1. Paramount Pictures (1991).Google Scholar

2. In fiction, John Grisham is noteworthy for his attempts to give readers an insight into some of the techniques and stratagems of litigation, including discovery. See especially Grisham, John, The Rainmaker (1995).Google Scholar Not even Grisham has attempted to portray the process of drafting individual interrogatories and requests for production in order to maximize the chances of uncovering important evidence or the process of drafting creative and disingenuous responses to them to avoid divulging damaging information.

3. As we use the terms, every weakening of one's own case is a strengthening of an opponent's case, and conversely. Because the rules for criminal discovery can raise concerns about self-incriminauon, we do not address them in this paper. The great preponderance of document discovery takes place in civil litigation.

4. Fed. R. Civ. P. 26–37.Google Scholar

5. Friedenthal, Jack H. et al. , Civil Procedure (2nd ed.) 379 (1993).Google Scholar

6. Wright, Charles Alan, The Law of Federal Courts (4th ed.) 540 (1983).Google Scholar

7. 146 F.R.D. 402 (1993).

8. Washington State Physicians Insurance Exchange & Association v. Fisons Corporation, 122 Wn. 2d 299 (1993).Google Scholar

9. “[A]ll experienced trial lawyers do some ducking and dodging” in discovery practice. (A Tragic Asthma Case, Seattle Post-Intelleigencer, 01 17, 1991, at Bl.)Google Scholar

10. This is not a full definition. A full definition of “important evidence” would have to deal with the usual complications of joint causation and causal overdetermination. The problem of joint causation is that there may be a set of documents which would together make a substantial difference to the outcome of the case, even though no individual member would make a substantial difference alone. In such a case, all are important in our sense, because of their joint effects on the outcome of the case. The problem of causal overdetermination is that there my be two (or more) documents that convey much the same evidence, so that failure to disclose one document would not affect the outcome of the case, as long as the other document was disclosed. In such a case, each document is important, in our sense, because each would substantially affect the outcome of the case were the other not to be disclosed. A complete definition of “important evidence” would not only have to incorporate these complications but would also have to incorporate iterated combinations of them (e.g., two sets of documents, S1 and S2, such that disclosure of the members of SI would substantially affect the outcome of the case, if the members of S2 were not disclosed, and vice versa). For simplicity, we do not here attempt to state the full, inductive definition that would be necessary to cover all these complications.

11. Perhaps the most disturbing example of the nondisclosure of relevant evidence has been the carefully orchestrated strategy of the tobacco industry to use attorney-client privilege as a dodge to prevent production of hundreds of thousands of documents over four decades. After reviewing more than 800 of these documents, Minnesota Judge Kenneth J. Fitzpatrick said they showed a “conspiracy of silence and suppression of scientific research” (Tobacco Documents Show ‘Conspiracy of Silence and Suppression,’ Judge Says, Wash. Post, 12 18, 1997, at A14Google Scholar). For examples from the asbestos industry, see Brodeur, Paul, Outrageous Misconduct (1985).Google Scholar For examples from the auto industry, see Lasso, Rogelio A., Gladiators Be Gone: The New Disclosure Rules Compel a Reexamination of the Adversary Process, 36 B. C. L. Rev. 479–526 (1995).Google Scholar Lasso includes a partial list of cases in which automobile manufacturers and their attorneys “have been found to conceal relevant information despite specific discovery requests and court orders” (n. 96 at 500), of which the most well-known is General Motors' withholding of crashtest results on its fuel tanks from more than 100 different plaintiffs who had sued G M concerning truck fires. These reports were not disclosed by G.M. or its attorneys until they were uncovered independently by plaintiff's attorneys in Mosely v. General Motors Corp., No CA-90V6276 (Ga. State Ct., Fulton County; 02 26, 1993).Google Scholar

12. Fed. R. Civ. P. 26.Google Scholar The “mandatory disclosure” provisions of the 1993 amendments are discussed in Section IV, infra.

13. Lang, Angela R., Mandatory Disclosure Can Improve the Discovery System, 70 Ind. L. J. 657, 675 (Spring 1995).Google Scholar

14. Baird, Douglas B. et al. , Game Theory and the Law (1994).Google Scholar

15. Unfortunately, in our view, out of a desire for deductive rigor, much of the game-theory literature has become so technical as to be impenetrable to most readers. In addition, the demands of deductive rigor lead to overly simple models based on implausible but mathematically tractable assumptions. In contrast, we employ game theory to explain certain tendencies that would be expected to hold, all other things being equal. Discussion of the technical assumptions appears in footnotes.

