I have some thoughts as a result of the recent news that the Senate failed to advance legislation on NSA reforms. My personal opinion about these reforms are ambivalent. The motivation for this post, however, concerns the particular objection raised by Senator Mitch McConnell “This is the worst possible time to be tying our hands behind our backs. The threat from ISIL is real”. I think he has a valid point in that the world is scary right now and NSA is providing some form of defense. My objection is that we should reject the legislation because of its consequence of “tying our hands behind our backs”.
Congress has no hesitation for passing legislation that “ties the hands behind their back” for otherwise useful and lawful industries. If this were a decisive argument, then we would not have passed restrictions on financial industry (such as Dodd-Frank) and on healthcare industry (such as HIPAA, and Affordable Care Act). Both of these are vital industries to the US economy and these laws undoubtedly impose tremendous burdens on the industries that impede their ability to perform their missions. So far, the industries appear to be handling the new burdens while still providing their services, but they would be able to deliver their services much more efficiently without these laws. These laws clearly handicap the abilities of these industries to do their work. The metaphor that NSA reforms ties their hands behind their back is an exaggeration. That exaggeration applies equally to what congress readily imposed on financial and health care industries. As a citizen, I have as much reason to be concerned about collapse of financial or health-care systems as I do about ISIL.
From my outsider perspective, I observe that these industries have so far been able to absorb the additional burdens imposed by congress. Certainly they are complaining and they are challenging provisions where they can. But they are also making the effort to comply with the regulation. The regulation has not prevented them from doing their business. Instead the impact of the regulation is to make their business more difficult. The regulations force the industries to change their practices.
These changes require the employees to work harder. The regulations are absorbed into the tears and sweat of the workers in these industries. Both financial and healthcare industries broadly burden their staffs with closer documentation of their daily operations and more restrictions on how they spend their working hours. This is especially true in healthcare industries where regulations have forced a sweatshop environment on the delivery of healthcare. The legislation does not prevent them from doing their business. They just have to work their staff a lot harder.
I regret the consequence of imposing on healthcare staff burdens on their time that approximate the workflow demands on late 19th century assembly lines. The nature of healthcare in particular depends on direct human-to-human delivery of care for benefit to both the patient (improved recovery outcomes) and the provider (better satisfaction of work). In this context, it is not an exaggeration to describe legislation as tying their hands behind their backs when that legislation demands that providers spend less time per patient and devote more time to paperwork instead of care.
We knew of these consequences when HIPAA and ACA were passed but we decided it was for the greater good that they be in place. We should be able to make the same calculations when it comes to regulating agencies within the US government.
For this post, I assume that the above mentioned regulations on financial and medical industries have mostly beneficial results. Although these regulations have imposed new burdens on the industry, the added burdens come during a time of increasing productivity especially in the areas of information technologies for handling administrative tasks. Lacking this demand for additional burdens, the industries may instead have used the productivity gains to reduce staff or to reduce job-requirements for filling positions. The new burdens from regulations raise the requirement for staff and skills that offset the productivity gains, but in a way that provides more value at least from the government’s perspective.
The government agencies are enjoying the similar productivity gains from the same technologies. Due to the way government staffing works, the gains are unlikely to result in reduction in staff. Instead the productivity gains results in less demanding jobs or a more relaxed working atmosphere. Without some external imposition of new job burdens, the productivity-enhancing technologies will not bring any real productivity improvement when measured in terms of labor units over months or years. The staff will continue to perform same quantity of work over that period but will see their work levels decrease as a result of technologies making their tasks easier.
As a result of investments in productivity technologies, the workforce in federal agencies have spare labor capacity in excess of what is needed to meet their mission.
When I started writing this post, I was trying to imagine a way for the voters to regain more control over agencies through their representatives in congress. I am concerned that once congress creates new agencies the bulk of the specific regulations get created outside of congressional control as described in this article:
The rise of the fourth branch has been at the expense of Congress’s lawmaking authority. In fact, the vast majority of “laws” governing the United States are not passed by Congress but are issued as regulations, crafted largely by thousands of unnamed, unreachable bureaucrats. One study found that in 2007, Congress enacted 138 public laws, while federal agencies finalized 2,926 rules, including 61 major regulations.
