Digital Law: Some Speculations on the Future of Legal Information Technology (2023)

Peter W. Martin
NCAIR Sponsored Program on
the Future of Legal Information Technology
May 1995

Introduction

The future of legal information technology in fewer than twenty pages-- however attempted the implicit velocity will be enormous. It will requireso fast a pace my aim can only be to identify some of the larger shapes Iforesee through suggestion and illustration. Careful or detailed developmentare out of the question.

These reflections are divided into five sections. The first and secondcontain a swift look back, positioning the future of this field against itspast. The third identifies, only to finesse, a clutch of important transitionissues that either do or ought to concern the participants and stake-holdersin the legal information system. Section IV sketches four areas of fundamentalchange made possible by digital technology. A concluding section (V) suggestsa few connections between these observations and the field of artificialintelligence and the law.

I. Predigital Law -- The Connection Between Information Technology andLaw (As a Process)

In a recent paper("How TraditionalModels of Distributing Official Information Have Been Influenced by AvailableTechnology"), I surveyed the five century history of printed legalinformation. Doing so I reached several core conclusions about the relationshipbetween information technology and the activity or process we identify as"law." Stripped of all supporting detail they include the following:

First: There is a powerful (but infrequently noticed) linkage betweenways societies think about law and the technology they use in the operationand distribution of it (law that is).

"Legal information technology" or "legal informatics" as it oftencalled outside the U.S. has deep effects on the domain in which we lawyersand legal academics work and on those ultimately affected by law. These effectsare infrequently noticed because for most days and most purposes legalinformation technology is as much in the background and as "transparent"as language, gesture, and even the stereotypes we hold so dear -- especiallyfor those who work with law on a daily basis.

The assertion that legal information technology has affected law doesnot refer to a distinct legal system or a particular corpus of rules andprinciples (the law of privacy or commerce or taxation), but refers insteadto the activity or function or phenomenon we call law, focusing especiallyupon the central role of communication in the law process. It is in thissense of the word "law" that one can observe, for example, that the commonlaw view of appellate decisions as precedent was far more a consequence oftheir distribution in printed law reports than the introduction of printedlaw reports was a consequence of the importance attached to judicial opinionsas precedent.

Second: The current level of change in legal information technology,though both profound and rapid, is on any reasonable historic scalenon-remarkable. Stability has not been a characteristic of legal informationsystems though the dominant impression is to the contrary.

The impression of stability can be attributed to the transparencyof the subject, already noted, and the fact that its fundamental nomenclaturehas remained constant for several centuries. It is all to easy to lapse intothe "book is a book, library a library" fallacy. To say that law has beencarried out using the same book and library technology for five centuriesis like saying that water transportation has not changed over an even longerperiod because it has continuously employed boats.

Third: Through the pre-digital history of this field there has been arecurring focus on the related themes of access and communication.

The present interlocking systems of topically indexed court reports,compiled and annotated statutes, collected and managed by informationprofessionals in libraries with associated finding tools, expert literature,and specialty services has been erected over less than a century. These systemsare at once a consequence of certain notions about the nature of law andof software innovations running on the book hardware (like forward citatorsand topically compiled and codified statutes and regulations, with annotations),but also a very direct product of reform efforts seeking to harness availableinformation technology to improve access to law or (to use a more activenotion) legal communication.

As Marc Galanter has observed, law "usually works not by exercise of forcebut by information transfer, by communication of what's expected, what forbidden,what allowable, what are the consequences of acting in certain ways." Galanterpoints out how such notions as "deterrence," "bargaining in the shadow ofthe law," and "legal socialization" assume an information flow. (Marc Galanter,"The Legal Malaise: Or, Justice Observed," 19 Law and Society Review 537,545 (1985).)

Consider what it would mean for the U.S. legal system if the decisionsof the Supreme Court were not widely available for years, and legislativeacts were available only at a limited number of national centers or libraries,chronologically stored and poorly indexed. Imagine further that policiesand individual determinations by government agencies were available onlyin the offices of the agencies themselves with whatever degree of order oruser assistance they managed to achieve. One doesn't have to travel backtoo many years in U.S. history to find a situation bearing strong resemblanceto this picture or today journey far around the globe. In 1995, there arecountries embracing the "rule of law" where effective distribution of highcourt decisions are a decade or more in arrears. In Tanzania, for example,with financing from the World Bank, the Faculty of Law at Dar es Salaam ispresently undertaking publication of the nation's Law Reports for the period1983-1992.

Since many legal norms do not operate through citizen self application,the quality of communication within the structure of government is vitallyimportant to how law is carried out. In areas like tax and social security,law operates through large government structures, which intersect the livesand activities of huge numbers of citizens. Qualities of performance suchas timeliness, consistency, and equity (like cases treated like, differentcases, with appropriate difference) are strongly influenced by how communicationof governing legal norms is accomplished within these agency structures.In areas of the law where judges or judges and law enforcement officialsare essential elements of the law application process, the concerns are quitesimilar even as the means of communication are likely to be different.

