13 May 1991
6,700 words

Streamlining Discovery with Computers

By Norman Bauman

It was a lot of work to put 115,416 pages into their data base, but it was worth it. After a day of searching, Hennigan & Mercer found the documents that won a $35 million settlement for their client in Abzug vs. Kerkorian et al.

"I don't remember who found it first, but I searched the key word 'PROJECTION' and it popped out," said Matthew Ghourdjian, litigation consultant on the staff of the 12-lawyer firm in Los Angeles.

The critical document was a valuation prepared by Bear Stearns to value the MGM film library. "They valued it at $470 million, but the handwriting on the edge of the document brought the value up to over $700 or 800 million," said Mr. Ghourdjian.

Abzug vs. Kerkorian was a class action suit that arose from MGM/UA's sale of all its assets to Turner Broadcasting Co., and Turner's resale of a subsidiary back to Kirk Kerkorian. The shareholders claimed that Mr. Kerkorian undervalued United Artists when he sold the parent company to Turner for $1.5 billion, and that Mr. Kerkorian subsequently bought back United Artists and the MGM/UA name in a sweetheart deal for about $470 million.

"Everybody had this document," said Mr. Ghourdjian. "It was obviously passed around at a meeting, because we got a copy of it from Ted Turner, from Drexel, from Bear Sterns, from everybody. And they all had different notes on them." With their Macintosh-based optical imaging system, they were able to find all the versions and coallate the handwritten notes.

Hennigan & Mercer probably would have found the document even without their computer, since it was delivered early in discovery. But their in-house MARS imaging system, with its 327-gigabyte Sony optical drive, made it easier and cheaper to manage a huge number of documents, and put the attorneys more in control of the document-management process. Hennigan & Mercer had been paying their service bureau $8 a page to enter and code documents, but they hope to get the incremental cost for the in-house system down to 60 cents a page.

Mercedes or Schwinn?

Hennigan & Mercer is a relatively small firm with an unusual amount of in-house computer power. Taking the small-scale approach to small-firm automation are individual attorneys who prepare their own PC litigation data bases on Symantec's Q&A (a surprisingly popular program). Paradox is a powerful data base, legal hackers say, but it takes too much time, and has no pre-packaged legal applications. When they reach the limits of PCs, the larger firms may go to Unix, and the smaller firms go to service bureaus.

What's the difference between MS-DOS and Unix? Suppose you were going to depose a witness, and wanted a report of every place that witnesses' name appears in half a million documents, explained Spencer Read, litigation support analyst at Orrick, Herrington & Sutcliffe, San Francisco, California. With a 256-megabyte Unix system, Mr. Read could have that report on your desk in four hours. With a PC, you couldn't get it till the next day, he said.

When discovery reaches 25,000 pages, it requires full-time personnel for coding and document management, and becomes difficult for a mid-sized firm to handle in-house. Service bureaus can enter full text with as much accuracy as you're willing to pay for, and they can code those documents with hard, objective information such as date, names, and type of document. For the subjective data fields, such as abstracting, some attorneys prefer to train their own paralegals in-house.

How the data base is used

Either the full document set or selected documents are scanned into a data base that always contains the date, names, and "Re:" line of the document; probably contains key words and an abstract; possibly contains a subject code; frequently contains the full text; and sometimes contains the bit-mapped image as well.

These data bases can then be used for two things: first, subtle searches for key words, and second, brute-force document management.

In some litigation, critical issues can be defined particularly well in terms of key words. A will is contested, and you want to know exactly what the deceased intended the words of the will to mean. A patent is contested, and you want to know exactly what the words of the claim meant as they were used by technical people in the field. In commercial litigation, documents can be reduced to a "blob" of searchable text that attorneys can probe for useful information about the terms of the agreement and facts of the case.