One example of these non-congressional regulations impacting the healthcare industry is the requirement to adopt the new medical coding standards called ICD-10. This is a major new requirement on the industry requiring new investment in technology and in staff training. Once implemented, the coding standard imposes more burden on the day-to-day operations of the delivery of health care. This requirement was not from an act of congress but instead it was a decision made within the department of Health and Human Services (HHS). The ICD-10 standard is very complex with detailed coding for every medical scenario. It is not a standard that could be created inside congress. However, the ruling to use it and the time when it must be used dramatically affects the healthcare industry. This is government decision imposed on industry without much if any recourse of using democratic processes through congress. The industry has to instead work out the issues directly through the HHS agency.
I believe the original intent of the constitution was that the voters would have a say in the crafting of all laws that will affect their lives. That original concept was the voter’s elected representatives would debate and agree on those laws, rules, and regulations. Instead, congress has transferred much of this rule-making business to federal agencies that in turn set rules without congressional interference. My hope was to find some way to restore more active role by congress in the agency’s activities in creating and revising regulations that affect individuals and their corporations. That same article deflated my hope with the observation that a Supreme Court case in 1984 decided that “agencies are entitled to heavy deference in their interpretations of laws”. The decision reflected a pragmatism of the fact that congressional laws can not cover every situation and thus the agency has an implicit power to make their own decisions when the legislation is silent or ambiguous. Although congress can revise their governing legislation, the agencies are granted broad autonomy to act without involving the congress. At least that is how I’m reading it.
In earlier posts (such as here), I introduced my preference that we participate in rule-making through a democratic process. That process gives us direct access to those who make the rules that affect our lives. This is not what we have today because of the size of government and the necessity of administering this government through autonomous federal agencies.
Undoubtedly the federal government is far larger today than the original framers of the constitution could have anticipated. Equally true is that our communication and information technologies have advanced far beyond even their imaginations. Compared to the early years of the republic, we have far less democratic control over the government apparatus and this is leading to a more autocratic government. As noted in the introduction to this post, we can not even attempt to reform an agency out fear that comes from our ignorance of what exactly they are doing.
In my earlier post on dedodemocracy, I described a form of government by data instead of by people. I believe this is occurring by default without our awareness. The government is acting on data it collects to make new policies. The authority of the data replaces the authority of democratic processes justify these policies. Much of this data is locked inside of government so we can not observe it. In that post, I proposed a way to regain a democratic participation in a data-driven government by making all of the decision-relevant data be available to the public, and by educating the public how to query and interpret this data (where interpretation includes scrutiny of its relevance).
There have been many attempts to use the democratic process through congress to reform various previously authorized agencies. These attempts have been very unproductive. Once the agency is created, it seems to quickly become autonomous and outside of reach of congressional oversight. The courts back up this autonomy as mentioned above in the 1984 supreme court ruling. Also, recently congress has explicitly deferred critical decision making to the agency as illustrated in the recent affordable care act (ACA) explicitly granting so much power to the secretary of health and human services. In my understanding, the wording of the ACA deliberately designs the provisions to be outside of congressional control and thus outside of political influence. I will generously assume this is an attempt to be sure that these critical decisions are based on data instead of political whim.
Modern information technologies, and especially data technologies (big data, data governance, predictive analytics, etc) offer the potential for better government. Realizing that potential requires restraining the human political influences. I describe this treating this as an appeal to the authority of data that encourages automated decision making. I think the much of the spirit behind the affordable care act is to get health care out of politics and fully within automated data driven decision-making. Conversely, the inability to reform an agency like NSA may be the result of the fact that the agency has developed elaborate data-driven processes that must be outside of political influence. In some sense, NSA is already doing government correctly. Similarly, the ACA is on the right track to eventually govern correctly. This correct form of government is a data-driven government, not a democratically debated government. It seems to me that this is the unifying vision of government that explains what our congress has been constructing over the past several decades. In particular, this seems to be an accurate way to describe the ACA.