Better access and improved communication have been consistent targetsthroughout the history of printed law -- from Sir Edward Coke who translatedthe classic Littleton's Tenures from "Law French" into English so that itmight be understood "seeing that ignorance of the law is no excuse," throughthe early 19th century statutes that required judges to write out their decisionsso that accurate copies might be distributed in print, the late 19th centurycodification and restatement movements that were premised significantly ona view that law derived from the mosaic of judicial opinions was tooinaccessible, to the Administrative Procedure Act and subsequent "plain Englishregulation" movements of the 20th. In some instances, concern that peoplebe able to know the grounds of their accountability, "ignorance of the lawbeing no excuse," captures the rationale for these reforms, but in many theaims are better understood affirmatively. That is to say whatever aims thelaw is seeking and through whatever intermediate means, the prime instrumentis communication. Efforts to make law more accessible, more understandable,more clearly expressed are ultimately efforts to make law more effectiveand in a democracy, more accountable.

Better information flow leads inexorably to information overload. Thattoo has been a recurring theme in the pre-digital period of legal informationtechnology. The century preceding LEXIS is filled with efforts to organize,to filter, to codify or condense the flood of book-based law.

II. Calibrating the Rate, Force and Direction of Change --
Eventsof the Last Twenty-four Years

A short twenty-four years ago, LEXIS first offered U.S. lawyers acomputer-based federal tax library, comprised of statutes, decisions, andagency material. It was a novelty then, greeted with huge skepticism. Froma present vantage point, one can observe that the introduction of digitallaw set loose momentous and unsettling change. In the decades that followedthe birth of LEXIS, computer-based law systems moved from being powerfulprint supplements used by a few to print replacements relied on by many.Following the initial revolution have followed more recent ones involvingdifferent digital delivery systems, notably CD-ROM and wide area networks.The resulting rate of change has become enormous.

Largely because of this technology shift, the multi-billion dollar U.S.legal information industry is undergoing massive realignment. Emerging arethree or four large enterprises possessing full print and electronic publishingcapability plus a host of new small ones. The digital portion of the legalinformation market has surged well past 50%. Pure print houses are sellingout or scrambling to acquire electronic publishing capability. Several seriousall digital houses have sprung up -- with names like Lawline andHyperlaw. The technology shift has also brought great stress to theunexamined "partnership" relationships formed earlier in this century betweenpublic bodies (which both generate and need legal information) and membersof the private sector legal information industry.

These interconnected changes have fueled battles over the degree of copyrightprotection of legal information and a struggle for media and vendor neutralcitation -- as those favored by past distribution patterns and citation normshave attempted to fighting off new competitors with both copyright claimsand established citation standards keyed to a single benchmark print version.The sums and energy devoted to these "technical issues" leave no room fordoubt that the stakes are very high.

Point and click, graphic user interfaces have come to law bringing imagesand color.

Last and hardly least, the Internet has exploded. When, in 1992, Tom Bruceand I, with NCAIR's support, launched the Legal Information Institute ofCornell Law School, putting the first law server on the Internet, it drewnotice outside the law but mostly puzzlement within. Less than three yearslater several of my colleagues' children have homepages, URLs appear in localnewspapers and mass audience national magazines, the best selling publicationof the American Bar Association is a book entitled: "A Lawyers Guide to theInternet" and public bodies in increasing numbers are placing legal informationdirectly on the Internet.

III. Transition Issues

Any time change occurs at this rate, it raises many complex transitionissues and generates numerous unanticipated side effects. There is a strongtemptation to dwell on them. They are both important and challenging. Theytouch on powerful vested interests, including the legal profession, and ondeeply seated work habits. But these issues are the present not the future.To reach a view of the future, I am going to assume (and ask you to assume)that certain key transition issues will ultimately be resolved responsibly,despite strong countervailing pressures.

What follows assumes that:

• Distribution of legal information in digital form will continue itsrapid substitution for the book/library system.
• Public bodies will rise to the opportunity and challenge, shaking offcozy alliances with individual publishers on the one hand and the temptationto leverage the strong demand for legal information into a revenue streamon the other and come to view the release of basic law data as a core publicfunction, for which they should charge no more than cost and should serveboth an archival and bureau of standards function.
• The private sector will be more competitive than it has during recenttimes, but the competition will emphasize values added over basic accessto legal data.

Place and space will, given this new environment, be far less important-- with disturbing consequences for those favored by proximity under thepreceding system, particularly those with a stake in the information collectionand distribution places we call libraries. I ask you to assume that the publicgains from this shift will be understood as being more important than thelosses suffered by the various interests that have enjoyed locational orprofessional advantage in a world of print libraries.