16. Descartes, Rene, Meditations (1641).Google Scholar

17. For more on the general subject of social epistemology, see Goldman, A. I., Foundations of Social Epistemics, 73 Synthese 109–44 (1987)CrossRefGoogle Scholar; Goldman, A.I., Epistemic Paternalism: Communication Control in Law and Society, 88 J. Phil 113–31 (1991)CrossRefGoogle Scholar; Goldman, A.I., Liaisons: Philosophy Meets the Cognitive and Social Sciences pt III (1992)Google Scholar; Goldman, A I.Argumentation and Social Epistemology, 91 J. Phil. 27–49 (1994)CrossRefGoogle Scholar; Goldman, A. I., Social Epistemilogy, Interests, and Truth, 23 Phil. Topics 171–87 (1995)CrossRefGoogle Scholar; Cox, J. C. & Goldman, A.I., Accurary in Journalism: An Economic ApproachGoogle Scholar, in Schmitt, F., ed., Socializing Epistemology 189–215 (1994)Google Scholar; Goldman, A.I. & Cox, J.C., Speech, Truth, and the Free Market for Ideas, 2 Legal Theory 1–32 (1996)CrossRefGoogle Scholar; and Goldman, A.I., Knowledge in a Social World (forthcoming 1999).Google Scholar

18. This historical segment follows closely the account of Langbein, John, Torture and Plea Bargaining, 58 Pub. Interest 43–61 (1980).Google Scholar

19. Tehan v. Shott, 382 U.S. 406, 416 (1966).Google Scholar

20. Fed. R. Evid. 102 (emphasis added).

21. Id. at 401.

22. Id. at 402.

23. Cleary, E., ed., McCormick on Evidence 23 (3d ed. 1984).Google Scholar

24. Fed. R. Evid. 804.

25. Id. at 804(b) (3).

26. Id. at 804(b) (5).

27. Id. at 603.

28. Id. at 403.

29. Id. at 701, Advisory Committee Notes.

30. Sward, E., Values, Ideology, and the Evolution of the Adversary System, 64 Ind. L. J. 326, 329 (1989).Google Scholar

31. Hickman v. Taylor, 329 U.S. 495 at 500 (1947).Google Scholar

32. Fed. R. Civ. P. 26(b)(1).

33. The criminal and civil justice systems both depend on accurate determinations of the truth. From the point of view of social epistemology, the most important difference between them is in how mistakes are weighted. In the civil justice system, the two kinds of mistake are weighted fairly equally—that is, the outcome in which a deserving plaintiff loses (and an undeserving defendant wins) is weighted about the same as the outcome in which an undeserving plaintiff wins (and a deserving defendant loses). In contrast, in the criminal justice system, the two kinds of potential mistake are weighted very unequally—that is, the outcome in which an innocent defendant is convicted is regarded as much worse (receives greater negative weight) than the outcome in which a guilty defendant is not convicted. This leads to different standards of proof for civil (“preponderance of evidence”) and criminal (“beyond a reasonable doubt”) matters, and, in some jurisdictions, to weaker requirements for verdicts in civil matters, than in criminal matters (e.g., less than unanimous verdicts in civil but not criminal matters).

We should also mention that although our primary focus is on the question of how to improve the civil justice system as a vehicle for determining legally relevant truths, any acceptable answer to this question must also have an acceptable economic cost The economic considerations operate as the following background constraint on our analysis: We only consider potential changes that will not increase the cost of the civil justice system. As we explain below, we believe that our proposal will decrease economic costs while increasing the reliability of factual determinations by the trier of fact.

34. Of course, the entire system may be lacking crucial evidence, which makes the task of accurate fact-finding extremely difficult. Critical eyewitnesses may be deceased, documents may have been destroyed, and so forth. Under such circumstances, however, no procedures could guarantee a high probability of truth.

