1 The growth and development of clinical legal education has been reported in annual surveys published by the Council on Legal Education for Professional Responsibility, Inc., New York City, from 1971 to 1979, under the following titles: Survey of Clinical and Other Extra-Classroom Experiences in Law Schools: 1970–1971; Survey of Clinical Legal Education: 1971–1972, 1972–1973, 1973–1974; and Survey and Directory of Clinical Legal Education: 1974–1975, 1975–1976, 1976–1977, 1977–1978, 1978–1979. A recent AALS-ABA report gives the following historical overview:. Clinical legal education is now a part of the curriculum of nearly all law schools…. Although only fourteen subjects in the law school curriculum were the subject of clinics in 1970, the range of subjects continues to grow; by 1978–1979, fifty-nine subject areas were involved in clinical fieldwork. The number of programs in which student fieldwork is completely supervised by the law school has also been increasing and now includes over one-half of the country's clinical legal courses. (Footnotes omitted]. Association of American Law Schools-American Bar Association Committee on Guidelines for Clinical Legal Education, Clinical Legal Education: Report of the Association of American Law Schools-American Bar Association Committee on Guidelines for Clinical Legal Education, at 7 ([New York: Association of American Law Schools-American Bar Association Committee on Guidelines for Clinical Legal Education], 1980) [hereafter cited as Guidelines Report].Google Scholar

2 The analogy to the teaching hospital is by no means of recent vintage. See, e.g., Frank, Jerome, Why Not a Clinical Lawyer-School? 81 U. Pa. L. Rev. 907, 917 (1933). But see Packer and Ehrlich's comment, “The analogy is not particularly apt, however, for in law there is no analogue to the hospital, and hence none to the teaching hospital,”in Herbert L. Packer & Thomas Ehrlich, New Directions in Legal Education 39 (New York: McGraw-Hill Book Co., 1972). For a strongly worded reply to Packer and Ehrlich see Leleiko, Steven H., Clinical Education, Empirical Study, and Legal Scholarship, 30 J. Legal Educ. 149, 159–65 (1979).Google Scholar

3 Leleiko, supra note 2, at 158–65. There is extensive discussion of the variety of clinical methodology in the Guidelines Report, supra note 1. The report reviews clinical teaching methods, including observation, role playing and simulation, performance of lawyering tasks by students on live cases, individual discussion and counseling sessions, classroom instruction, and video tape review and analysis. Id. at 14–24, 47–82. The following Consultants' Reports, printed as Part III of Guidelines Report, also analyze various clinical teaching methods: Joseph D. Harbaugh, Simulation and Gaming: A Teaching/Learning Strategy for Clinical Legal Education, at 191; Robert Bastress & Joseph D. Harbaugh, Examining Lawyers' Skills, at 223; Elliott S. Milstein, The Design of the American University Criminal Justice Clinic, at 238; Harry I. Subin, Clinical Pedagogy—the Educational Program of the New York University School of Law Criminal Law Clinic, at 254. Milstein states:. Any discussion of the design of an on-going clinical program necessarily represents work-in-progress rather than a finished product. The program is different every year, perhaps every day. Each change in persounel, in the legal environment, in the dynamics of the student group, in the teaching materials, and even in the office space available for the clinic affects the execution of its goals or necessitates their re-definition. Guidelines Report, supra note 1, at 253.Google Scholar

4 The computer “assembles” a document by piecing together parts of boilerplate text and items of client data to form a document that is customized to the needs of a particular client. The boilerplate text is extracted from form documents stored within the computer's memory, and the client data are gathered during an “on-line” interview. The entire process is controlled by the form documents themselves, and no computer programs are required. See infra, sec. II.Google Scholar

5 See generally James A. Sprowl, Automating the Legal Reasoning Process: A Computer That Uses Regulations and Statutes to Draft Legal Documents, 1979 A.B.F. Res. J. 1. The ABF processor is also intended to be used for computer-assisted instruction, but this experiment tested its effectiveness only as an automated tool for drafting legal documents in a law school teaching clinic environment.Google Scholar

6 See summary description of prior tests and training efforts in Sprowl, supra note 5, at 77–79. See also sec. IIIC, infra.Google Scholar