Then there is a brute-force application to document data bases as well. Sometimes the game is to bury the other side in paper--hundreds of thousands of pages. One of the skills in modern litigation is knowing how to manage a mass-production paper factory, in which documents must be scanned for relevance, cataloged, and kept available in case they turn out to be useful. The object of the game is to first, manage the documents cheaply, and second, locate the documents that are helpful to your case.

Winning by intimidation--or losing the underdog advantage?

Most litigators like to flaunt their strengths, including their computer power, like two stags roaring at each other before they lock horns, each hoping that the other will settle on favorable terms without actual bloodshed. "An opponent fears somebody who is well prepared," said Peter Brown, of Brown, Raysman & Millstein, New York City, "whether that preparation is by paper records, or computer documents."

Neil E. Aresty, of Aresty & Associates, Boston, Massachussets, recalled a witness being deposed while he was searching prior depositions and testimony for inconsistencies on his laptop. "They were intimidated," he said. But, he added, "Sometimes in trial you may not want to come across as a city slicker. Having a computer could be a distraction and may have the wrong impact on the jury." An attorney who opposed Hennigan & Mercer wondered whether they lost their "underdog advantage" by bringing their fancy equipment into court.

Tipping your hand to discovery?

Some attorneys preferred not to reveal their use of the computer, even though it should be obvious. One attorney refused to be identified for fear that a judge might decide that all or part of his data base is not be covered by the attorney's work product protection. This is unlikely, said attorneys in the field--but it is an outside possibility. "I don't know of any cases in which this was a problem," said Mr. Brown.

Back in the old days, said old-timer Paul S. Hoffman, the usual strategy was to keep quiet about your use of computers, for fear of creating discovery problems, rather than to flaunt your computer power, for purposes of pushing a settlement. But nowadays, you can be fairly sure that everyone in major litigation is using a computer, said Mr. Hoffman, a sole practitioner in Croton-on-the-Hudson, New York.

An argument for discovery

One attorney's problem is his adversary's opportunity. If your opponent is abusing the discovery process by burying you in paper, could you convince a judge to save you needless duplicative expense, by giving you access to your opponent's data base directly?

"There is discretion of the court," speculated Mr. Aresty, "to say that these documents are available in hard copy, but only available in an inconvenient and expensive form, and the electronic form is cheaper and much more efficient, and the interests of justice would dictate producing the documents through electronic form."

"If I were seeking that information in discovery, I would argue that there is no way we could get this information in an efficient manner, and we merely want a printout of the name of the document, the document type, the abstract, and the location of the document," said Mr. Aresty.

"Whether a court would buy that, I don't know," said Mr. Aresty. If he were seeking to protect that information, he would argue that the attorney was involved in the design of the data base from the beginning, and even the objective classifications, done by the outside service bureau at the attorney's direction, are protected.

The very fact that an attorney has decided to put a document into a data base is useful and time-saving information. One big job in large-document discovery is sorting out the small number of useful documents from the huge mass of junk. So the Bates number field alone could be useful.

If the court did buy it, said Mr. Hoffman, a specialist in computer contracts, the data base would probably be divided into a priviledged part and a disclosable part. If one party went to a great expense in preparing the data base, the court could reasonably order the other party to reimburse part of that expense, he said.

In fact, Hennigan & Mercer discussed sharing the cost of computer document preparation with one adversary in a case that was running into 600,000 documents, said Mr. Ghourdjian. But the deal fell through. "They didn't want to do anything to help us, even if it would save them money," he said.

Three case histories

Below are descriptions of how three firms use computerized discovery. Hennigan & Mercer uses a state-of-the-art document management system for massive commercial litigation like Abzug vs. Kerkorian. Orrick, Herrington is keeping its options open with both PCs and Wang/Unix, and Mr. Read described how they defended a will in a case which involved computerized depositions, hearing transcripts, and daily trial transcripts. An anonymous midwestern patent attorney, who programmed computers himself in his previous career as a mechanical engineer, explained how he sets up simple data bases in Q&A, and sends the big jobs out to a service bureau.