35. For criticisms of the adversary system from this vantage point, see Luban, D., Lawyers and Justice 68–74 (1988)Google Scholar; Simon, W.H., The Ideology of Advocacy: Procedural Justice and Professional Ethics, 1978 Wis. L. R. 29–144 (1978)Google Scholar; Golding, M.P., On the Adversary System and JusticeGoogle Scholar, in Bronaugh, R., ed., Philosophical Law 106–12 (1978)Google Scholar; and Goldman, A.H., The Moral Foundations of Professional Ethics 112–16 (1980).Google Scholar

36. See Langbein, J.H., The German Advantage in Civil Procedure, 52 U. Chi. L. Rev. 823 (1985).CrossRefGoogle Scholar

37. The adversary system has already been subject to some dilution and modification in various domains of American adjudication. See Sward, , supraGoogle Scholar note 30. Sward includes under this heading: class actions, specialized courts, masters and court-appointed experts, case managers, and especially alternative dispute-resolution devices. Recendy, there have been even more radical experiments, including experiments with written questions of witnesses from jurors. New Plans for Jury Improvements: In the Works: More Judicial Supervision, Questioning by Jurors, Better Screening, Nat'lL. J. A17 (11 13, 1995).Google Scholar

38. On our view, the adversary system itself is not inherendy just or unjust It is merely an instrument for achieving just results. We return to this topic in Section VII below.

39. Schwarzer, W. W., The Federal Rules, the Adversary Process, and Discovery Reform, 50 U Pitt L. Rev. 703, 712 (1989).Google Scholar

40. This is not quite true. Even in a single agent the evidence may be distributed in different parts of memory, which creates a problem of access that is somewhat analogous to the access problem for organizations. See Cherniak, C., Minimal Rationality 49–71 (1986)Google Scholar and Goldman, A. I., Philosophical Applications of Cognitive Science 9–13 (1993).Google Scholar

41. It is in theory possible for both sides to obtain net benefits from mutual disclosure, and thus in theory possible for both sides to be rationally motivated to agree to mutual disclosure. However, our impression is that lawsuits in which mutual disclosure would, on balance, benefit both parties are rare. And even in such cases, it would be very unlikely that the attorneys involved could agree upon terms for disclosure that would make it rational for each party to believe that the others would actually comply with the terms of the agreement Thus, in the absence of court rules backed up by sanctions, we would expect little or no pretrial discovery.

42. Thus, the U.S. Supreme Court has stated that the purpose of discovery rules is “for the parties to obtain the fullest possible knowledge of the issues and facts before trial” (Hickman, v. Taylor, , supra note 31, at 500Google Scholar) and to “make a trial less a game of blind man's bluff and more a fair contest with the basic issues and facts disclosed to the fullest practicable extent” (U.S. v. Procter & Gamble Co., 356 U.S. 677 at 683 (1958)).Google Scholar

43. See Fed. R. Civ. P. 37(b) (2) for the variety of sanctions that may be imposed, of which refusing to admit the material into evidence is only one alternative, and not the harshest.

44. We believe that discovery of expert witnesses is handled satisfactorily by current practice under the existing discovery rules. See Fed. R. Civ. P. 26(b)(4). The introduction of expert testimony at trial is quite another matter. It is one of the scandals of the U.S. legal system. The current system of expert testimony has given rise to a profession of expert witnesses who function as “hired guns” by tailoring their testimony to the purposes of the party paying their bills. The result has been a great increase in the expenses of litigation and often very little if any, contribution to the determination of the relevant facts. From the point of view of the Truth Rationale, this is one more disturbing feature of the adversary system.

45. Fed. R. Civ. P. 26(b)(1). Prior to the 1993 amendments to the federal discovery rules, the tools for discovery included written interrogatories (questions) directed to a party for written replies (Fed. R. Civ. P. 33Google Scholar); written requests for production of documents and other things (Fed. R. Civ. P. 34Google Scholar); written requests for entry upon land for inspection and other purposes (Fed. R. Civ. P. 34Google Scholar); written requests for physical and mental examinations of persons (Fed. R. Civ. P. 35Google Scholar); written requests to a party to make admissions of fact (Fed. R. Civ. P. 36Google Scholar); and the taking of testimony under oath from potential witnesses and others with relevant information (Fed. R. Civ. P. 30, 31Google Scholar). The 1993 amendments did not eliminate any of these tools, but added some “mandatory disclosure” requirements and set limits on the use of the pre-1993 tools (Fed. R. Civ. P. 26 (a) (b)).Google Scholar

46. Supra note 8. Our summary of the case is based on the Washington State Supreme Court's summary at 306–10, and on reports in the press.