7 Boyer and Cramton noted in their 1974 survey of research on legal education that there is a need to learn more about how law professors and law students spend their time. Boyer, Barry B. & Cramton, Roger C., American Legal Education: An Agenda for Research and Reform, 59 Cornell L. Rev. 221 295–96 (1974). Assuming that methods for controlling the accuracy of the data accumulation process can be implemented, computer management systems such as the one we have developed may prove to be useful vehicles for gathering nationwide data on the activities of clinical educators and students. Our computer management system is described in a companion article to be published later.Google Scholar

8 A more extended introduction to computers as they relate to law can be found in Sprowl, James A., Evaluating the Credibility of Computer-generated Evidence, 52 Chi.-Kent L. Rev. 547 (1976).Google Scholar

9 In the ASCII (American Standard Code for Information Interchange) code, every printable character and a number of commands are represented by numbers between zero and 127. The number 65 is assigned to A, the number 66 to B, and so on. See American National Standards Institute report ANS-X3. The ASCII code is adhered to by most terminal manufacturers and by many computer manufacturers, including Digital Equipment Corporation, IBM generally adheres to the ECDIC (Extended Binary Coded Decimal Information Code) code in which printable characters and commands are represented by numbers between zero and 255 and in which 193 is assigned to A, 194 to B, and so on. Many terminals support the EBCDIC code as well as the ASCII code. Control Data Corporation has its own extended display code which includes the numbers zero to 63 and which represents printable characters and commands by sets of one or two of these numbers. In this extended display code, I is A, 2 is B, and so on. These three are the dominant code conventions at the present time. The incompatibility of these three code conventions is one of many reasons why it is difficult to move programs from the computers of one manufacturer to those of another.Google Scholar

10 The computer will prepare reports summarizing how much time each attorney spent working on each case each month.Google Scholar

11 The computer will report on the cash disbursements made for each case each month.Google Scholar

12 Client bills will be written including varying amounts of detail from the attorney time and cash disbursement information.Google Scholar

13 Searching through abstracts of litigation exhibits and through deposition transcripts, the computer will facilitate the management of large amounts of evidentiary information.Google Scholar

14 The lawyer's computer will communicate with the court computer by telephone to make docket book entries and also to file motions and briefs.Google Scholar

15 A computer conference is a discussion conducted by parties who leave each other messages within a computer. Much of our present-day motion practice could be carried out using a computer-conferencing arrangement established between the judge and the attorneys for both sides. One attorney would file a motion by submitting it to the computer, and the motion would be delivered to the opposing attorney the next day. The opposing attorney would then file a “response” with the computer. The next day, both the motion and the response would be placed before the judge who would type an “order” into the computer for delivery to the parties the next day. By avoiding court appearances and by never requiring the judge and the attorneys to physically meet such motion conference arrangements could significantly reduce the cost of pretrial practice.Google Scholar

16 Such a module is called a compiler if it actually translates an English-like language into numeric instructions. But often such a module merely “interprets” the English-like language and then tells the computer what to do, in which case it is called an interpreter or interpretive processor for the language.Google Scholar

17 The term processor is often used interchangeably with the term computer. We use processor here in a limited sense. In this article, a computer is a machine that becomes a processor for a specific English-like language when the computer contains a module designed to translate the English-like language into the numeric commands that can control the computer.Google Scholar

18 The most commonly used English-like languages have been COBOL (for business and accounting applications), FORTRAN (for scientific and engineering applications), PL/1 (supported primarily by IBM), BASIC (for time-sharing computers and microcomputers), PASCAL (widely used by computer scientists on microcomputers), ADA (a grandchild of PASCAL, sponsored by the United States Department of Defense, which may soon emerge as the most widely used language), APL (a very powerful, concise language used by many computer scientists), SNOBOL (a language for manipulating strings of text), and LISP (a language used in designing “artificial intelligence” systems that simulate human thinking). While standards exist for most of these languages, the versions supported by different computer manufacturers are not always compatible—another bar to moving programs from one computer to another. For more information on these and other computer languages see sources cited in Sprowl, supra note 5, at 14 n.23, and Peter Grogono, Programming in PASCAL (rev. ed. Reading, Mass.: Addison-Wesley Publishing Co., 1980); Laurent Siklóssy, Let's Talk LISP (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1976); and Peter Wegner, Programming with Ada: An Introduction by Means of Graduated Examples (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1980).Google Scholar