The document strategy

Hennigan & Mercer does "sophisticated commercial litigation," said Mr. Ghourdjian. "In this litigation, we have the advantage of having a paper trail."

"I don't care how big the case is, or how small the case is, it comes down to a handful of documents," said Mr. Ghourdjian. "If it's any bigger than that you're going to have a hard time. This system lets you manage that incredible amount of information, and cull out all that garbage, and get down to what you need, and it puts that power into the hands of the attorney."

"That MGM case was like most valuation cases, a case where business records document a transaction," said Mr. Ghourdjian. "You can rely less on what people say and more on what documents say. That speaks to our state of mind much better than our individual recollections three years later."

"At the beginning Kerkorian was 50.1% owner of MGM/UA," said Mr. Ghourdjian. "At the end of this series of transactions he had 87% of the new UA. So he basically squeezed the shareholders out of 30% of the value."

"It came down to the value of the MGM film library and operating company, and the UA film library and operating company," said Mr. Ghourdjian. "Mysteriously enough, MGM had no records documenting the valuation," he said. "Except for a few scarce documents that we got from Bear Sterns, Drexel Burnham, and Turner Broadcasting System, there was essentially no valuation documentation at all. What we were looking for was tidbits of information from which we could derive a value."

"We did that by being able to peruse this enormous data base," said Mr. Ghourdjian. "You could run the word 'VALUATION,' you could run the word 'PRO FORMA,' you could run the word 'FILM LIBRARY.' The way the software allows you to browse, it's incredibly efficient."

The computer litigation strategy

"First of all we were able to take all these documents, scan them, turn them into digital pictures, and also turn them into text," said Mr. Ghourdjian. "So now on the computer in searchable form is this blob of free texts that you can do boolean searches on.

"So Michael Hennigan and Jim Mercer could do on their own desk everything that a paralegal could do," said Mr. Ghourdjian.

"The first thing they did was make sure there were no documents in the file that supported a good business judgment valuation for MGM/UA lower than $1 billion," said Mr. Ghourdjian. "And we didn't find any. So we knew they couldn't suprise us with anything. Our firm does plaintiff and defense work, and we play it straight and give the other side everything they deserve."

The proverbial haystack needle

"A secondary search was culling out all the documents that had to do with valuation," said Mr. Ghourdjian. "One document popped out of this search. It was a 1-page valuation pro forma, done by Bear Stearns, who was hired to value the company. It had all kinds of math errors that lowered the value of the company. It also had notes in the handwriting of the financial vice-president of MGM/UA that took this low value and added all kinds of additional items to it that brought the value back up."

(An attorney on the other side had a different interpretation. "There was no math error by Bear Sterns," he said. "There was a column of numbers in which one number wasn't included in the total, but that was done intentionally because it was an item in a particular year that was not going to be received. Plaintiffs never found that out because the case settled before the witness got to testify at trial.")

Documenting the alleged conspiracy

"Our theory," continued Mr. Ghourdjian, "was that Bear Stearns had undervalued the company." In this interpretation, MGM/UA said, "Bear Stearns has undervalued the company, let's take advantage of it, and put this deal together."

"One of the crucial pieces of evidence was that none of MGM's management had originally had any substantial holdings in old MGM/UA, but in the new deal, the big key people bought to beat the band," said Mr. Ghourdjian. "The former president of MGM/UA bought $3 million worth of stock, the chief financial officer bought $1 million worth of stock. These were people who only owned a couple of thousand shares before. Obviously the people in the know knew this was a good deal."

(The opposing attorney had a different story about this too. "The public shareholder subscription was in October 1985, and at that time, many people in MGM's management elected not to buy the stock," he said. "The people who bought the stock he's referring to bought it in February 1986, when the new UA company was starting up, and there's nothing improper about management in a new venture getting stock options or stock.")

How much is the Home Entertainment Group worth?