As I have mentioned previously (such as here), social cohesion required a super-majority consent of the governed. By consent, I mean that an overwhelming majority support the government to the point of being willing to defend it against rebellious dissenters. I argued that our history proves that we can obtain this consent without coercion through the use of accountable decision making. The decision maker is a human who is trusted to make a human decision that considers both the evidence and his informed fears and doubts. The accountability becomes important when decisions produce disastrous consequences for some group. The accountable decision-maker needs to provide a persuasive argument either to convince the aggrieved that the decision was a good one, or to convince the super-majority that the aggrieved is being unreasonable.
In contrast, the data-driven approach eliminates the human accountability. One way to retain social order is to coerce participation through the demand that people believe that the data-driven decisions are the best possible decisions. The only way to maintain social order in a data-driven government is through forceful applications of penalties for not cooperating. At least that seems to be the approach we are taking today. The democratic-impenetrable regulations of agencies require agency-issued penalties to obtain compliance to the regulations. This is necessary because the ones who must cooperate had no access to the data that is justifying the regulation in general and justifying this specific enforcement in particular. From a human perspective, the enforcement of regulation appears indistinguishable from authoritarian rule by the whim of some more powerful person.
In my discussion of dedodemocracy, I suggested a way to regain democratic participation in a government that is becoming entirely data driven. In a data-driven government, the decision making must become automated based on the available data and the public’s cooperation must become an obligation. However, data driven decision making relies on new data to make future decisions. The rationale to accept suffering the consequences of a past decision is that this will produce new data for future decisions that we hope will avoid this consequence. The way to restore something comparable to democratic participation in data-driven government is to allow the public to have direct access to the data that is contributing to the decisions. Democratic access to data does not result in a democracy that can change the decisions (although this is a possibility), but it will allow a person to convince himself that the data justifies the decision.
Also, this data access will allow a person to adjust his life choices to have more favorable outcomes based on decisions he can anticipate are coming. Assuming that the decisions are purely based on decisions, access to the same predictive algorithms will allow the public to anticipate the future decisions.
I started this post discussing a frustration that congress experiences trying to reform government agencies to conform better with democratic political pressures for reforms. This frustration is based on an anachronistic view that such decisions must be democratically decided. Increasingly the decisions made within government agencies are based on data or at least are defended as being based on data. The real problem for the modern public is that they have incomplete access to this data. NSA provides an extreme example of this necessarily hidden data.
A more productive approach for congress to reform federal agencies may be to focus instead on making decision-making data available to the public. The agencies can continue their autonomous decision making, but they must make the data behind those decisions available to the public. This is not a new concept. Congress has already passed Freedom of Information (FOIA) laws. Congress also has subpoena power to request specific data. These powers are consistent with what I mean by democratic access to data.
The problem with both FOIA and subpoenas is that the data is very expensive to obtain. Every request for data must be handled manually in order to review and redact irrelevant data. The resulting long delays and incompetent redactions greatly limits the value of this data having a role in democratic participation in the decision-making process.
This experience suggests my suggestion for government reform. Instead of focusing on revising the missions and limitations of agencies, we could instead demand a better data management system to promote freer flow of information. With modern technology, we should expect that FOIA and subpoena requests be answered immediately where the delivered data is automatically free of irrelevant data (that otherwise would require manual redaction). In order to facilitate this kind of responsiveness to data queries, the government needs a different approach for recording its work products.
A very useful reform for all of the federal government would be to demand that all of the government job duties be cataloged into distinct work products that can have a particular number assigned to it. These numbers would be specific enough to distinguish different types of activities but general enough so that they will occur frequently among the staff and over time.
The reform introduces to government business practices something similar to the ICD-10 medical coding system that is imposed on healthcare industry. The ICD-10 system is designed to capture specific conditions that are treated. A particular case may involve multiple codes where each code captures information such as type of condition, location of the condition, and the events that lead to the condition. The codes provide information about the actual case that requires medical attention. The system of codes is very exhaustive to cover every medical condition with specificity to distinguish the condition from other conditions and yet general enough to apply to multiple cases (although some codes are likely to be very rare).
It should be possible to create a similar system of codes describe all of work products of government workers (civil service or contractors). Like the ICD-10 system, these codes describe the case to include both a description of what type of effort is involved and what external cause required this task to be performed. Also like the ICD-10 system, there may be codes to distinguish whether the task was initial assignment or a follow-up for revision or addressing some complication.