To reiterate, I ask that you suspend any skepticism or disbelief you mayhave concerning these propositions so we can turn to possible or likely longerterm effects technology will have on law and its central institutions. Inbrushing past these transition matters, I do not mean to belittle them. Ittook a generation after Gutenberg printed the first book before a printerwas able to throw off the idea that a printed book had to look like (andwork like) a manuscript.

IV. Four Important Areas of Potential Change

Bold in the belief that the dimension of change opened by digitaltechnology is enormous -- not simply as a means of doing law, as we haveknown it, at faster speeds, over greater distances, as it surely will, butchange of a more fundamental order -- I shall try now to reach as far asmy imagination can see, at the risk of errors of foolishness or miscalculation.In order to stimulate and provide plenty of target for discussion, I'll bemore specific than general, inviting you to extrapolate my particular pointsto fields you know. Obviously, that gives you another option which is toconclude that a perspective that comes from working with Social Security,Intellectual Property and Land law and in non-profit publishing on the Internetis seriously skewed.

A. Digital technology can and is likely to remove or reduce communicationbarriers hidden in the book/library legal information technology.

Contemporary law books are not very useful except in the contextof large libraries.

Vol. 437 - New York Supplement

This book standing by itself holds little value. It is one volume of overnine hundred delivering the output of New York Courts to date. This one containsa thousand or so decisions rendered by New York Court in 1981. Without itscompanion New York Supplement volumes, the 13 volumes of Shepard's Citationsreporting which of its opinions were subsequently reversed or distinguished,bookcases full of annotated New York Statutes and Regulations, book accessto New York legal information is incomplete. In fact, given the importanceof federal law on one side and municipal law on the other to such topicsas civil rights, business finance, or real estate development, such a collectionwould be incomplete for a citizen or enterprise or law firm concerned simplywith New York State. For a regional, national, international actor accessto appropriate legal information in book form means access to a very largelibrary.

Such libraries do exist, and they provide up-to-date and comprehensivelegal information of an unparalleled quality to those who work close by themand who possess the necessary skills to use their resources.

Plain language you cannot access does not communicate effectively.

The information system law libraries embody works far less well forordinary citizens and all who live and work any distance away. The publicinstitutions founded in the U.S. to expand access to legal information -notably court and county law libraries - have been seriously underfunded.They have, as a consequence, inadequate collections, staffing, and services.

By contrast the exploding reach of the Internet promises to bring an enormouscollection of legal materials into small offices and schools and homes invery remote settings.

The links that work so efficiently for professionals (legal citations)are written in a code few non-professionals can read.

There are greater barriers than miles separating most citizens fromrelevant and important legal information. Ethan Katsh's new book, "Law ina Digital World" details how law delivered through law libraries remainsat a great informational distance from non-professionals, no matter how closethey may be geographically. Encoded in professional language and linked usinga citation system that requires special skills to follow, law book/law libraryinformation is distant from even highly literate citizens.

Decisions of the U.S. SupremeCourt Accessible Through the LII World Wide Web Site

In a point and click networked environment a high school student orprofessional in some field other than law can retrieve particular decisionsby aSupreme Court justice or follow aprecise statutoryreference without knowing "legal citation."

The same ease of use considerations make it possible for disk-based materialsto open a path for non-experts from broad description of a legal point tothe most specific regulations or a closely analogous case. That path cangive the user a degree of control over direction and depth of research thatonly very skilled researchers can achieve using print.

Point and Click Navigation Can Bring a User of SocialSecurity Plus To a Relevant Text
Overview and Fromit to the Cases, Statutory Provisions, and Regulations Most Directlyin Point

B. Digital technology can and is likely to loosen the law's tie to words.

The use of color and image in law: of green-gold pressing pads,a "festive eating atmosphere", and the flag

On March 28 the U.S. Supreme Court delivered an important trademarkopinion inQualitex Co.v. Jacobson Products Co. (No. 93-1577). Writing for a unanimous Court,Justice Breyer construed the language defining the reach of theLanham Act --"word, name, symbol, or device" -- as encompassing color. Qualitex had registered"a special shade of green-gold" as a trademark for pads it sold to dry cleaningfirms. The litigation that brought this issue to the Court arose when acompetitor, Jacobson Products, began to sell pads of a similar color. Priorlaw on this point was far from settled, but market realities had alreadybroken down narrow readings of the Act. As Justice Breyer noted: "The courtsand the Patent and Trademark Office have authorized for use as a mark aparticular shape (of a Coca-Cola bottle), a particular sound (of NBC's threechimes), and even a particular scent."

Does it seem natural to you, or as discomforting as it has come to seemto me, that the Court's decision explores and rules on the role of colorin identifying the Qualitex brand of pressing pads without including a fullcolor picture of this "green-gold" object and its "similar" competitor? Orthat the presentation of the issue to the Court was similarly "colorless"?Among the arguments rejected by the Court was that of "shade confusion" --one resting on skepticism about the competence of courts to distinguish betweenshades in infringement suits. Was the legal conclusion sounder for havingbeen rendered without having looked at the "confusingly similar" coloredpressing pads?