"One of the methods we used to determine the value of MGM/UA was to value a subsidiary of MGM/UA that had been repurchased," said Mr. Ghourdjian. "That was the MGM Home Entertainment Group, which owned the video rights to all the films in their library. MGM/UA had bought back a 15% interest from another company, so we could extrapolate the value from the repurchase price. In a matter of moments we could pull out every document in which the Home Entertainment Group was mentioned, and look for everything that had to do with its valuation. Jim Mercer could do that at his desk, as he got the idea. We wound up thinking that the subsidiary alone was worth probably $1 billion."

Attorneys replace paralegals

"The research capability is amazing," said Mr. Ghourdjian. "One of the attorneys wanted to know whether Turner Broadcasting had been indemnified. That's a pretty simple search. Just run the word 'INDEMNIFY.' A few seconds later you find out that it was not."

The information you need to build a complex case like this can only be recognized and found by attorneys, not by paralegals, said Mr. Ghourdjian. "When you're talking about a physical tort, that's something that a jury can understand," said Mr. Ghourdjian. "It's a physical act. But when you start to talk about an ethereal thing, like conspiracy to breach a fiduciary duty, you're starting to talk about something that is beyond the reach of paralegals, and concepts that only exist in the minds of attorneys."

"So Jim Mercer is sitting in his own office, searching the documents, and in a couple of days he was able to come up with a thesis," said Mr. Ghourdjian.

"We've taken a lot of the research out of the hands of paralegals and put it into the hands of the attorneys," said Mr. Ghourdjian. "We have our senior attorneys tied into the office through 24-hour-a-day digital lines, and these digital lines allow them to browse their files at home. A guy gets a hot spot, he can pursue it right then and there."

5 minutes to search 100,000 pages

"The day we got our exhibit list ready for trial, it took us an hour to get all our documents together, put them into a word processing document, and send them off to the other side," said Mr. Ghourdjian, "because they'd already been flagged as trial exhibits."

Jim Mercer came in and said, "I want to take the deposition of the chief executive officer of UA tomorrow, and I want every document that he's ever written, I want every document that he's ever mentioned in, and every document that he's ever received," recalled Mr. Ghourdjian. "It took 5 minutes to search 100,000 pages, and come up with 50-75 documents, and 30 minutes to print them out."

Fielding ARCO's production

"We're suing Atlantic Richfield Corp. (ARCO), in a case rising from the wage and price control freeze of 1974, in which we represent a class of wholesalers," said Mr. Ghourdjian. "They thought they'd bury us with paper." ARCO informed Hennigan & Mercer that they would be giving them 600,000 pages. "They gave us 250,000," said Mr. Ghourdjian. "We sucked it right into the Mars system. No big deal. ARCO saw that and decided to buy a Mars system."

"We were originally going to give them the documents on disk without any proprietary coding," said Mr. Ghourdjian. "They suggested sharing documents." The idea was for each party to put their documents onto their own MARS system, and share the costs. "We didn't have any objection to it, but they decided against it," said Mr. Ghourdjian. "They didn't want to do anything to help us, even if it would save them money."

MARS Mac-based optical imaging with full text search

Hennigan & Mercer uses the Macintosh-based Multi-user Archival and Retrieval System (MARS), which is marketed by Micro Dynamics, Ltd., Silver Spring, Maryland, with proprietary software and off-the-shelf hardware. They have three Fujitsu high-speed scanners and one HP Scanjet Plus, two Calera OCRs, a 327-gigabyte Sony jukebox optical disk drive holding 50 6.5-gigabyte disks, five Macintosh IIfx servers with 8 MB RAM, and hard drives of 40 MB to 1.3 gigabytes. The search text is kept on a 1.3 gigabyte CDC Wren-Runner II hard drive, and the images are kept on optical disks. Each 12-inch Sony optical disk can hold up to 200,000 images of the original documents, along with graphics and handwritten notes like the ones that were so valuable in Abzug vs. Kerkorian.