19 Sprowl, supra note 5.Google Scholar

20 This definition of the term algorithm is essentially the meaning that is taught to students of computer science. A much more precise definition of the term can be found in Bryan Niblett, Legal Protection for Computer Programs 13 (London: Oyez Publishing Ltd., 1980), where Niblett, a barrister as well as a professor of computer science, defines an algorithm as. a set of systematic rules or statements for solving a problem encountered in the design of a program. In his work on the art of computer programming, Knuth … says: ‘The modern meaning for algorithm is quite similar to that of recipe, process, method, technique, procedure, routine except that the word “algorithm” constitutes something just a little different.’ Knuth goes on to distinguish five important features of an algorithm: that it contain a finite number of steps; that each step be precisely defined; that it have zero or more input quantities; that it yield one or more output quantities; and that it be effective in the sense of consisting of basic steps which can be carried out exactly in a finite time.Google Scholar

21 Discussed in Sprowl, supra note 5, at 36–37.Google Scholar

22 The document is typed into the computer and is stored within the computer as a single unit, no matter how long it may be. Northwestern's computer stores the document on a spinning magnetic disk. At midnight, the Northwestern computer automatically produces two magnetic tape backup copies of each newly entered or altered document, and these backup copies are stored away from the computer. Backup copies of computer-stored material are essential. Many law firms are very negligent in not maintaining backup copies of the material they have stored on magnetic cards, floppy disks, or hard disks even though it would cost thousands of dollars to rekey the material if it were accidentally destroyed or erased.Google Scholar

23 This is not an actual will suited for client use but only an example of how the ABF language is used. The will system designed by Seibel for use by the clinic staff is not reproduced in this article. Seibel plans to publish a separate article describing the will and trust systems.Google Scholar

24 Passages that are to be duplicated are bracketed by [REPEAT] …. [END REPEAT]. Such passages must contain a bracketed statement that indicates the condition that is to stop the duplication of the passage. See Sprowl, supra, note 5 at 25–30.Google Scholar

25 For a complete explanation of the way documents are described in the ABF language, see id. at 17–30.Google Scholar

26 Actually, the computer does not “convert” itself into anything. The ABF processor module simply turns control of the computer over to the word-processing module. The word-processing module then takes whatever action is necessary to bring the client-customized will into its “workspace” where it can be reviewed and revised. A separate printing module later controls the printing of the will. We found it essential to arrange the ABF processor to pass the completed will to the word processor automatically, because beginning trainees became confused if they were required to fetch the will into the word processing environment by issuing a FETCH command following the client interview.Google Scholar

27 We use the term client interview here and elsewhere to refer to an interview such as the one just described. The exchanges are actually between the attorney (or secretary) and the computer about the client and the service needed. The client may be present, or the questions may be answered. from information obtained earlier from the client and placed in his or her file. The interview process could be set up for client self-interviews if adequate safeguards were included to be sure that the questions were understood and answered completely and accurately. The automated practice systems described in this article were not designed to be used by unsophisticated clients who were not assisted by a lawyer or a paralegal.Google Scholar

28 See Sprowl, supra note 5, at 54–56 for a more detailed explanation of how replacement questions and directives are actually created.Google Scholar

29 See id. at 35–50 and 56–60 for a complete explanation of procedures.Google Scholar

30 For a complete explanation of the rationale underlying the design of the ABF processor, see id. at 4–17. Columbia University law professor Robert Hellawell, who learned computer programming to design a tax-planning computer system, analyzed the difficulties of using a computer specialist to design such a system:. A possible obstacle to using computers for legal analysis will be that writing programs on complex legal subjects is a difficult and time-consuming job. Moreover it is doubtful that a nonlawyer professional programmer could take on as much of the work in legal programming as he can in many other areas. My work on CORPTAX, for which I did both the legal analysis and the computer coding, convinced me that an expert on the legal problem being programmed will need to be closely involved throughout. Most of the coding is the coding of legal decision points. The plan or algorithm of the program is almost wholly one of legal decisions. A legal expert, or a junior lawyer under expert supervision, will at least have to write the plan, and that task will be very time consuming. Indeed, it will be the bulk of the work. In contrast, much standard nonlegal programming may require a comparatively smaller component of professional decision-making. Hellawell, Robert, A Computer Program for Legal Planning and Analysis: Taxation of Stock Redemptions, 80 Colum. L. Rev. 1363, 1393 (1980). For a comparative discussion of the ABF processor approach to system design and the approach taken by most computer professionals, see Sprowl, supra note 5, at 6–9, 12, and 15.Google Scholar