During oral argument there was a largely frustrating exchange betweenthe justices and counsel over whether trademark registration of colors mightever proceed without a showing of "secondary meaning." It took the form ofhypotheticals framed ineptly by such words and phrases as "weird" colors,"lime yellow", and "hot neon pink." Visual examples were not used.

The Court's earlier decision inTwoPesos, Inc. v. Taco Cabana, Inc., 112 S. Ct. 2753 (1992) dealt withthe question whether "inherently distinctive" trade dress is protected bysection 43(a) of the Lanham Act "without a showing that it [has] acquiredsecondary meaning." The factual background of that case was a jury findingthat Taco Cabana's restaurant decor had not acquired secondary meaning butwas "inherently distinctive." About this "inherently distinctive" trade dressthe Court tells us:

"[It consisted of] a festive eating atmosphere having interiordining and patio areas decorated with artifacts, bright colors, paintingsand murals. The patio includes interior and exterior areas with the interiorpatio capable of being sealed off from the outside patio by overhead garagedoors. The stepped exterior of the building is a festive and vivid colorscheme using top border paint and neon stripes. Bright awnings and umbrellascontinue the theme."

Would not this critical concept of trademark law, largelyworked out in connection with words and phrases, have been more effectivelypursued by the Court had it been able to see and use graphic images? Aboutthe defendant's restaurant the Court says merely: "Two Pesos adopted a motifvery similar to the foregoing description of Taco Cabana's tradedress." [Emphasis added.]

The world to which law must relate has color, shape, and texture. Lawthat speaks almost exclusively through words and without color is severelyhandicapped (more in some areas than others).

Legal systems that take in information, to inform judgment or to recorda situation or individual claim, are handicapped as well by the dominanceof words over image -- with perhaps even greater consequence. Given thatcolor can, under Qualitex , be a registerable trademark and can, aswas firmly established before, be a significant element of a design registeredas a mark, how does the registrant submit the mark for registration? Theregulations call for drawings of a mark -- black ink, no shadings -- andthen lay out a crude system for encoding color in black and white. See37 C.F.R. § 2.52 (1994). At this one of several places in the Code ofFederal Regulations where color is the very subject of a legal claim, linerepresentations of color are used.

How text bound is current U.S. law? Out of 72,161 federal court decisionsfiled in 1994 and loaded into LEXIS only 130 included a chart, map, photographor other graphical element. No sections of the U.S. Code directly includesuch material (although specific maps are occasionally incorporated byreference), and only 913 sections (out of over 200,000) in the Code of FederalRegulations do.

At the risk of trivializing this point as we leave it, let me quote thetext of 4 U.S.C. §§ 1 and 2:

§ 1.
The flag of the United States shall be thirteenhorizontal stripes, alternate red and white; and the union of the flag shallbe forty-eight stars, white in a blue field.
§ 2.
On the admissionof a new State into the Union one star shall be added to the union of theflag; and such addition shall take effect on the fourth day of July thennext succeeding such admission.

The thought experiment I ask you conduct is this: Imaginesomeone being handed this text and being instructed to make or draw a flag.Better yet imagine someone being prosecuted for mutilating one of these orusing it in an advertisement.

I have read about but not seen a 16th century Flemish criminal law bookthat contained a beautifully detailed engraving depicting each crime itdiscussed. My experience as a law student offered nothing approaching it.During recent centuries pictures and especially images including color havebeen squeezed out of ordinary law books, statutes, regulations, and evenjudicial opinions by very practical factors -- sometimes at great sacrificein clarity of communication.

Digital law can change that. Today, ordinary desktop (and even laptop)computers deliver color and in line images, not just from local files butvia the Internet from anywhere. Color printers now cost no more than an IBMselectric used to. What might it mean for law to have images and color addedto its language? Ethan Katsh writes that: "Illustrations are not a part oftraditional legal education because they would make it more difficult forlaw students to develop the habits of thought that are considered to be atthe core of being a lawyer." ("Electronic Media and the Transformation ofLaw", at 257.) While I am less confident in attributing aim or purpose, Iagree with Katsh that the effect has been profound: "The effect of excludingimages, full-colored images, including flowcharts, and diagrams and illustrationsof accident scenes or allegedly defective products, or the dust-casting cementplant set against its surroundings, is to diminish the effectiveness of ourconsideration of important legal issues and to push law toward the abstractor conceptual." And therefore I agree with Katsh that bringing visual elementsinto law will bring important change. To conclude this point, once more usingKatsh's words, (Id at 262.): "As limits on the form of informationare relaxed, limits on thought are lifted as well. ... The information-reductionfunction of print that fostered the development and use of abstractions willbe challenged as we are given more information to think about and someinformation that, like music, poetry, or drama, may touch us or move us inways that are difficult to articulate."