The firm has 70 Macintoshes, and most attorneys have Mac IIci's with two-page displays on 19-inch, 16-level gray scale monitors. "We're phasing out the color monitors," because most attorneys prefer the larger screens, to the color screens, said Mr. Ghourdjian. They did buy two Radius 2-page 19-inch color displays, to get large screen and color, but at $4,000 the are too expensive. Every attorney has a Mac at home, linked either by modem, or by 24-hour ADN 56KB digital lines to view images, which cost about $280 a month for the 300 miles to San Francisco. They run 10/Base T Compliant Farallon Start controllers, and are planning to go to Cabletron 10/Base T Concentrators when they move.

The initial cost of the MARS system was about $400,000, and it costs $200,000 to $250,000 a year to operate, including personnel costs. Mr. Ghourdjian estimates that, exclusive of amortization of equipment, the incremental cost of adding a page to the system should be about 60 to 70 cents. That compares to $8 to $11 a page they used to spend on a service bureau.

The old system

Partner J. Michael Hennigan had been an IBM salesman and programmer before going to law school, so the firm had been highly automated from the beginning, starting with IBM mag cards, and expanded to a network of PC compatibles, NBI word processing terminals, and Toshiba portables, some of which were "gathering dust" in the offices of attorneys who didn't have time to figure them out.

The firm had been using a San Francisco service bureau, which scanned the document, entered summaries into a customized database, and classified the document by key word. That system had several problems, said Mr. Ghourdjian.

First, the litigators were required to choose their key words so early in the case that they often chose the wrong words.

Second, the searches themselves were unwieldy. The attorney would start by calling San Francisco to specify the information he wanted, said Mr. Ghourdjian. Then a consultant would call back asking for a clarification. The report would arrive a week later. Then the attorney would revise the search field, or the key word. "It was an iterative process," he said. They had proprietary terminals in-house linked to the service bureau, but most attorneys weren't comfortable with them. Most complex searches had to be done with an attorney together with an on-site consultant who charged $150 an hour. The billings for the system in its last year were about $1.5 million.

After Mr. Ghourdjian arrived in 1988, he brought in Macintoshes. The main benefit of the Mac was that "we felt the attorneys were going to use the Macs," unlike the PCs. "Jim Mercer, who never turned on his Toshiba except to do a Nexis search, is absolutely a new person," said Mr. Ghourdjian with admiration. "He spends his afternoons at Egghead Software."

Mr. Hennigan, a PC-using holdout, ignored the Mac II in his office, until Mr. Ghourdjian brought him a Mac Excel spreadsheet. Mr. Hennigan wound up using it until 2:00 a.m., and woke up a convert to the Macintosh. "He discovered that using 1-2-3 on the Toshiba was taking an additional 2 or 3 steps for each function," noted Mr. Ghourdjian. The firm liquidated the PC system, and moved to Macs. They continued to use their Xenix-based Altos system for data base needs.

"We looked at a lot of PC- and work station-based systems," said Mr. Ghourdjian. "We didn't go out looking for a Mac system, but the MARS was more of a refined product, with more features, and the speed of it was incredible," especially with the FreeForm text searching software.

They still use deposition digests, which enable an attorney to quickly assemble material related to a specific case, said Mr. Ghourdjian. Each document is coded with standard search fields, but most searches are full-text.

Mr. Ghourdjian advises firms not to scrimp on the hardware, or on hiring a good manager for the LAN. Many firms try to get away with Mac SEs, but those machines are only good enough for document summaries, not for document imaging. The system requires a sophisticated LAN, which requires sophisticated management.

The adversary's assessment

The opposing attorney, who declined to be identified, gave his perspective on what it feels like to come up against the Hennigan & Mercer computer. "They're very good lawyers, and I'm sure they would have found the document with or without the computer system," he said. "There were five different copies of this document, and some of them came early in the litigation when they only had a few documents."