31 The Legal Services Center, located on the first floor of the IIT/Chicago Kent College of Law building at Wacker Drive and Monroe Street in downtown Chicago, includes 16 private offices and 5 large rooms. Three of the private offices are small rooms designed to hold a computer terminal and, at most, 2 or 3 chairs. These offices are used for computer programming and for delivery of legal services using the computer. The 5 larger rooms are a library, 3 student workrooms, and a work area for 3 secretaries plus a computer terminal.Google Scholar

32 “Demand for, and use of, lawyers' services by people of moderate means are both highly elastic. They are determined by certain elements of competition, including: the quality of the service; its price; its accessibility; and public knowledge and attitudes about law, lawyers, and lawyers' services.” Barlow F. Christensen, Lawyers for People of Moderate Means: Some Problems of Availability of Legal Services 38 (Chicago: American Bar Foundation, 1970). The center's service of moderate income clients for fees was a new and growing part of its program. These client groups had produced some cases that would not usually occur in a poverty law office: for example, real estate closings, income tax return preparation, estate planning, evictions representing the landlord, probates, and divorces involving property settlement arrangements.Google Scholar

33 For example, in 1979 the center opened approximately 20 new cases each month for clients seeking simple estate planning or a new will. During the experiment, the center normally received from 150 to 300 “new client matters” each month. Approximately one-half of these new client matters were resolved during the initial consultation; the remaining matters required additional legal services of one type or another.Google Scholar

34 The “sheltered environment” was created by the generous financial support this experiment had from the Council on Legal Education for Professional Responsibility (CLEPR) and by support from the law school, CLEPR authorized $100,000 to support the ABF processor experiment and the other experiments in automating the center's activities. The center considered the CLEPR grant to be seed money funding the establishment of a self-supporting or partially self-supporting clinical education program, built on a clinic serving moderate-income clients and charging modest fees. It was hoped that the computer technology would form a cornerstone of this moderate-income clinical practice and would enable the clinic to provide services of higher quality at reduced costs. If the computer technology did not assist the clinic in providing efficient, cost-effective legal services, then its use might have to be discontinued after the end of the experiment. But to date, the computer is still in operation.Google Scholar

35 During the experiment, there was a significant turnover of both professional and nonprofessional personnel at the center. So far as we can determine, the computer did not cause lawyers and legal secretaries to take jobs elsewhere. But this turnover did have an unexpected beneficial impact on the experiment: because new faces often appeared, we were provided with many opportunities to test different training strategies and different designs of the processor itself.Google Scholar

36 The center also accepted a variety of clients from other sources to provide a variety of lawyering experiences for the students. For example, Staudt was teaching a course on employment discrimination in the clinic, so he and his students frequently represented clients who claimed they were victims of job discrimination. Other cases requiring significant amounts of litigation were accepted to provide trial opportunities for student interested in litigation. In January 1979, the center began a criminal litigation program and added two criminal litigation experts to its faculty. The cases that originated from the sources just listed were not sufficiently repetitive to be suitable candidates for automated document processing within the confines of our small clinic. But we see the need to provide automated drafting systems for pleadings and motions for such cases in the future when it will be possible to market automated form systems to attorneys throughout a large county or a state or even throughout the country. For example, though the center did not have a sufficient volume of evidence-suppression motions in criminal cases to warrant automating the drafting of such motions, there is clearly a sufficient volume of such motions filed in Cook County to warrant the development of a county-wide motion-drafting system that could be shared by many defense attorneys. The litigation faculty experimented briefly with using sorting and searching features of our computer to help them sort through abstracts taken from depositions originating in a criminal action involving multiple defendants. Our flexible command module, described in sec. IVB, infra, enabled us to build up a set of special commands to facilitate this work.Google Scholar

