34
CHAPTER 6. SUMMARY AND DISCUSSION
166
6
administration of groundwater resources, and wildlife management. … Separate
agencies administer these laws at federal, state, and tribal levels. … Each entity
has different missions, authorities, and modes of operation.
What we are left with is a patchwork of statutes that recognize jurisdictions of
state, federal and tribal agencies regarding countless issues affecting a watershed.
Where these authorities overlap, it is often difficult for governmental entities to
cooperate and share power among themselves as well as the regulated
community.”
This watershed situation illustrates a situation in which contradictions may occur,
intentionally or unintentionally. Contradictions may occur when different agencies are
promulgating regulations and these regulations overlap. Contradictions in the regulations
may also happen within a single agency’s regulations by accident due to carelessness,
when one person is combining regulations from different areas of the agency into one
document, or when revisions are made over time that interfere with previous rules.
43
As mentioned in Chapter 4, the research work presented in this thesis relies on first order
predicate calculus (FOPC) for modeling regulations, and this form of logic was sufficient
for approximately modeling regulations in this research. There is an extensive body of
research on other types of logic [44, 45, 85]. The work on advanced forms of logic could
be used to investigate modeling regulations more precisely. One reason FOPC was
selected for use in the research work presented in this thesis is that it is a standard form of
logic that has well-developed and efficient tools for reasoning with it. This may not be
the case for some of the alternative forms of logic that may be more expressive, and is an
issue that would have to be overcome in order to use them for compliance checking.
43
In the course of this research work we found an error in the federal used oil regulations, 40 CFR 279,
which were first promulgated in 1992. A subsection, 40 CFR 279.64 (e), is currently incorrectly titled
“Secondary containment for existing aboveground ta nks” instead of “Secondary containment for new
aboveground tanks.” While this is a simple subsection titling problem rather than a more serious
regulation contradiction, it illustrates the difficulty of managing massive regulations and attempting to
keep them flawless.
29
CHAPTER 6. SUMMARY AND DISCUSSION
167
7
There would be a distinct advantage to using a form of logic that allowed greater
precision in modeling regulations than FOPC. Greater precision in logically modeling
regulations would result in enhanced capabilities for providing regulation compliance
assistance services. In addition, more precise logic modeling of regulations would also
enhance the identification of contradictions and ambiguity in regulations.
There are many areas of future research possible in precisely modeling regulations. An
area of future research that would advance the work presented in this thesis would be to
investigate the application of advanced logics to modeling and computing regulation
rules within the XML framework for the purpose of compliance checking.
6.2.3.3 Rulemaking with Logic Representation
What if, when regulation rules are being drafted, the rulemaking agency had to include a
logic-based specification of the regulation, in addition to the regulation texts? What
effect might this have on the rulemaking process? Answering these questions is an
important area for future research, and a preliminary sketch of the area is provided in this
section.
An issue that lends support to the idea of developing regulations that contain a logical
component from the start is that ambiguities and contradictions must be resolved
eventually. Resolving these issues in advance, during the regulation-writing period,
makes their resolution more public. The regulation-writing process is designed to be
formal and open. The process is designed such that interested parties can keep informed
about what is going on, and have an opportunity to influence the process. When
ambiguities are “resolved” after a regulation is written, the process is generally not nearly
as open. For example, an agency may use its discretion with regard to ambiguous
provisions when deciding whom to prosecute and whom not to prosecute for violations.
It is generally not possible for interested parties to keep informed about what is
occurring. They may never know about potential prosecutions that were not pursued for
30
CHAPTER 6. SUMMARY AND DISCUSSION
168
8
unstated policy reasons. These issues all point to the idea that regulations should be
complete, i.e., containing minimal ambiguity and contradictions, when they are
promulgated.
Including a logic component in regulation writing could remove the ability of agencies to
“paper-over” issues. A logic component might force the agency to identify more
potential contradictions in regulations and resolve more ambiguities, since these
contradictions and ambiguities are difficult to model in logical form. Since this would
require very careful examination of the regulation rules, this process would result in very
carefully structured regulations. However, such an approach would also reduce the
discretion available to people enforcing the regulations.
There are also benefits in terms of knowledge sharing and regulatory quality that could
result from modeling regulations. Researchers in the Netherlands, working with the
Dutch Tax and Customs Administration to model new legislation, have found that there
are significant side benefits to modeling legislation for knowledge-based systems. They
note that, “these knowledge-ba sed systems proved to be useful by themselves, but
perhaps even more important were the side effects caused in creating them” [31]. The
researchers cite three major positive side effects of building knowledge-based systems.
First, knowledge about the legislation was made explicit through the representation of
this knowledge in the document. The explicit representation of knowledge about
legislation is important because the knowledge would otherwise remain implicit, and a
later need for this knowledge would require identifying an expert on the relevant part of
the legislation. Second, explicitly defining knowledge from all the appropriate experts
for a piece of legislation makes it easier to determine the validity of an expert’s
knowledge about the legislation. Third, explicitly defining the knowledge of various
experts about legislation allows wider dissemination of this information, thus allowing an
organization to more effectively use their likely sparse set of experts and improve the
quality of its enforcement actions.