What impressed the opposing counsel most was "the ability to turn out very impressive graphics from their own computer, whereas we used an outside graphics service."

He did worry about an ostentatious display of equipment. "As counsel for a defendant being cast as a giant company that's throwing its weight around, the last thing you want to do is show up with all kinds of computer equipment, because it makes the other guy look like more of an underdog," he said. "When you're a plaintiff against a big company, you also have to be careful when you come in with tens of thousands of dollars worth of computer equipment, because then you lose sympathy in front of the jury."

PCs and Unix in a bond firm

Orrick, Herrington & Sutcliffe, San Francisco, California, is one of the premier bond firms in the country, said Spencer Read, litigation support analyst. With 280 attorneys, Orrick, Herrington also has a large public finance department, and does all types of general litigation, including securities, product liability, and intellectual property.

They have also had high-powered computer capability for some time. The firm uses a both 386 PCs and Wang/Unix for document control, which includes full-text document retrieval for deposition, hearing and trial transcripts. They also have a small desktop graphics department, and they find useful the graphics functions on data base and spreadsheet programs, notably Quattro Pro.

Defending a will

Mr. Read recalled a challenge to a will that Orrick, Herrington defended. "We had 30 or 40 depositions in the case, plus daily trial transcripts, plus hearing transcripts, all of which were put on line and searchable," he said. "We got the disk from the trial court reporter in standard ASCII on a daily basis. We could load it on our PC and have access to the prior day's testimony in a matter of seconds. This helps prepare for the following day's examination."

"We'd reviewed nearly half a million documents, and put 150,000 pages of them on line in coded format," said Mr. Read. "An outside company coded the objective information, such as date, document type, and names, and we hired people internally to do the more subjective coding, which included an elaborate issue outline, a description, and the significance, or value, or assessment of the document." This enabled the firm to print them out sorted by priority, "and look at the hottest ones first," he said. They could then compare the documents by issue-coding.

"So we could search for a name, within a timeframe, we could prioritize it, we could set it up any way we wanted it," said Mr. Read. "We could pinpoint our preparation with stunning accuracy."

Blew their socks off

"Quite frankly, we blew the socks off our opponents, who were not computerized," said Mr. Read.

"I used to get the floppy disk between 8:30 and 10am the day after the testimony, and I would search on certain key topics," said Mr. Read. "Then I would sit down with the attorney during the noon break, and discuss with him what he wanted."

A typical attorney request would be: "This guy said something yesterday about what the deceased intended her will to say. Find me where he said it, and find the exact words he used, because I think he said something wrong."

A request like that could probably be answered with a proximity search like "WITHIN 5 LINES OF INTEN* OR WISH OR SAY OR INDIC*," said Mr. Read. "After you've been using it for a while, you can usually find what the attorney's looking for with little effort."

Four hours in Unix, next day on the PC

Unix can support 256 megabytes of RAM, noted Mr. Read. "This means that, in a big case, with half a million documents on line, you can have a report on every document in which a witness' name appears on the attorney's desk in one to four hours," he explained. "With a PC, you couldn't give it to him till the next day."

Orrick, Herrington used a proprietary litigation support package written in Informix. "This software was the Maseratti of litigation support at the time," said Mr. Read. A similar package, written in the Unix data base Ingress, is available from Advocate Systems.

"We tend to use something like Paradox, and every time I want to work on a new case, I have to spend quite a bit of time and effort to design a new system for it," said Mr. Read. "Advocate Systems has put a lot of thought into creating a generic litigation support product. In about a week, you can get your coding set up and start loading a million documents into this data base," he said. "With Paradox, you could create a simple index in an hour, but if you wanted an elaborate tracking system, with full entry screens, and pre-set search and reports options, it could take a couple of months."