37 See notes 41, 43, 44, and 45 infra for a description of the terminals we selected for this experiment.Google Scholar

38 The modern trend in text entry and revision systems is toward screen-oriented systems that permit one to display a document on a viewing screen, move a cursor (or bright spot) to the point where a correction is to be made, and then type in new material or erase unwanted material (erasure is accomplished by passing the cursor over the unwanted material). Screen-oriented systems work best when only a small number of revisions are required and when they are run by full-time operators; they are therefore generally relegated to word-processing centers. Those most skilled at operating screen-oriented systems are frequently not as highly skilled in typing, spelling, grammar, and style as are many legal secretaries, who often prefer to use standard electric typewriters and who, by making fewer errors, can often outperform less skilled operators of the screen-oriented systems. The complexity of the screen-oriented systems is such that attorneys almost never attempt to operate them. The Northwestern word-processing module was not screen oriented. It displayed the text of a document with a number assigned to each line. To delete material, one typed the letter D followed by the line number or range of line numbers to be deleted. To insert material, one typed T followed by the line number of the insertion point, and then typed in the insertion. In addition, the module could search single lines, sets of adjacent lines, or an entire document for words or phrases and could replace these words or phrases with others. To use the Northwestern module, one needed to master only four commands—insert, delete, search and replace, and display. The simplicity of this module made it possible for everyone in the clinic to understand its use. Attorneys could produce and revise wills in the evening without secretarial assistance.Google Scholar

39 See note 53 infra.Google Scholar

40 On one occasion, it insisted upon inserting a randomly selected character into the first page of a will. We never were able to trace down the source of this particular problem, but when we made a trivial change in the will, the problem disappeared and never returned.Google Scholar

41 In addition to automated law practice systems, we also developed management and accounting systems during the course of the experiment. When we purchased the terminals to begin our work on the ABF systems, we still did not know what kind of technology we would use to perform the accounting and management functions. Therefore, we purchased general purpose terminals that could be used in a variety of ways. As we explain in an article to be published, our accounting and management systems were eventually built within the Control Data computer that also contained the ABF processor system. Access to both systems was obtained through the same terminals, using the same front-end command module (described in sec. IVB). The management and accounting systems utilized several special features of our terminals and printers such as the special-function pushbuttons, the automatic page advance, and the automatic fill-in-the-blank questionnaire devices. None of these were used by the ABF processor module.Google Scholar

42 Exactly and technically expressed, the transmission speed was 300 baud.Google Scholar

43 This unit was an A. B. Dick Magna I automatic “daisywheel” typewriter that is compatible with IBM Mag Card II equipment. We equipped the Magna I with an optional communications package that enabled it to communicate with the Northwestern computer. We then leased a special telephone that included a modem or computer-telephone interface. To avoid switchboard interference problems, we installed a dedicated telephone line for this terminal.Google Scholar

44 This unit was an Anderson Jacobson automatic “daisywheel” terminal. We connected it to a ComData 300-baud modem that contained a cradle for a standard telephone handset. This terminal was equipped with a forms tractor that enabled it to print out lengthy administrative reports on accordion-fold paper with perforated tear-away edges. At an earlier point in the experiment, we had leased a Teletype 41 printing terminal from the telephone company. The Teletype 41 coupled directly into any phone line without the need for a separate modem. It was quite compact and had readable type, but its dot-formed characters were not suitable for court filing.Google Scholar

45 These were microcomputer-controlled Teleray terminals, model 1061. One was connected to a ComData modem having a cradle for a standard telephone handset. This terminal operated at the slow 30-characters-per-second rate (300 baud). The other was connected to a special data telephone by a Vadic modem which operated at a 120-characters-per-second rate (1,200 baud). The faster data rate meant that this terminal was considerably more vulnerable to telephone static. The static problem was particularly severe when we had to dial from Chicago to Evanston. After Northwestern installed a Chicago entry port to its computer, we were able to contact the computer by a local call placed within Chicago, and then the static problem was much less severe. But occasional garbled lines were always a problem with this terminal, possibly because of some slight mismatch between our equipment and Northwestern's.Google Scholar

46 Assembling a will for a client, a task that normally took 5 or 10 minutes, could take as long as 30 or 40 minutes in the late afternoon or at the end of a school term because of the much heavier student use of the computer at those times. This was largely due to the inefficient design of the prototype ABF processor, which needed 60,000 computer memory locations to function—a very large amount of memory compared to the less than 10,000 locations that Northwestern's word processor module required. At times when many small jobs were being run, the automated job scheduler on Northwestern's computer penalized the ABF processor because it required such a large block of memory.Google Scholar

47 The Northwestern computer functioned flawlessly most of the time. But some weeks it would “crash” (cease operating entirely) several times during each day until some major technical problem was diagnosed and corrected. If such a crash occurred when one had just about finished a will interview, and if the “job” was not “saved” and resumed, then one had to start all over again.Google Scholar