47. The trial court reduced the award for loss of professional consultations from $150,000 to $2,250, but left standing an award of $1,085,000 for injury to professional reputation and an award of $2,137,500 for physical and mental pain and suffering. The State Supreme Court disallowed the awards for loss of professional consultations (WSPIEA v. Fisons, supra note 8, at 332Google Scholar) and the award for physical and mental pain and suffering (WSPIEA v. Fisons, supra note 8, at 318–23Google Scholar), but left standing the $1,085,000 award for injury to professional reputation (WSPIEA v. Fisons, supra note 8, at 332).Google Scholar

48. In WSPIEA v. Fisons, the proceedings were governed by the Washington State Superior Court Civil Rules. Though the Washington rules were not identical to the pre-1993 federal discovery rules, they were modeled on the federal rules, and the relevant provisions, including rule and section numbering, are practically equivalent for our purposes. The pre-1993 federal discovery rules are still the basis for the discovery rules in almost all of the states, and in the nearly one-third of federal district courts that have “opted out” of the “mandatory disclosure” provisions of the 1993 amendments.

49. The classic work on the theory of two-person, zero-sum games is John Von Neumann & Oskar Morgenstern, Theory of Games and Economic Behavior (1944). Following the usage of von Neumann and Morgenstern, we use the term “game” very generally to refer to any strategic interaction.

50. In the U.S. legal system, as in the medical system, the predominant model is payment for services rendered, and thus one of the major monetary motivations for attorneys is to maximize the number of services rendered. It is easy to see how this fact can give attorneys on opposite sides of a lawsuit a common interest (in multiplying legal services and thus their legal fees), which potentially conflicts with the interests of their clients. This is one of many ways in which part of the interaction between the attorneys better fits the model of what Axelrod calls an “iterated Prisoner's Dilemma” (Axelrod, Robert, The Evolution of Cooperation (1984))Google Scholar and what Taylor calls a “Prisoner's Dilemma Supergame” (Taylor, Michael, The Possibility of Cooperation (1986)).Google Scholar

51. Baird, et al. , supra note 14, at 89–90.Google Scholar A striking example of the unraveling principle at work comes from biology. Because the deepness of a toad's croak is positively correlated with its size, and because size is an important variable in determining evolutionary fitness, it might be thought that when a female announces her interest in mating, only the largest toads (with the deepest voices) would croak in reply. But all toads will croak, because they all gain an advantage over toads smaller than they are by croaking, and no advantage over toads bigger than they are by not croaking. So the result is that, if a female announces her availability, all the toads croak in response, thus making it easy for her to determine which is the largest. See Frank, Robert, Passions Within Reason 105 (1988).Google Scholar The classic statement of the unraveling result is Akerlof, George, The Market for ‘Lemons’, 84 Q. J. Econ. 488–500 (1970).CrossRefGoogle Scholar It should also be noted that here, and throughout, we limit our discussion to information that is verifiable in the sense in which used by Baird, et al. , supra note 14, at 89—Google Scholarthat is, that it can be readily checked once it is revealed.

52. Id. at 90–95.

53. Id. at 95.

54. Id. at 262–66.

55. The amount ($10 million) is chosen to make the amounts comparable to the amounts reported in WSPIEA v. Fisons, supra note 8Google Scholar, and to simplify the arithmetic. As a limit on the defendant's liability, the amount is, if anything, too low. For example, in August 1994, an Oregon jury awarded $28.6 million to a man who suffered brain damage after using a common asthma drug. See Asthma sufferer wins $28.6 million award, Seattle Times, 09 3, 1994Google Scholar, at Bl. Also, the cost to a party of disclosing negative evidence may far exceed the potential loss from the current lawsuit, because the disclosure will also increase the probability of further lawsuits by other plaintiffs, and will increase the probability of the other plaintiffs' prevailing also. For example, it was not until the disclosure of the Sumner Simpson papers that the tide turned in favor of plaintiffs in the asbestos litigation. See Brodeur, , supra note 11, at 109–11.Google Scholar The cost of those papers to asbestos defendants might well be measured in the billions of dollars.

Our model is an extreme oversimplification. To make our model more realistic, it would be necessary to define two probability density functions: (1) a probability density function over all possible dollar measures of damages, from $0 to $10 million; and (2) a probability density function over degrees of liability of the defendant from zero (defense verdict) to 100%. Adding these complications would not require any substantial changes in our conclusions, but they would make the mathematical details much more daunting, so we ignore them.