Benefits of Ipso Facto

"For the PC, we're using Ipso Facto primarily," said Mr. Read. "Ipso Facto is easy to use, it's quick to load into a searchable format, it permits you to search not only transcripts but also memoranda, letters, and other documents in the WordPerfect or other native format, and put them all into the same data base," he said. "It allows you search as many files as you want in rapid succession, and has a nice feature of allowing you to create separate scratch files comprised of the juiciest, most valuable portions of the documents. If I'm searching 15 depositions I can go through them interactively, and once I find a couple of paragraphs that I like, I can highlight those and cut and paste them off into another notepad, and keep cutting and pasting any of the sections I want into as many notepads as I want." Furthermore, Ipso Facto doesn't charge for converting diskettes into proprietary formats, he added.

Patent litigation with Q&A, Discovery ZX

An attorney at a mid-west intellectual property firm declined to be identified because he had concerns about whether the data base would be the subject of discovery in continuing litigation. "I've never heard of any cases on the subject, but judges have lots of latitude," he said.

Furthermore, he said, "I don't know that we'd want our competitors to know what we're using."

"For the big jobs, we've used CDC's Quorum, time-sharing on a mainframe by modem," he said. "For small and medium jobs, Q&A."

Paradox "has a lot of power, but didn't seem to be worth the agony," said the attorney, who used to be a mechanical engineer, and did a "fair amount" of programming, before law school.

They use Discovery ZX for full-text deposition tracking. "Discovery ZX is very easy to use, because the court reporters will prepare the disks for you and you can just pop them in," he said. "I would warrant that any case of any size is worth spending the extra $35 to get the disk."

Q&A or service bureau?

"We do a balance between patent prosecution and litigation," he said. "I've been mostly involved in the litigation. Typically the clients are big companies that have a lot of documents. The nature of the litigation is typically document intensive, so it's an area that is ripe for using the computer to keep things organized." They use a service bureau with Quorum to keep track of the names and dates in big cases, of 250,000 pages and up, and Q&A in-house for the same applications, "only I find it much easier with Q&A to customize the data base, add fields as needed, and generate reports."

The decision to go with the service bureau or in-house depends on the volume, with the crossover point at about 25,000 pages, he said. "For a medium to small-sized law firm it's impractical to hire people to do coding yourself in-house," he said. "For the big jobs we want someone else to handle all the coding and maintenance of the data base." And for the larger data bases, PC hardware is not practical.

"We've probably put 10,000 or 15,000 documents into data bases coded into Q&A, and it seems to work well in that volume," he said.

The document strategy

"The beginning of the process would be an attorney, or group of attorneys, deciding what information they want to glean from the documents," he said. The next step is to create a list of key words, and other document attributes they want to capture. The most useful information is the name, date, and document title or "Re:" line, and if the job is important enough, he would include key words. The paralegals would highlight the key words and synonyms. "It's very useful for chronologizing," he said. "If you put in titles and key words, the big picture can come together very quickly. If you have coders that are sophisticated enough, they can give you a capsule in place of a 'Re:' line that can serve as a substitute." Then a secretary or typist would key the information into Q&A.

"It's very easy, prior to a deposition, to get a list of all the documents that someone authored, or all the documents that contain that individual's name," he said.

"Typically either an attorney or paralegal would, as the need arose, perform queries into the data base, either to generate printed lists, or to just review the particular subset of the data at the computer itself," he said.

What is the meaning of the words in the claims?

"Patent cases are probably the bulk of litigation of intellectual property firms, because patents involve particular technical concepts and indeed particular words that define the invention," he said. "You typically want to key-word anything that has to do with the language and claims portion of the patent. The claims define the invention, and a lot of times it boils down to, 'What do those words mean?' so you want to be able to find all the documents that refer to the language, or synonyms of the language, so you can determine how those words were used."

"Typically you can use the computer to create subsets of pertinent documents that relate to a particular issue," he said. "If one of the issues in the suit revolves around the construction of a particular word in a claim, you can use that computer to locate the way that word was used everyday by the technical people who were working with the subject matter."