48 The experiment was designed to last only a short time, and our resources were not sufficient to warrant the purchase of in-house computing equipment for the experiment. Nor do we wish to criticize the staff of the Vogelback Computing Center at Northwestern University. At the outset of this experiment we were aware of the limitations just described, and the fact that the experiment was quite successful in spite of these limitations is, in large measure, due to the support we received from the Vogelback staff.Google Scholar

49 We permitted EQUALS and = to be used interchangeably; we permitted THEN to be omitted; and we made numerous similar changes. There were also some language constructs we had permitted in the original prototype that we decided were too dangerous to be used by the trainees. For example, we learned that it was essential to have the computer check for spelling errors in the capitalized ABF language words such as IF and END IF. We learned that the trainees often misspelled these words, and the early version of the processor simply assumed the misspelled words were part of the statements within the brackets that described client data. No error indication was given, but error checking was clearly needed here to assist the trainees. To correct this problem, we forbade the use of capitalized words within brackets except for a limited number of special words such as IF and END IF. The computer was then instructed to assume that any other capitalized word encountered within brackets was misspelled, and it flagged all such words as errors. But we also determined that the computer could safely ignore any spelling errors that occurred beyond the first four letters in special words such as OTHER WISE. The computer could correct such spelling errors without drawing them to the trainee's attention. In many other small ways, we tightened up the error checking where it helped the paralegals and we loosened it up where it simply slowed them down. Feedback from users, essential to making such fine adjustments, was supplied by a group of five law student volunteers who constructed an intestate probate system and a short form 1040 income tax return preparation system during the fall of 1978. Northwestern law student Gregg Maryniak, now a Chicago attorney, was particularly helpful.Google Scholar

50 A short description of the concepts introduced at the first five classes is as follows: Class 1: Use of the computer terminal; how to contact a remote computer by telephone; creation, modification and destruction of documents; introductory word-processing commands; introductory BASIC, including programming a computer to solve a simple compound interest problem; and introduction to repetitive operations in BASIC. Class 2: More discussion of repetitive operations and terminals; additional text revision commands; introduction to temporary document storage; library storage of documents, and the commands needed to gain access to libraries and to manipulate documents stored in libraries; and introduction to the postprocessor and the commands that control centering, paragraphing, and the like. Class 3: Introduction to the ABF language; the types of documents a library may contain; generating repetitive passages using the ABF language; explanation of the differences in notation between BASIC and ABF language; repetition-controlling commands; each class member assigned to write a portion of a divorce system, beginning with the complaint or petition. Class 4: Additional instruction on ABF processor protocols; the formatting of numbers; altering questions; review of the text-editing commands, system commands, and library commands; further work on the postprocessor; and a quiz reviewing all commands and procedures. Class 5: Discussion and examples of the difference between short names and file names: one being the temporary name that the computer gives a document or record in temporary storage, the other being the name the ABF processor gives a procedure or question; problems were raised by class members, particularly about the use of arrays and variable lists.Google Scholar

51 The law students had mixed reactions to the project. One student who had previously worked with computers thought it was interesting and fun. She maintained her interest in the field test and later helped design the trust system under Robert Seibel's supervision. The second, who had not previously worked with computers, was neutral. The third, who had been a programmer and had gone to law school to get away from programming computers, reacted quite negatively to the project toward the end but ultimately completed his part. All the trainees found the commands frustrating, as we shall explain.Google Scholar

52 Stahulak was the only secretary to attend all the class sessions with the attorneys and law students. The others were trained separately to do document entry and revision work, and they were not taught the ABF commands. They were supposed to help the attorneys enter and revise the stored documents.Google Scholar