56. In an actual case, it would be unusual for there to be a single “smoking gun” document. As mentioned above, in Wspiea v. Fisons, supra note 8Google Scholar, there were two. Often there will be no individual “smoking gun” documents, but a group of documents will collectively amount to a “smoking gun”—e.g., the Sumner Simpson papers discovered in the asbestos litigation (Brodeur, , supra note 11, at 111).Google Scholar It simplifies the discussion of our hypothetical case to assume a single “smoking gun” document.

In our hypothetical example, we assume that the defendant's attorney knows of the existence of the “smoking gun” document from the outset of the case. In WSPIEA v. Fisons, the chronology was more complex. The attorneys for defendant Fisons did not know of the “smoking gun” documents when they made their first responses to discovery requests in November 1986. They became aware of the first “smoking gun” document in March 1987. At no time after they became aware of it did they supplement their previous responses; nor did they disclose it in their responses to subsequent discovery requests made in June 1987. For information on the chronology of discovery in WSPIEA v. Fisons, see Taylor, Stuart Jr., Sleazy in Seattle, 16 Am. Law. 5, 74–79 (04 1994).Google Scholar

57. In a simple shell game with n shells (n > 1), one player (Hider) hides a pea under one of the n shells and a second player (Finder) guesses under which shell the pea is hidden. Finder wins the game if she guesses correctly. Hider wins the game if Finder guesses incorrectly. In this simple two-person, zero-sum game, Hider's best strategy is to randomize so that the probability that any particular shell harbors a pea is 1/n. As a consequence, Finder has no better strategy than to choose randomly among the shells. In the Discovery Game that we describe in the text, the Hider (defendant's attorney) does not actually hide the pea, but her best strategy, if possible, is to give Finder (plaintiff's attorney) only information that makes Finder's best strategy to choose randomly among the shells. Thus, the Discovery Game is a close relative of the actual shell game.

58. In a typical case, there would probably be more than one “threatening” request for production. For example, in WSPIEA v. Fisons, there were requests for production from both the child's attorney and the attorney for the child's doctor that would have required disclosure of the “smoking gun” documents had not the defendant's attorneys objected to them (supra note 8, at 348–50).Google Scholar It simplifies the discussion of our hypothetical case to assume a single “threatening” request for production.

59. Cf. WSPIEA v. Fisons, supra note 8, at 307–308.Google Scholar

60. Here we follow very closely the facts of the Fisons case. See WSPIEA v. Fisons supra note 8, at 346–54.Google Scholar

61. To keep the example simple, we describe only one side of the Discovery Game where the defendant's attorney takes the role of “Hider” and the plaintiff's attorney takes the role of “Finder.” In a typical lawsuit, both attorneys will play both roles.

62. This is exactly the move made by the attorneys for defendant Fisons in WSPIEA v. Fisons. The court said this about the unilateral narrowing of the scope of the requests for production: “There was no clear indication from the drug company that it was limiting all discovery regarding Somophyllin Oral Liquid to material from that product's file. Nor was there any indication from the drug company that it had … information about Somophyllin Oral Liquid that it was not producing because the information was in another product's file. The doctor was justified in relying on the statements made by the drug company's attorneys that all relevant documents had been produced and he cannot be determined to have impliedly, albeit unknowingly, acquiesced in limiting the scope of discoverable information” (supra note 8, at 353).Google ScholarSee also infra note 66. In WSPIEA v. Fisons, the attorneys for defendant Fisons had to be even more creative than our simple example would indicate, because they had to argue that one of the “smoking gun” letters that began “Dear Doctor” was not really a “Dear Doctor” letter, and thus did not fall within the request for “Dear Doctor” letters made by the attorneys for the defendant doctor, even though it and letters like it had been referred to as “‘Dear Doctor’ letters” in internal Fisons communications (Taylor, , supra note 56, at 75).Google Scholar

63. Objections of each of these four types, and many other types besides, were raised by the attorneys for defendant Fisons in WSPIEA v. Fisons. See infra notes 66, 67, and 70.Google Scholar

64. Thus, the solution to the Discovery Game is formally analogous to the solution to the Shell Game—defendant's attorney tries to give the plaintiff's attorney no clue as to which “shell” might have a “pea” under it, so that from the point of view of plaintiff's attorney, each “shell” is equally probable.

