December of 2009 and FAST will require changes to accomodate more records.
- The majority of the work on FDsys after the release of the first phase of Release 1 has been centered on fixing problems and dealing with emerging issues. GPO is focusing more on fixing and upgrading a deployed system than on building the final system.
- The FDsys Program has performed little to no analysis, planning, design, or development work for Release 2.
- The “large number  of deployments [production builds] over a ten month period reflects the obvious fact that the originally deployed system contained numerous deficiencies.”
- The lack of clear definition of the system and the lack of a detailed implementation plan prevent GPO from determining realistic cost estimates for future development and endanger the ability of GPO to develop and deploy the final system.
- The consultants say that GPO does not have sufficient system engineering expertise to direct and oversee the development of FDsys and that this has resulted in a system with incomplete functionality, design problems, and numerous deficiencies. They recommend that GPO hire a senior system engineer and say that, without one, these problems will continue and future releases will likely take longer and cost more than anticipated. GPO management, however, completely disagrees with this recommendation.
There is more in the 38 page report, but the above gives you the gist of the problems.
The positive, the negative, and the risks.
There are some positive things. Much has been accomplished. Twenty-five of the most complex collections have been transferred from GPO Access to FDsys. The project managed to incorporate a significant design change during implementation to accomodate “Collections with numerous granules.” The project also was able to create a new capability to support public access to FDsys information via the Data.gov website.
But these accomplishments are overshadowed by the numerous problems that the report documents. Even the already-deployed system is apparently overwhelmed with problems. The report documents 232 problems that adversely affect the accomplishment of an operational or mission-essential capability and notes that the many unresolved problems with the system create “a serious risk that the overall goals for FDsys … will take much longer and require significantly more funding to achieve.”
The failed “paradigm shift.”
Since 1993, GPO has been championing a “paradigm shift” in responsibilities in which GPO arrogated to itself the responsibility for both access and preservation of government information and diminished the role of FDLP libraries. (See, for example, the discussion on GPO’s draft regional libraries report and FGI comments.) We at FGI have been concerned about this shift from the beginning for a number of reasons. One of those reasons is the danger of entrusting all preservation and free access to any single organization. Any interruption or failure of that organization (financial, technological, political, etc.), could mean a catastrophic loss of access to government information for everyone.
We have been hopeful that interruptions would be small and short and failures would be in the future. But we have hoped to persuade the FDLP library community, including GPO, that it would be wiser and more prudent and more durable to build on the existing FDLP model of sharing responsibility for access and preservation across many institutions. These institutions with different infrastructures, governing bodies, technologies, and communities of users, would, we have argued, do a better job collectively than any one institution could do by itself.
We have feared the day when Congress would cut back GPO appropriations after we all were irreversibly dependent upon GPO — a day when it would be too late to create a new system of free, permanent, public access to government information. With the release of this report, we worry that that day may be closer than we had imagined.
The original design and specifications for FDsys were expansive and ambitious. That was a good thing. It would be wonderful if GPO could support FDsys and all of its almost three thousand system requirements and features. OAIS compliance, persistent naming, metadata management, and support for RSS are among the features we look forward to. And we hope for other features: maybe, someday, APIs and OAI-PMH support, for example. But what do we do if GPO does not have the resources or the expertise to fully develop FDsys? It is hard to read this new report without being concerned that this is exactly the reality we face today. We worry that this report is “the writing on the wall” that is telling us that “the paradigm shift” will not work and is not sustainable.
Many FDLP libraries (or at least the directors of those libraries, and, in many cases, the FDLP librarians as well) have been hoping for almost two decades that they could rely on GPO to provide the services that the FDLP libraries themselves used to provide. If this is proving to be a false hope what will happen next? Is it only a matter of time before Congress pulls the plug, or GPO throws in the towel, or the private sector raises a stink?
What are our options for the future?
Naturally, we hope GPO can continue to develop FDsys. It would best for access and preservation to have FDsys in place. But it would be better if FDsys was not our only resource for preservation and access. It would be better if we had more systems in place to complement FDsys. It would be better if we had a digital FDLP that shared responsibility for access and preservation.
What options do we face right now? The obvious status-quo next step is for GPO to get more resources. It needs more money and more expertise so that it can deal with existing problems and move forward faster and with better planning that will make it easier for it to succeed and do so in a reasonable time frame and on budget.
But FDLP needs a “Plan B” to deal with the real possibility of GPO not getting adequate resources to finish or maintain FDsys. What will happen if we don’t have a plan in place? We can imagine at least three generic scenarios: One, GPO will scale back and provide less access, or less secure preservation, or fewer collections, or some combination of those. Two, preservation and access will remain government-provided, but will become completely fee-based (somewhat like NTIS and STAT-USA). Three, the private sector will move in and demand, perhaps under OMB regulations, that GPO shouldn’t have undertaken this job in the first place and that the government shouldn’t provide a system that the private sector could provide. It would argue that raw information should be given to private sector companies who will produce their own preservation and access systems that will be fee-based. (We almost certainly will see a proposal to replace GPO’s single-entity model with a private-sector, fee-based, single-entity model. Ithaka is already laying the groundwork for such a proposal. [See: Ithaka report on the future of the FDLP.] To those of us at FGI, this seems the worst of both worlds: relying again on a single organization rather than a community of organizations and moving that model to a fee-based system.)
There is also the possibility that none of these will happen and we will simply lose access because no one will take responsibility.
A better option; a more durable future.
But there is one other possibility: a collaborative effort in which GPO deposits digital files with FDLP libraries and those libraries preserve those files and make them accessible. This would be a real digital depository system with shared, distributed responsibility. It would have many advantages but, in the context of the current report, it has one major advantage over the current system: it has no single-point of failure (which is what we have with the GPO, single-entity, paradigm-shift model).
Such a system will take planning and resources and will not be trivial to implement. But the time to start planning for such a system is now. It would be much worse to wait until FDsys is in technological or budgetary crisis. At that point it could be too late.