DRAFT 7/2000

By John O'Looney

There are numerous avenues by which citizens can influence government behavior, for example, through electoral politics, service on boards and advisory groups, petitions, lobbying, and attendance at public hearings and forums. One of the most effective ways to influence government behavior, however, is through participation in rule making. Rule making occurs whenever an executive department is charged with deciding how to implement a strategic policy developed by the legislative branch. Typically, rules represent the "details" that make all the difference between an effective and a "gutted" policy or law. Traditionally, participation in rule making has been limited to policy insiders and groups with a strong interest in a policy area. The lack of interest in a topic will always limit participation in rule making. However, the logistics and inconvenience of participating in rule making has also meant that citizen participation has also been limited by lack of access to high-quality, up-to-date information and convenient channels for making one's voice heard. There is hope, however, that advances in government and private sector Internet technologies and service will address this situation.

Before identifying these developments, it is important for the reader to have some understanding of the procedural rules underlying the rule-making process. Both the Federal government and most of the state governments have administrative procedure acts that define the exact steps for how rule making will take place in their respective jurisdictions. For our purposes, these procedures can be categorized into four types: 1) notification; 2) format, 3) consideration, and 4) reporting.

1) Notification: Notification is the stage at which resource-poor citizens are most disadvantaged by the traditional rule-making process. Governments typically provide for at least 30 days' notice of a proposed rule-making decision and often will mandate that a short summary of the rule also be created so that citizens do not have to read a long, complex notification document to determine a possible interest. The system tends to break down in the publication and distribution stages of notification. At the federal level, notification through the key publication sources (i.e., the Federal Register and Commerce Business Daily) possesses the weakness of all comprehensive information sources (or non-personalized Web portals) that are too broadly defined to be useful. That is, in order to know that a single bureau in a U.S. government department has issued a new proposed rule, a citizen must get access to these publications every day and review them fairly closely. Administrative procedures acts will often provide citizens or interest groups with a more time-husbanding information services--if the citizen is willing to pay. For example, in Georgia the Act reads as follows:

"The notice shall be mailed to all persons who have

requested in writing that they be placed upon a mailing list which

shall be maintained by the agency for advance notice of its

rule-making proceedings and who have tendered the actual cost of

such mailing as from time to time estimated by the agency;"

While this service is indeed better than being kept in the dark, it imposes a resource barrier that runs counter to the spirit of democracy. Fortunately, intelligent notification technologies holds some promise that for those on-line at least customized notification of proposed rules in one's interest area can be sent via e-mail or a personalized Web portal. The beginnings of such a service has already been developed for the State of Alaska (see: http://notes.state.ak.us/pn/pubnotic.nsf/). This Online Public Notice web site provides comprehensive listings of proposed rule changes as well as meeting and contract opportunity notices. The listing are searchable and are organized by category, department, location, and publication date. Unfortunately, the service still presents the disadvantage of having to drill down through various departmental information sources in order to be able to identify the rule or notice in which one has an interest. However, once the site is further developed, it should be able to manage an e-mail list of persons interested in being kept informed within particular topical areas. Based on an interview with an employee of Radian, Inc., a firm based in Northern Virginia, there is every indication that this firm has plans to create such a sophisticated notification system for citizens interested in being notified about federal government rule making.

2) Format: The format of citizen participation is also frequently specified in governments' administrative procedures acts. In Georgia, for example, the agency proposing a rule or rule change must "afford to all interested persons reasonable opportunity to submit data, views, or arguments, orally or in writing."

There are additional rules specifying when the government must afford opportunities for oral presentations. Unfortunately, procedural rules such as this one may be unwittingly biased against electronic transmission of views, opinions or evidence. This is the case because it is possible for courts to interpret electronic transmissions as not being writing. Such a situation occurred in Georgia a few years ago when a state department refused to consider a document submitted by fax as being "writing," the format for submittal specified by law. This confusion has since been rectified by new legislation, but the fact that such confusion can occur suggests that the problem faced by public managers wanting to open up rule making to electronic delivery may not always be technical ones. For example, Keith Johnson, the U.S. Department of Agriculture manager who first experimented with electronic rule making at the federal level, had to first get approval of the general counsel before initiating the project.