28
CHAPTER 6. SUMMARY AND DISCUSSION
169
9
A key feature of modeling regulations as they are developed is that it provides a central
repository of expert knowledge. Because expert knowledge, that would have gone
undocumented, now is encoded in the regulation, there will be greater sharing of this
information within the regulatory community. Modeling regulations in a form of logic at
the time they are developed could help ensure regulations are clear and free of logical
problems [32]. This is significant because, from a policy standpoint, undocumented
ambiguity in regulations is problematic in that it can lead to different enforcement
policies and unequal rulings.
One drawback to developing regulations that include a logical specification is that it
would add a significant burden to the regulators’ task of developing regulations. This is
important to note, since it would further increase the workload for already overburdened
agencies. In addition to the issue of adding work to develop a regulation, if regulations
were not ambiguous, perhaps they would not be passed in the first place. Forcing a
resolution of all issues during the development of the regulation might be so
overwhelming that the regulation would never be completed.
The issues described above are fertile ground for future research on the effects of logic
modeling on regulation drafting. Developing a better understanding of how logic
specifications could positively or negatively affect the development of regulations could
impact the adoption of logic representations in the regulatory community. This is an
important area for future research with important implications for the way regulations are
drafted.
6.2.3.4 Regulatory Implications
The research work presented in this thesis does not directly change the regulatory
process, but it could affect it in the future. In addition to potentially laying a foundation
for using logic representations in the rulemaking process, this research work could have
28
CHAPTER 6. SUMMARY AND DISCUSSION
170
0
an effect on the regulatory process if it includes a feedback loop that provides regulators
with more information.
There are many forms of feedback information that a regulation assistance system could
gather for regulators. The following are forms of feedback that our regulatory contacts
indicated would be of value to the regulatory community:
• A regulation assistance system could help regulators identify provisions that users
feel are ambiguous or confusing. Provisions of this nature could be identified by
the frequency with which users select the “I don't know” option during
compliance checks.
• A compliance assistance system could provide statistics on what regulatory
provisions people are viewing or working with most often. Information of this
nature would be straightforward to collect from a regulation assistance system and
could provide statistics such as what provisions are the most or least commonly
used.
• A compliance assistance system could provide information about how companies
are complying with regulations. This includes information about what types of
companies are using particular provisions. This could help identify regulation
provisions being used differently than expected. Perhaps a minor exception has
taken on great importance for many companies. With this information, one could
also better predict impact of changes in the regulation. This information might
also be used by trade groups to advise members on best practices, or regulators to
optimize the structure of regulatory regimes.
• A regulation assistance system could provide more direct feedback from the
regulated community. For instance, people could file feedback during a
compliance check, with reference to a particular provision. This could assist not
29
CHAPTER 6. SUMMARY AND DISCUSSION
171
1
only in identifying confusing provisions, but would also quickly bring new
industry processes to the attention of regulators.
Online information gathering tools would also be useful for identifying how effective a
regulation assistance system is in achieving the goal of improving the rate of compliance
and understanding of regulations as a result of using the system. The EPA has a
requirement to measure how effective any compliance assistance efforts, such as training,
on-site visits, or paper guides, are in terms of changes made by the user as a result of the
compliance assistance effort. The approach used by the EPA could be applied to the
evaluation of a software regulation assistance system tool as well.
The EPA typically evaluates the results of compliance assistance efforts along three
dimensions, applied to regulated entities receiving direct assistance.
44
The first metric is
the percentage of entities that report an increased understanding of environmental
requirements as a result of the assistance. The second metric is the percentage of entities
that report the improvement of environmental management practices as a result of
assistance. The third metric is the percentage of entities that report a reduction in
pollution as a result of assistance, along with how much pollution was reduced.
Identifying the best way to provide the above types of information, in addition to any
other important information that can be gleaned from a regulation assistance system, are
important areas for future research. Assessing the impact of both regulatory feedback
and system evaluation information on the regulatory process also could be an important
area for future research. Gathering this type of information, however, raises some other
important questions in the area of privacy and security.
44
Per email correspondence with an EPA regulatory advisor.
29
CHAPTER 6. SUMMARY AND DISCUSSION
172
2
6.2.4 Privacy and Security Issues
The ability of a regulation system to gather the information described in the previous
section in a way not previously possible raises a number of privacy and security
concerns. In fact, even if a regulation-assistance service provider performs no explicit
information gathering, the exchange of sensitive information in the compliance process
makes security and privacy a major concern. While privacy and security concerns are not
a research focus in this project, it is useful to briefly address them here because they are
of such significance for future research.
An important research question is how a regulation assistance system might maintain
security and gather aggregate compliance information from the regulated community.
Even if regulators have access to information about how companies are complying with
regulations, it is important to protect an individual company's information from being
inappropriately accessed by regulators or competitors.