Sound

Campbellv. Acuff-Rose Music, Inc., decided in March of 1994, 127 L. Ed. 2d500, is an important Copyright "fair use" decision. The litigation that broughtthe issue to the Supreme Court arose out of a 2 Live Crew commercial recordingof the Roy Orbison song, "Oh, Pretty Woman." Acuff-Rose, copyright holdersof the Orbison song, had refused permission for the recording. 2 Live Crewasserted that their version of the song was a parody and that as a parodyit was protected by the doctrine of "fair use," notwithstanding its significantcommercial success. Among the elements of the "fair use" analysis appliedby Justice Souter's opinion for the Court is an investigation of "the amountand substantiality" of the portion of the copyright work that has been reworkedin the parody. The Court's analysis of this issue not only concerns the lyricsof the two versions (which it attached in an appendix to the decision) butthe different sound of the "parody." We are told by Justice Souter that "2Live Crew not only copied the bass riff and repeated it, but also producedotherwise distinctive sounds, interposing 'scraper' noise, overlaying themusic with solos in different keys, and altering the drum beat." While bothradio and TV news broadcasters were able to and did illustrate the Court'sdecision the evening it came down with recorded excerpts from both versions,the Court was limited by the technology available to it to reprinting thelyrics. Indeed, it appears likely, though not certain, that the members ofthe Court decided the matter without themselves having access to the originalsong or the parody.

Interaction

There are many law domains in which substantive rules and even morecommonly procedural ones lay out a process, with sequence, a time line, branches,conditional elements. Representing such a dynamic flow in text gives us someof the intricate cross-referenced texts law teachers drill their studentson -- often, I might observe, with the help of a schematic or flowchart.Hypertext technology allows us to see that printed text [being flat, linear,static] is clumsy in its representation of process and dynamic relationship.Add color, graphics, and sound and you have vastly superior capability forrepresenting diverse legal topics ranging from how a determination of eligibilityfor Social Security disability insurance should be made, through the resolutionof competing priorities under Article 9 of the U.C.C., to consideration ofentity alternatives for a start-up business.

A Point and Click Practice Guide from Social SecurityPlus

We will, I predict, see exciting experimentation in the use of interactivediagrams of law coming out of the current rush of commercial law publishersto CD-ROM. And government agencies should not remain forever stuck in observerstatus, limiting themselves to print bulletins, brochures, forms, and"regulations."

The static quality of printed texts severely limits their capacity tocommunicate changing law. The best printed codes provide a strobe-lit frameof the provisions in effect at a particular moment together with editorialnotes that allow readers to assemble the set of provisions that applied totransactions, persons, events of a prior time or those that will operatein the future. In many legal areas, important current problems must be solvedwith "old" law. Planning for next year and beyond requires an eye on lawto come. Imagine what it will mean to have code representation systems thatcan assemble the set of operative provisions in real time as a user pursuesa problem.

Legal Algorithms

Finally, there are numerous settings where legal rules are obligedor strongly pushed to describe a surface that is not flat, a line that isnot straight, or a mathematical function that is non-lineal. Working withwords and numbers performing basic mathematical operations, the results tendtoward a clumsy compromise. Digital law can contain more flexible algorithmswith accompanying representation of their operation.

C. Digital technology can and is likely to improve law's implementationby enabling better/faster communication of legal information within governmentagencies (and courts).

The many consequences that flow from the difficulties of communicatingwithin government using print

Those governed by the law are "assumed" to know it. Perfect communicationis hardly reality, but "ignorance of the law is no excuse." With varyingdegrees of acknowledged discomfort, we operate with fictions like assumedknowledge of the law and "legal notice," speedy trials, and fast-track agencyreview. And we deal with error within the system using after the fact reviews,that carry both delays and other costs that cause them to be underutilized.

Those who apply the law (judges and administrators) are in one sense heldto the same standard, for to the extent their work is subject to review byhigher authority, their interpretation or application or knowledge aboutthe law can be upset without pause or deference.

Combined these two propositions pose serious risks to even the mostconscientious citizens. Consider any one of several important legal regimesthat have the following characteristics: an intricate set of rules that arechanged from time to time, a large to enormous effected group with fair amountsat stake. (The set includes tax laws federal and state, farm subsidies, floodinsurance guarantees, social security and medicare.) Such programs involvemore than a rule promulgator, those affected, and a cast of adjudicators;they are implemented through agencies. Spreading legal information withina large and geographically distributed organization has in pre-digital timesrequired training programs, supervision, spot checks or quality controlmonitoring, and massive manuals. The POMS or Program Operations Manual Systemof the Social Security Administration fills a large bookcase in my officeand is regularly cursed by the member of our library staff who is responsiblefor keeping it up to date.