53 A further explanation is needed of why those trained in the fall of 1978 had more difficulty with the commands than did those trained the preceding spring. The Northwestern word-processing module included an excellent mechanism for storing documents in separate “libraries.” Accordingly, we designed the ABF processor to fetch documents directly from these word processor libraries. Form documents stored in any such word processor library could thus be processed by the ABF processor directly with no complications. We used this library mechanism in the spring of 1978, and the trainees had no difficulty with it. Unfortunately, as the number of documents stored in the word processor libraries grew, our use of this library mechanism became very costly. The storage charge for a document stored in one of the word processor libraries was 25 cents per day, no matter how small the document was. Since our law practice system libraries often contained from 50 to 120 documents and since many of them were only one to three lines long, we soon found that our monthly charge for storing documents on the computer exceeded $500. Document storage charges were rapidly using up our rather limited project funds. The word processor's library mechanism also caused each document, if it was not used for several weeks, to be pulled off the computer and archived individually on a reel of magnetic tape. Thus, when an automated legal practice system had not been used for several weeks, its documents and procedures would all be archived, and it might take the Northwestern computer an entire morning to retrieve all the individual documents and procedures of the system from the many separate reels of magnetic tape, each of which had to be hand mounted on the computer. We were thus forced to substitute a new document storage module of our own design for the word processor library mechanism. The new module stored all the documents relating to a single legal practice system together as if they were a single document. Hence, the 25 cents per day charge applied to an entire set of documents, rather than to each individual document. Then if a set of documents was not used for several weeks, all the documents in the set were archived together on a single reel of tape from which the entire set could later be retrieved in only 10 to 15 minutes. But the commands needed to operate the new document storage module were considerably more complex than were the commands associated with the word processor library mechanism. Those trained in the fall of 1978 were thus required to master a more complicated set of commands than those the trainees had used the preceding spring.Google Scholar

54 Those with computer experience may be interested to know just how this command module functions: Norstadt was charged with maintaining Northwestern's word processor module at the time of this experiment. To increase its flexibility, he integrated that module with a separate interactive text revision module designed by Mausner to enable those using the word processor module to initiate the execution of tasks performed by other modules within the computer. Using this mechanism together with a separate mechanism within the word processor module that is normally used to initiate the translation of English-like computer programs into numeric form, Sprowl caused the text-editing command GO to prompt the operator by displaying COMMAND? and to accept whatever single-word command the operator supplied in response. The computer then checked to see if the single-word command corresponded to the name of a document in a special library of command documents. If a document was found whose name was the same as the command supplied, this document was delivered to Mausner's interactive text revision module. Questions were generated, and the answers supplied controlled the revision of the document. After it was revised, the document took over control of the computer and caused it to carry out a series of tasks. All one had to do to create a new command was to prepare a document and store it in the special library of command documents, assigning to the document a name that then became the name of the new command. Each document in the command library contained a mixture of text-editing commands, commands to the computer's operating system calling for control of the computer to be turned over to various modules, and special text revision commands that caused Mausner's program to ask questions and then use the answers supplied to control the revision of the document. The command module was not particularly efficient, but it made an excellent breadboard structure for testing different arrangements of the commands as the experiment progressed.Google Scholar

55 One legal secretary, Teresa Stahulak, ultimately used many sophisticated commands commonly used only by skilled programmers. She now opens and closes computer accounts, arranges for making tape backup copies of files, and generally functions as the computer expert for the center even though she has never received any formal training in computer science. The attorneys and legal secretaries who participated actively in the design of automated delivery systems reached intermediate levels of proficiency that reflected their enthusiasm, their talent, and the amount of time they spent working with the computer. The remaining staff members, who used the computer only to produce client documents, remained securely within the innermost protective sphere where they seemed to function very well. Our trainees thus sorted themselves out in a voluntary manner into a variety of proficiency categories, and the computer's command module was flexible and powerful enough to meet the needs of users within each category.Google Scholar

56 During the first lecture, Sprowl explained how to use the terminals; introduced the new commands and the new computer-aided instruction facility; and discussed word processing in a very summary fashion, covering only the most basic word-processing commands. The trainees were then immediately sent to their terminals to begin their first laboratory session. One week later, Sprowl gave a slightly more elaborate explanation of word processing and of the commands for retrieving, revising, storing, and printing documents, and he introduced the class of the ABF processor and handed out copies of the ABF primer. Then once again, the trainees went directly to terminals to work on their assigned automated legal practice systems.Google Scholar

57 From 1972 through the end of 1978 Seibel was employed as an associate at Ropes & Gray, a large Boston law firm. His practice there was concentrated in estate planning for moderate and large estates. Mark Neil, a Chicago attorney with estate-planning experience, assisted Seibel in the early weeks to draft language appropriate for Illinois wills.Google Scholar

58 The legal secretary assigned to work with Seibel on the will system was also a new addition and had not participated in the fall computer classes. Therefore, both members of the team assigned to set up the will assembly system started in January 1979 with no computer experience and no ABF processor experience. They worked as a team for several months until the secretary left the clinic to be married, after which time Seibel did his own work with the computer, assisted by a number of other legal secretaries.Google Scholar