Johnson's experience with electronic rule making suggests the following lessons regarding the format for soliciting citizen input:

• If one is only interested in quantity and ease of commenting, e-mail is the best vehicle for channeling comments. However, use of e-mail is more likely to reduce the overall quality of the comments and suggestions.
• Using a web form for comment solicitation presents a number of advantages, including:
• The ability to capture name, address, and legislative district information that can then be used to develop customized (e.g., jurisdiction-specific) reports for legislators.
• The ability to structure information in ways that facilitate citizens' understanding.
• The ability to store automatically the information in databases for analysis.

The most important benefit of electronic rule making may come from choosing to publish in real time citizens' input and to allow further posting and comments that reply to this initial posting of citizens' views. This real-time posting format in turn appears to lead to:

• 1) more dialog, feedback and responses among the citizens making comments; and
• 2) reduction in the "wait till the last moment to submit one's comments so that no one has time to respond to one's own arguments" syndrome that Johnson had observed in the typical written comment process.

Johnson also suggests that attempts to have citizen users of electronic rule making technologies identify the specific rule section on which they are commenting will often fail. Users will typically choose the "All Sections" default choice even when they are commenting on a specific section. Users do not want to have to work to make the lives of bureaucrats easier. As such, the intelligence for organizing the comments will likely need to be transparent to users. In this regard, one method that could be tried would be to have hypertext links for web forms or e-mail pop-up messages built into a web version of the document. With these in place a citizen could, for example, click on a link that brought up a form for commenting on that particular section or point. The link itself would contain information about the section, so the user could simply make a statement such as: "I don't agree with this" without having to cite the section itself. On the receiving end the public manger would read something to the effect that "Jane Jones does not agree with the part of the proposed rule (Section 10-b) that calls for limiting the use of pesticides on soils containing more than 20% residuals…"

3) Consideration: Consideration concerns the level of attention that a public employee is required to give to a citizen's views or comments. In the Federal Administrative Procedures Act, the standard for consideration is "fully and completely assess." The parallel act in my state simply asks that the agency "consider fully all written and oral submissions." What this means in practical terms, however, is not always clear. A brief review of cases involving alleged violations of the Administrative Procedure Act indicate that agency actions are only likely to be set aside where the agency has entirely failed to consider an important aspect of the problem or has offered an explanation for

its decision that runs counter to the evidence before the agency ( See Motor Vehicle Mfr. Ass'n v. State Farm Mutual Auto. Ins., 463 U.S. 29, 43 (1983)).

While agencies are unlikely to be reversed by the courts for minor violations of the spirit of the mandate to "fully and completely assess" citizen comments, agency legal counsels are very likely to rule against the use of new technologies that would indicate a purposeful attempt to avoid. Yet as agencies open up their rule making to easier participation by citizens, there may be a need to use automated content processing and analysis technologies to effectively provide full and complete assessment without having to hire dozens of new employees just to read the mail. The problem of giving consideration to each and every citizen comment came to the fore during the first use of electronic rule making at the Department of Agriculture. Keith Johnson reported that his office received 275,000 public comments related to proposed rules on organic produce. This was many hundreds of times the number of comments normally received through the mail. When Johnson's staff had completed assessing the comments, they calculated that approximately 160,000 or about 58% were comprised of cut-and-paste or forwarded form letters that had been sponsored by interest groups. However, a "full and complete assessment" of these letters requires as complete a reading as one would conduct were the comments original. This is the case because without a full reading one cannot be sure that someone's comments do not diverge from the form letter contents half way through the text. Because of the twin factors of "ease of commenting via a electronic networks using a form message" and the obligation to "fully assess" each message, government offices that take the initiative to encourage citizen participation by bringing rule making on-line are essentially punished for their effort. Highly educated and fairly well-paid policy analysts are condemned to read the same text over and over again. This is hardly an inviting prospect for staff, and it represents additional government expenditures.