Most citizens have no alternative but to rely on a government agency ormore accurately whichever one of its people they encounter to be both accurateand authoritative on questions involving "its" law. But, of course, systemsfail and in most instances U.S. law today casts the risk of such failureson the citizen. In 1990, the U.S. Supreme Court held, in Office of PersonnelManagement v. Richmond, 496 U.S. 414, that a former civilian employeeof the Navy was ineligible to receive six months of a disability annuitybecause he had exceeded the statutory limit on earnings -- even though hehad done so only after seeking advice about that limit from Navy employeerelations personnel and receiving assurance in writing that his earningswould not cause benefit loss. As it had in an earlier case involving SocialSecurity, the Supreme Court ruled that "erroneous oral and written advicegiven by a Government employee to a benefit claimant" does not bind thegovernment. When the initial error by a public law-applier leads to paymentsof an excessive amount or undercollection of tax the result is typicallysimilar; the citizen bears the risk of errors of law within government agencies.

Through better information and communication it should be possible bothto reduce the incidence of government error and, with better accountability,to increase the authority of front-line government workers to issue bindinginterpretations of the law.

Digital information technology and the flattening of legal hierarchies

The work of Shoshana Zuboff ("The Age of the Smart Machine" (1988))and others strongly suggests that digital information technology has a corrosiveeffect on top down control within organizations. Information can be and thereforetends to be more widely distributed which invites non-managers to "a broaderview" of the enterprise and a "deepening understanding" of their own tasksand roles. Thinking of this tendency in relation to public bodies ratherthan the financial institutions, paper mills, or auto companies studied byZuboff, I see serious possibilities for "reinventing government." Top down,rule-based control, is the model on which we have constructed both industrialenterprise and the administrative state. It is what the phrase "rule of law"means to many. Recently, at a bookstore I saw: Philip Howard's "The Deathof Common Sense" and Nicholas Negroponte's "Being Digital" side-by-side.For me the books are, indeed, related. I share Negroponte's optimism aboutdigital technology and believe it permits, though hardly assures, governmentmeasures that are less rigid, compartmentalized, and hierarchical and facilitatesmore richly informed discretion. The layers, delays, departmental boundaries,and rigid rules against which Howard rails are, in my view, tightly associatedwith the use of print-based systems to communicate law.

D. Digital technology can and is likely to narrow the gap between "lawon the books" and "law in action."

Reading the law in outcomes not just in "precedent"

An increasing fraction of the decisions of the U.S. Courts of Appeals(and indeed many appellate courts throughout the United States) are denominated"unpublished" and "non-precedential" by the deciding court. According todata from the Federal Judicial Center, the percentage of "merits dispositions"by Courts of Appeals that are "unpublished" climbed from 58% in 1987 to 66%in 1991. So long as the only effective distribution channel for the decisionsof these courts were books published by the West publishing company, thedecision not to publish effectively kept a decision the court thought ofas adding little to established law from entering the legal information system.The on-line systems changed all that. A very high percentage of the "unpublished"Court of Appeals decisions are distributed in full text by LEXIS, all ofthem carrying warnings similar to that used for ninth circuit decisions:"THIS DISPOSITION IS NOT APPROPRIATE FOR PUBLICATION AND MAY NOT BE CITEDTO OR BY THE COURTS OF THIS CIRCUIT EXCEPT AS PROVIDED BY THE 9TH CIR. R.36-3."

By my calculations, 64 percent of the Court of Appeals decisions in LEXISfor 1991 carried such a label. Why does LEXIS load decisions that courtsdeclare non-precedential and forbid parties to cite? Quite simply becausethose decisions have value to lawyers, albeit for a relatively short periodof time. Lawyers are eager for any information that will allow them to reada court, not just the court's precedential opinions.

Greater consistency through better feedback

Finding appropriate means to provide consistent but equitable outcomesin a high volume adjudication system (whether judicially or administrativelyrun) is a challenge that neither top down rule or precedent guided approachesaddress satisfactorily. Instead of pulling you into the details the of theU.S. Social Security disability determination process, I'll try to illustratethe point with the more widely visible issue of criminal sentencing. In themid to late eighties, the U.S. was not alone in focusing on the seriousdiscrepancies in sentencing outcomes that appeared when one studied or reviewednot individual sentencing decisions but decisions coming from all judgesof a particular court over a period of time. Through great controversy andwith equally uncertain success, the U.S. dealt with this problem by establishingsentencing guidelines.

In New South Wales, Scotland, and British Columbia a quite different approachrelying on putting better information in the hands of those involved insentencing decisions has been the subject of reform and experiment. New SourthWales offers the world a working model of this approach. A Sentencing InformationSystem launched there in late 1990 provides several complementary databasesto sentencing judges. Two contain law data of traditional sorts -- namelythe statutory provisions laying out the range of dispositions and factorsto be considered and appellate decisions dealing with sentencing. A thirdprovides information on the availability of special facilities or personnel,relevant, say, to including assignment to a drug rehabilitation program aspart of a sentence. A final "Penalty Statistics Database" is for our purposesthe most interesting for the premise on which it rests is that the hundredsof thousands of sentencing decisions made by sentencing officers in New SouthWales along with reasonably full offense and offender information viewedthrough effective statistical analysis tools and graphic display should beavailable to guide the judge's selection of an appropriate penalty in aparticular case.