There is a large literature on performing secure transactions over the Internet. Various
encryption techniques are generally used to provide for secure communication over the
Internet [94]. Mediation technologies for creating secure management of information,
which could be used to manage access to compliance information, have been recently
developed [55, 108]. A great deal of this work could be applied to designing a secure
regulation assistance system.
In his thesis on tailoring online legal documents to individuals, Christopher Royles
explores a software system design involving a proxy server and a local program to
improve the security and privacy of his document customization system [83]. Under
Royles' model no significant user information needs to be sent across the Internet, where
it might be intercepted, or stored on a remote machine.
The regulation assistance system described in this thesis is currently implemented as a
web application that is run on a remote server. The system could also be implemented to
28
CHAPTER 6. SUMMARY AND DISCUSSION
173
3
run on a client's local machine, as a proxy server, or using some combination of these
design configurations. For example, one could implement the regulation assistance
system to run on a proxy server maintained by a trade association or some other trusted
source. Encryption could be used to protect information exchange between the company
performing a compliance check and the trusted proxy server. Another alternative would
be to implement a regulation assistance system that runs entirely on a client's machine,
only requesting updated regulations to be downloaded from the trusted source.
Determining which of these many possible designs would be best for providing a
compliance assistance service could be a focus area for future research work.
Research on how to protect privacy and maintain security while providing compliance
assistance services is an important area for future research in order for a regulation
assistance system to become a practical tool. In addition, the policy question of how
much information regulators should have access to with regard to how companies are
complying with regulations could prove to be an interesting area for research.
6.2.5 Implementation Issues
This section examines the issue of how this research work might be moved beyond a
research prototype and implemented for production use. Our brief exploration of the
topic here is in terms of who might implement a compliance assistance system and what
legal standing such systems might have.
If the EPA implemented a logic-based regulation assistance system they would most
likely treat it as a guidance document. That way the EPA would not be required to abide
by the compliance/noncompliance rulings of the system, since the EPA is not legally
bound by its own guidance documents. Although the regulated community would
probably be more enthusiastic about the technology if the EPA treated logic-based
systems like a regulation equivalent, since the system would then be legally binding and
32
CHAPTER 6. SUMMARY AND DISCUSSION
174
4
thus offer more legal protection, the regulated community would probably still be
interested if the logic system was considered a guidance document.
A disclaimer designed to protect against litigation in the event the system provides
incorrect advice, something all of the current expert systems described in the introduction
possessed, might also accompany any implementation by the EPA. For example, the
following disclaimer accompanies all the expert systems that are part of OSHA's elaws
suite of systems:
45
“STATUTORY AND REGULATORY DISCLAIMER
The Department of Labor is providing this information as a public service. The
regulations and related materials are maintained on this system to enhance public
access to information on Department of Labor programs. This is a service that is
continually under development. While we try to keep the information timely and
accurate, there will often be a delay between official publication of the materials
and their appearance in or modification of this system. Therefore, we make no
express or implied guarantees. The Federal Register and the Code of Federal
Regulations remain the official source for regulatory information published by the
Department. We will make every effort to correct errors brought to our attention.
Further, the advice about rights and obligations provided by this interactive
compliance assistance guide depends on the accuracy and completeness of your
responses to the questions asked.”
Another possibility is that someone other than the EPA, someone with fewer constraints,
could implement such a system. For example, a trade group, commercial service
provider, or a company’s internal IT group mi ght build a regulation assistance system. A
trade association or other third party building such a system might also add a liability
disclaimer. While the regulated community would probably prefer that the EPA
implements the system and thereby provides some legal cover, they would probably still
45
Located at the web address http://www.dol.gov/elaws/, accessed on July 19, 2003.
32
CHAPTER 6. SUMMARY AND DISCUSSION
175
5
find the system useful.
46
Although the regulatory assistance system provided by a third
party would not have any legal standing, it might demonstrate good faith efforts to
comply with the regulations if any regulatory action were taken against companies using
it. In addition, the record-keeping benefits of such a system would still have significant
value for companies. Another issue to consider is that if the system were implemented
outside the EPA, it would be easier for the ambiguities in regulations to continue to exist.
This is important if the EPA did not want to remove the ambiguities, since it would result
in less pressure on EPA to remove them from the regulations.
Research work on the best way to create a practical service for providing regulation
assistance services could be an important area for future research, since this might
increase the rate at which such systems are adopted for use.
6.2.6 Summary of Future Directions
This research has attained the goal of building an initial document repository, XML
structure for regulations, and a regulation assistance system. There are a variety of
important technical research issues that remain to be addressed, however, particularly
with respect to the regulation assistance system. These research issues are important for
enhancing the capabilities of the system and resolving important issues to make the
system more practical outside of the research domain. Some of the more salient research
questions for future research are:
• How can automated tools be built to help entities find information on state and
federal laws, as well as identify sources of assistance with compliance questions
and problems?
46
In presenting our research work for industry contacts, several companies expressed interest in a
commercial version of the system. This provides anecdotal evidence that even if a regulation assistance
system does not have legal standing or support from the EPA, companies might still be interested in
taking advantage of its functionality.
Documents you may be interested
Documents you may be interested