Because of this dilemma, government agencies have not been quick to adopt electronic rule making. What is needed to address the problem is not self-evident. However, it is likely that some combination of technology and new procedures for citizen on-line commenting will be needed to make electronic rule making more viable and cost effective. A number of possible reforms could be instituted that would have the effect of easing the burden of consideration.

While each of these methods takes a different approach to the problem, the use of artificial intelligence/natural language processing may present the greatest promise in the long run since this method would require little or no change in citizen behavior. However, employment of natural language processing does involve citizens being willing to trust that the technology will lead to equivalent level of consideration to that provided for in current practice. Should we trust such a technology? Unfortunately, unlike data base applications or applications for data analysis (e.g., SAS or SPSS) or even qualitative data analysis (e.g., Nudist), there are no truly off-the-shelf natural language processing programs that can give citizens a good feel for how they work. Moreover, the applications of such technology that citizens are aware of and use (e.g., natural language search engines such as "Ask Jeeves") have not exactly inspired confidence. The reason for this is that natural language processing programs are only as good as the underlying knowledge base. A knowledge base outlines the contexts and relationships among concepts, objects, actions, modifiers, and qualifiers within a knowledge domain. In order for a natural language processing application to be able to simulate (e.g., through accurate summarizing of thousands of documents) the kind of consideration that human bureaucrats can give to a citizen's comments, the knowledge base must first be sufficiently developed. There are currently a few efforts of this type being supported by government agencies such as the National Science Foundation and a number of private firms (e.g., Brightware) that are working with government and other organizations to develop practical uses of knowledge ware. Unfortunately, it is too soon to know whether an agency that invests in the development of a knowledge base for one rule making procedures will then have a sufficiently rich and easy-to-modify template so as to enable inexpensive and accurate use of this technology in numerous other rule making tasks. However, even if the expense and effort involved in developing elaborate knowledge bases delays the full-fledged employment of natural language processing in rule making, simpler versions of these NLP technologies could be put to work in the near future. Drawing on the experience of the Department of Agriculture with electronic rule making, it is evident that the department could have benefited from a relatively simple NLP tool. Specifically, if it had an application that could identify and count all the form letter messages of a particular type, government staff could be freed from this onerous task.

Reporting: Administrative procedure acts also tend to specify the reporting on the adoption of a rule. In Georgia, for example, agencies, if requested to do so, must issue a concise statement of the principal reasons for and against its adoption and the reasons for giving consideration to a specific course of action (see Georgia Code: 50-13-4). At this stage in the process, the effective employment of information technology is contingent on the nature of the rule in question. For example, some rule decisions can be based on an objective analysis of data gathered during the rule making process. In these cases, the government department could provide both the data set and access to freeware analytical tools that would allow citizens to reproduce or extend the analysis that the agency has conducted. Because it is rare that rules are amenable to a definitive analysis, it could be argued that the agency would be opening its decision up to more extensive criticism. However, a good argument can be made that any agency that backs up its support for the values of an open society by offering to put raw data and analytical tools on-line is one that will earn the respect of the majority of citizens and will protect itself against charges of closed-loop decision making. While it is unlikely that such data will be examined by a large number of citizens, the symbolic value of having complete access to the information used by decision makers is probably worth the small cost of electronic publishing.

Electronic policy making is still in the infant stages of development. We can expect, however, that much more will be learned about this process as public managers explore the various combinations of network protocol, data structuring, and automated information processing. Before wholesale adoption of any one electronic policy making process, it would be worthwhile to sponsor studies. One interesting study would be to have two parts of the same agency conduct rule making in both the traditional and in the most technologically-enhanced manner, and then compare the two.

.