I predict that the future of legal information will include distributionof much more outcome data representing full populations of decisions, notjust those explained by the decider in a form picked up for "publication".

The data on all dispositions in civil matters filed in the Federal courtsrecently opened to the Internet by Cornell's Legal Information Instituteis suggestive in this regard, but incomplete as a model. While it offersnew and exciting opportunities for statistical analysis -- comparing jurytrials and bench trials for duration or plaintiff success, for example --it provides no way to review outcomes by judge. For a brief time in the eightiesthe Social Security Administration kept outcome data on Administrative LawJudges deciding disability insurance and old-age and survivors insurancecases, which members of the public (including lawyers) were able to secureunder the Freedom of Information Act. I foresee widespread implementationof the idea that adjudicators can and should be known by the pattern of whatthey do, not simply by what they write in explanation of some decisions.The perceived advantages that go with the knowledge of a decider's propensitiesobtained by repeat players and less reliably through hearsay suggest theimportance of such information to litigants. I believe that we will growcomfortable with the idea that an independent adjudicator can, consistentwith that independence, be reviewed on the basis of a pattern of performancenot just in individual instances, evaluated in isolation, upon appeal.

Applied to agency rulings on individual cases this relates back to theearlier point about technology's push toward flatter but more effectiveadministrative bodies.

Tax law changed dramatically in 1976 when, following rulings by the U.S.Court of Appeals for the D.C. Circuit and other federal courts holding thatletter rulings were subject to the Freedom of Information Act, Congress addeda new section to the Internal Revenue Code, 26 U.S.C. § 6110. It providesfor access to such agency documents (letter rulings and technical advicememoranda) and their "background file documents" but also states that "Unlessthe Secretary otherwise establishes by regulations, [they] may not be usedor cited as precedent." Id. § 6610(j)(3). Courts are nonetheless referredto them and do, on occasion, draw upon them for evidence of agency practiceor interpretation. See Rowan Cos. v. United States, 452 U.S. 247, 261 n.17 (1981). More importantly there is strong evidence the IRS itself seeksto use such rulings as an instrument of internal consistency. And in thelight of that tax practitioners now routinely use them in providing counsel.What has allowed both agency and tax lawyers to use this unwieldy corpusof agency law is technology that arrived more or less with the prying ofletter rulings from the IRS -- namely, digital storage, search, and retrieval.For over a decade, they have been available on-line, first on LEXIS andsubsequently WESTLAW, as well.

"No action letters" by the SEC represent a similar phenomenon and aninstructive information story. These letters (and the related interpretativeletters) are written by the SEC staff in response to specific requests andindicate that on the basis of facts presented the agency will not view particularactions or transactions as falling within a statutory requirement to register.Being written by staff (and not the Commission) they don't bind the government,not even in the matter on which they were issued. See Loss & Seligman,"Securities Regulation" 524 n. 29. On the other hand, the Securities lawliterature stresses that they are not only honored by the SEC in the particularmatter but trace the SEC staff's unfolding interpretation of registrationrequirements in gray areas. Important topics of securities law previouslyunclear have been mapped in detail through a succession of no-action letters.

These theoretically non-binding but practically highly reassuring letterscome at a substantial cost -- delay. Yet they are sought and issued at arate averaging well over 1,000 per year. To those who see law only in statutes,regulations, and binding agency or judicial precedent these do not qualify.The academic lord of securities law, Louis Loss, pronounced them: "ninetypercent ... sheer, unadulterated, repetitious garbage." But to those counselingcommercial concerns they constitute a valuable information asset. Where academicexperts may see needless repetition, others find comforting company.

In pre-digital days, several large NYC firms doing high volume securitieswork created a pool of the no-action letters they had received for clientswhich they (the participating firms) but no others could consult in advisingclients. Like the IRS letter rulings these were in time pulled into morepublic distribution, initially in a printed loose-leaf service, then throughthe two on-line services WESTLAW and LEXIS, and most recently on CD-ROM.(Except where confidentiality reasons force a longer delay, the letters aremade available to the public 30 days after their issuance. 17 C.F.R.§200.81.) In the latter form particularly, they have become a widelyused resource -- widely used and easily usable by those who are not themselveseither repeat players with the SEC nor securities specialists. As an insurancecompany's general counsel explained to me recently -- having this materialin digital form has allowed her office to give management swift advice aboutfinancing alternatives without having to turn to expensive outside securitiesfirms.