59 In July 1981, Seibel moved to the clinical law faculty of the University of Maine.Google Scholar

60 For a variety of reasons, low-income individuals in Cook County, Illinois, sought to obtain the transfer of legal custody of children for short periods of time. For example, such a transfer is necessary when a sole surviving parent wishes to enter military service and cannot take the child to the military base. The probate court in Cook County was unable to find lawyers to represent individuals who wished to have such uncontested procedures completed. Therefore, in conjunction with the local legal aid program, the Legal Services Center agreed to represent individuals seeking uncontested guardianships on referral from the probate court. During 1979 our automated guardianship system was used for 70 clients.Google Scholar

61 This system was extremely simple and did not require sophisticated or complex clause selection or document drafting. The text of the documents was short and concise so that automated printing did not offer great advantages over manual typing. Only four documents were produced, none of which exceeded one page. In fact, as described in the text, the only difficult aspect of this system was designing instructions to the computer's document-printing module that prepared the document caption for multiple parties. The difficulty was not attributable to the ABF system but rather to the printing module, which had no provision for preparing such a caption. Sprowl used the RABF system to compensate for the deficiencies in the printing module.Google Scholar

62 Seibel's time sheets indicate that from January through August 1979, he spent approximately 135 hours in writing, rewriting, testing, and debugging ABF computer systems. During February, March, and April much of this time was spent working on the will system. Some of his time in the 8 months was spent on the guardianship and trust systems. In May 1980, Seibel stated that he probably spent more time than was indicated on his time sheets working with the computer, but he agreed that 200 hours is a reasonable outer limit on the billable time he spent working on the will system.Google Scholar

63 With an in-house computer, the costs could have been either larger or smaller, depending in part on how many other tasks the shared computer could be used to accomplish. The cost of computing power is dropping rapidly, and within three to five years it is likely that a computer as powerful (for our purposes) as Northwestern's can be purchased for $15,000 to $30,000 (in 1981 dollars).Google Scholar

64 This phenomenon is illustrated by comparing Hughes's and Seibel's experiences in developing their systems. Seibel was supplied with a model system, and he successfully used that model as the departure point for his own successful design efforts. Hughes was not initially supplied with a model system, and her first attempt to design the divorce system was unsuccessful. Her second attempt succeeded partly because she had available the unsuccessful first system as a departure point. A number of student projects not described here also demonstrated the importance of a model system.Google Scholar

65 The actual “protocol” or convention for transmitting data over telephone lines has been standardized for some time. West has adopted additional protocols defining commands that cause the remote terminal to perform different tasks such as writing on the terminal screen at selected locations or erasing at selected locations. West has adopted the command protocol of the Hazeltine Model 1520 terminal as its standard protocol. Perhaps a better protocol would be that recently promulgated by the American National Standards Institute.Google Scholar

66 This is the PCUP project (Professional Computer Utilization Project) funded by the Murdock Charitable Trust. Sprowl spent the summer of 1979 in Portland training the PCUP team in the ABF technology. The PCUP software is written in the language PASCAL and is presently installed on a PDP 11/780 (VAX) minicomputer manufactured by Digital Equipment Corporation. The PCUP team hopes to install this system on a much smaller microcomputer in the near future.Google Scholar

67 This excellent system is intended to demonstrate the “human engineering” principles that must be incorporated into the law office computer systems of the future. Farmer is a psychologist as well as a law professor.Google Scholar

68 Saxon, a professor at Eastern Michigan University, is an attorney as well as a computer scientist. He rewrote the “machine language” portions of the Northwestern ABF prototype in FORTRAN and revised the remaining FORTRAN programs to be compatible with IBM computers. The Ann Arbor system is an all-FORTRAN system that runs under the MTS (Michigan Terminal System) operating system on an Amdahl computer. It is almost finished as of this writing.Google Scholar

69 See, e.g., Boyd, William E. & Saxon, Charles S., The A-9: A Program for Drafting Security Agreements Under Article 9 of the Uniform Commercial Code, 1981 A.B.F. Res. J. 637, where the authors describe how they used a commercial law office microcomputer marketed by LMS, Inc., Ann Arbor, Michigan, to automate the assembly of complex security agreements. See also Hellawell, supra note 30, where the author describes his computer program that helps analyze the tax consequences of corporate stock redemptions, taking into account the constructive ownership rules of Internal Revenue Code § 318. The program is written in BASIC-PLUS-2 to run on a Digital Equipment Corporation DECSYSTEM-20 computer.Google Scholar