"There have been instances where the language of a patent claim was construed to be one thing in the law suit," he said, "but construed to be something else by all the technical people who used it in their everyday work."

Can your data base be subpoenaed?

In litigation, some attorneys like to flaunt their strengths, including their computer strengths, in the hope of intimidating their opponent into a more favorable settlement, but this attorney likes to play it more conservatively. In particular, he was worried about subpoenas from the adverse party, which is why he didn't want to be identified.

"If you put a field in with work product or mental process, the other side could subpoena everything else, or it's possible that a judge could say that I don't think that field should be kept out," he said. "You don't want to pay a lot of money and go through a lot of effort to establish a data base just to hand it over to them." And that could tip off your strategy. "If you've coded 100,000 documents of 1 million, the other side is going to want to know that that's what you're interested in.

"There's probably a kind of gray area as to whether these data bases are protected by the work product doctrine," he said. "The work product doctrine protects attorney impressions, mental thoughts and things like that. The issue is whether factual information that you have coded from documents is protected by the work product doctrine."

No cases, but probably not discoverable

"I don't know of any cases in which this was a problem," said Peter Brown, of Brown, Raysman & Millstein, New York City, Chairman of the Computer Law Committee of the Association of the Bar of the City of New York. "Simply because you use computer technology, no different rules apply than if you had made long and detailed lists on paper. On the whole, an opponent fears somebody who is well prepared."

Neil E. Aresty, of Aresty & Associates, Boston, Massachussets, didn't think it was likely either, but he thought out the arguments for both sides. "To the extent that the attorney has helped to design the structure of the data base, as well as code the subjective elements--the issues of the data base--I would hold that that is the attorney's mental notes and impressions, and is absolutely privileged," he said. "The fields that contain the objective information and bibliographic elements are not privileged."

"If the attorneys were involved in the design of the data base from the very beginning, it would be more likely to be protected," said Mr. Aresty. "It might be argued that the classification of document type could be a protected work product, even though that would be an objective data element."

Strategy and tactics of paper wars

Paul S. Hoffman, a sole practitioner in Croton-on-the-Hudson, New York, and specialist on computer contracts and software ownership, thought out the possibility that, if there is information in the data base that is pertinent to settling the case, the judge may lean towards disclosure that will expedite the litigation.

"If the content is merely transcribed depositions, which are frequently now available on automated media anyway, I doubt that a judge would order that data transferred," said Mr. Hoffman. "The second type of content would be where one side had prepared a data base of factual material, such as, in a patent case, identification of relevant patents, and that, I would think, would lean a lot more towards the discoverable side, if only because the other side would not have had an equal opportunity to develop it themselves," he said. "A third type of content gets into judgmental data base information, as to the accuracy of information, the analysis of information by experts or by lawyers, or paralegals," he said. "That kind of data is unlikely to be very discoverable."

If the game is to bury your opponent in paper, then discovery of your data base can give your opponent an advantage, said Mr. Hoffman. "In any large data base you're going to find an enormous amount of junk," he said, "and there's a question on the other party's side whether they have the right to obtain this information after the junk has been excluded, or whether they have to go through it and throw out the junk themselves. The identity of documents on which the party will want to be able to rely is probably discoverable."

What's a fair compensation?

"However, if counsel to one party invests an awful lot of time and effort into putting together and organizing a data base, then there's a question of essential fairness if the other party for either no expense or a nominal expense, picks up the same capability," said Mr. Hoffman.

"If there is a proposed order, it might very well be in terms of optional stages," said Mr. Hoffman. "This chunk you get for nothing, this chunk you should be able to get by paying not only the cost of producing it but a reasonable share in the cost of preparing it."

"I think there are situations in which the judge could rationally force the disclosing party to operate in a semi-service bureau manner," said Mr. Hoffman.