Digital technology makes it possible for agency legal interpretationsto be at once offered to those who seek reassurance in advance of structuringa transaction or employment practice and for such advisory rulings to beavailable to others needing to "read the agency" not just its formallypromulgated policy. High stakes corporate tax and securities law issues providethe most highly developed examples of this approach, but opinion lettersof the Labor Department's Wage-Hour Administrator dealing with Fair LaborStandards Act points suggest that the greater gains may be realized by extendingit to lower stakes but widely recurring legal issues.

V. Possible Implications for A.I.

When the field of Artificial Intelligence and the Law first emergedas a discipline, it was digital science applied to a print-based activity.That reality is rapidly changing. The field's future is surely as a digitalscience applied to work carried out through digital information and communicationsystems. This transformation will be at once liberating and challenging.

A.I. will be burdened less and less by the problem of applying digitalscience to print bound information.

One of the major limits on industrial strength AI projects has beenthe need for researchers to digitize law data required by their work. Solong as AI researchers are prepared to go where the data is (i.e. allow theirprojects to be drawn to areas with digitized law data) that is or will soonbe no problem.

A.I. will have increased opportunities to be "relevant" to Legal informationend users and legal information system authors/builders/publishers (toolsfor intermediaries rather than law knowbots for every citizen)

As digitized law data proliferates there is a greater and greaterneed for intelligent agents to organize (and filter) it on the user's behalf.As some would paint the future scene we will all have software agents consciousof our ongoing information needs harvesting new material of importance tous and also ready to venture off on our bidding to acquire information pertinentto an arising need. But a more plausible future, certainly one for whichthere is great demand today, is for the agent's agent (tools that serve authors,editors, lawyers or judges).

The risk that A.I. must avoid, given the rate of change in the legalinformation field, is the risk of addressing law as it has been rather thanlaw as it is becoming. That necessitates taking the communication elementseriously, as well as the changes sketched above.

As Kevin Ashley wrote in the introduction to "Modelling Legal Argument": "Expert systems designers need to focus on improving a system's abilityto explain its advice." Explanations are both critical to understanding thesystem's advice and a means of evaluating its quality. Expert systems mustalso focus on improving pre-advice communication, the interaction with theuser that brings a problem or transaction or case into the system's solutionspace.

As an interested and friendly observer of the field of Artificial Intelligenceand the Law, I see a deep tension in the field over whether it should bepure science or applied, despite the frequent assertions that it can be both.Some have abandoned the term AI altogether because of this duality. DouglasHofstadter takes pains in his new book to explain why he now chooses to callhis field "cognitive science" rather than artificial intelligence. Perhapsbecause law itself as a domain has long been comfortable with working theoryand practice in joint harness, those doing A.I. and Law remain relativelyoptimistic about the connection between understanding and modeling legalargument and legal reasoning and building useful computer systems.

But that straddle, with its confidence that those goals can be pursuedsimultaneously, will become more difficult if, as I suspect, it developsthat the more useful computer systems in law require a scale quite differentfrom the individual human intelligence doing legal reasoning. The modelunderlying much AI work including that in law is of a computer system doing(to use the words of Marvin Minsky) "things that would require intelligenceif done by a man." The serious limitation of that view, as I see it, liesin the greater value and opportunity for improving legal systems throughbetter information and communication systems than through the constructionof software that will simulate some portion of the functionality of individuallawyers or judges no matter how expert.

Finally, those working on A.I. and the law in university departments mustfight off the habit of defining the field as bounded by the academy. It oncewas, but no longer. Unless A.I. work in law is carried out with full appreciationof all that publishers and governments are doing to deliver law effectivelyin digital form it will be increasingly irrelevant.

"Turbotax" Tax Preparation System

Four million Americans used software like this to prepare and review theirincome tax returns this year. Many them used it to file electronically orby means of the 1040 PC print data form. A definition of Artificial Intelligencethat ignores the elements of expert system now showing up in commercial productsand agency systems is a definition that denies AI a significant future.

Closer connections between those entities developing and deploying thenew legal information technologies and those doing applied work in A.I. andthe law is of mutual benefit. But in the current turbulent environment, theyare not likely to develop spontaneously.

VII. Conclusion

To recapitulate:

• Digital law should be far more effectively and equitably distributed.
• It is likely to be colorful, tuneful, even noisy, and capable of movementand interaction (i.e. fully communicative).
• It may be characterized by flatter administrative agencies -- with lesstop down control, delay, and rigidity, but greater consistency.
• And finally it should offer both greater opportunity and greater influencefor those working in Artificial Intelligence and the Law.

In conclusion I wish to acknowledge and thank our patron -- The NationalCenter for Automated Information Research -- for sponsoring this forum, forsupporting my Social Security CD-ROM project in its critical first year,and for furnishing the start-up funding that has allowed the Legal InformationInstitute at Cornell to explore the value of Internet and hypertext on diskas means of delivering legal information in new ways to old customers andeven more importantly to a wide variety of new ones.