An example may be a task to present a briefing to a meeting. That task may have a code to indicate it requires being in a certain place at a certain time, it involves speaking, and the speaker is a peer of the audience members. This meeting may involve multiple codes to capture whether other resources are required (a meeting room, a teleconference number, a projector, etc). The topic of the meeting and the briefing may also be coded.
Preparation of the meeting will involve still more codes for reading, communications (each form having a different code), writing, editing, reviewing, printing, etc. A single meeting may result in recording dozens of different codes for what went into that effort.
Each code would be attached to a particular documentation of the effort involved. For example, an specific email containing specific information would have one or more codes attached to it. The author would be responsible for assigning the appropriate codes.
Each specific piece of information in an email or a document would be flagged as to which code applies to it. For example an email that replies to an earlier email may quote the original e-mail. The original e-mail involves codes concerning reading the email and reacting to it, while the new content would have codes indicating what effort went into preparing that content.
Each specific piece of information in a document that resulted from a specific individual’s work effort would be flagged with a code indicating the nature of work to produce that information.
There would need to be codes for idle activities such as breaks and non-work-related conversations.
These codes will be recorded both to describe a particular work product (presentation, or e-mail in these examples) and the labor of a particular worker. The work-product codes should be specific enough to describe work in set periods of time, such as 15-minutes, one hour, or half-day. At the end of the day, the staff’s attendance record will account with one or more codes for what he is doing during each period during his duty-day.
Coding at this level will introduce a significant new burden on the federal workers. However, this level of burden is comparable to the burdens that congress imposed on the financial and health care industries. These industries had no choice but to accommodate these regulations through a combination of technology upgrades and additional staff duties and training. As I mentioned earlier, the regulation on these industries leverages recent productivity enhancing technologies in a way to keep staff busy providing new value instead of using productivity to make job duties lighter or allowing for reduced work-forces. The productivity improvements get more work out of workers instead of allowing workers to enjoy idle time on the job. In the example of the healthcare industry, the regulation fills the worker’s day with new tasks as maintaining electronic records.
The proposed reform imposes a comparable burden on the federal workforce and they can adapt in the same way. The reform adds burdens to more precisely document their work-products and how they spend each time-unit of their duty hours. This will increase their productivity by converting the idleness gained from productivity enhancement into a useful product of generating data that the describes what they are working on and how much time it requires of them.
The benefit of this reform is to produce more data about what is going on within government agencies. The coding concept will allows this data to be structured in a way that it can be easily queried or analyzed in aggregates.
Having highly-specific codes for each work product will allow for quick queries for the purposes of satisfying subpoenas and FOIA requests. We can be satisfied that the codes are precise enough if such queries can be automated without any need for manual review or redaction. The need for such manual intervention is evidence that the coding is insufficiently precise and needs more work.
The goal for example is that a specific FOIA request should be automated to the point of being able to provide an immediate result that delivers all of the relevant information and without any irrelevant (or sensitive) information. The diligent and precise and complete coding of individual work-products can make this possible.
Another goal is to evaluate the labor costs for work-products especially in aggregate analysis to observe the distribution of labor costs across the staff performing similar tasks and over time. This will allow for better understanding about resources the agency requires to accomplish its mission involving a multitude of work products. It will also permit better optimization to identify under-utilization of resources or inefficient deployment of resources.
The above introduces an approach to reform federal agencies where the reform focuses on improving transparency of the agency’s activities instead of trying to change what they are doing. As discussed, the latter project of changing agency activities is very hard for congress in part because of the lack of awareness of what the agencies actually are doing. By introducing a reform that requires more extensive documentation of work products (comparable to regulations congress imposes on industries such as my examples in finance and healthcare), we can obtain the benefit of easier access to data such as faster FOIA responses without the need for manual intervention delays. One possible reform approach is to use a decimal coding system for federal agency work-products similar to the ICD-10 system that the healthcare industry uses for individual healthcare cases. The benefits will be similar in making more precise data available for querying and to establish reasonable reimbursement rates for each code.
The FOIA responses can be open (and affordable) to the public in the form of open-data initiatives so that everyone can observe what is going on within agencies and in particular to anticipate future changes to regulations. This access to data will permit a stable transition to a government by data with democratic participation in the form of access to the same data.