65. For example, in WSPIEA v. Fisons, in response to particular requests for production from the attorneys for the child's doctor, the defendant's attorneys incorporated into their responses two pages of general objections, attached as Exhibit A (supra note 8, at 346, n. 86).Google Scholar

66. For example, in WSPIEA v. Fisons, the defendant's attorney preceded his responses with the following general objection, among others:

“Requests Regarding Fisons Products Other Than Somophyllin Oral Liquid. Fisons objects to all discovery requests regarding Fisons products other than Somophyllin Oral Liquid as overly broad, unduly burdensome, harassing, and not reasonably calculated to lead to the discovery of admissible evidence” (supra note 8, at 348).Google Scholar For an even more extreme example of the “piling on” of objections, see the defendant's response to Interrogatory No. 2 quoted in note 86 (supra note 8, at 346).Google Scholar

67. In WSPIEA v. Fisons, in response to a “threatening” request for production from the attorneys for the child's doctor, defendant's attorneys first invoked their general objections and then continued: “Fisons further objects to this request to the extent it calls for expert disclosures beyond the scope of CR 26(b) (4) or which may be protected by the work-product and/or attorney-client privilege. Without waiver of these objections and subject to these limitations, Fisons will produce documents responsive to this request at plaintiffs' expense at a mutually agreeable time at Fisons' headquarters” (supra note 8, at 349Google Scholar, emphasis added). We discuss the work-product doctrine and attorney-client privilege in Section VI below.

68. A similar strategy of this kind is to limit disclosure to a specified time period (e.g., all documents dated within three years of the injury), where the time period is defined by the responding party's attorney to carefully exclude the production of known negative evidence (e.g., “smoking gun” documents dated four years before the injury).

69. In many cases, local rules require that the attorneys discuss the basis for discovery objections as a precondition for filing a motion to compel discovery. If, in our hypothetical example, the plaintiff's attorney were to call the defendant's attorney to discuss her discovery responses, we would expect the defendant's attorney to be motivated to seem cooperative, without disclosing any information that might help to identify the location of any “peas.”

70. In WSPIEA v. Fisons, we know that the attorneys for the child or the doctor discovered that the defendant manufactured another theophylline-based medication called “cromolyn,” because they explicitly requested production of documents regarding cromolyn products. The defendant's attorney objected as follows: “Defendant Fisons objects to this discovery request as not reasonably calculated to lead to the discovery of admissible evidence, as overbroad in time, and as incredibly burdensome and harassing. This discovery request encompasses approximately eighty-five percent of all documents in the subject files and departments—millions of pages of documents. Neither cromolyn (which should be referred to as cromolyn sodium), nor any cromolyn product, nor the properties or efficacy of cromolyn is at issue in this litigation. Furthermore, Fisons objects to this discovery request as calling for the production of extremely sensitive trade secret and proprietary material” (122 Wn.2d 299 at 350). We did not include in our hypothetical story the discovery of the cromolyn sodium product by plaintiffs attorney, because we do not believe that production of the “smoking gun” documents should have depended on the plaintiffs attorney discovering the existence of the cromolyn sodium product We agree with the court in WSPIEA v. Fisons that, regardless of where they were filed by the defendant, documents regarding theophylline-toxicity were directly relevant to the toxicity of Somophyllin Oral Liquid and should have been produced in response to requests for production of documents relating to dangers of the Somophyllin product. Cf. WSPIEA v. Fisons, supra note 8, at 353.Google Scholar

71. In WSPIEA v. Fisons, in support of a motion for a protective order (apparently to limit disclosure to Somophyllin product files), defendant's attorney filed an affidavit that stated: “Following receipt of plaintiffs' First Request for Production, I traveled to Fisons, in Bedford, Massachusetts, in order to ascertain firsthand the scope and extent of documents responsive to plaintiffs' request for production. At that time I confirmed that to produce all of the documents responsive to plaintiffs' catch-all requests would be extremely burdensome and oppressive to Fisons. Between one and two million pages of documents, most of which have no colorable relevance to the issues in this action, would have to be located, assembled, and made available for review or copying. The time, expense, and intrusion upon the day-to-day business activities of Fisons would be immense” (supra note 8, at 351).Google Scholar In a memorandum in support of the motion for protective order, defendant's attorneys pointed out that the Fisons' Regulatory File on Somophyllin Oral Liquid had already been made available to the plaintiff's attorneys and argued disingenuously. “In short, Fisons' Regulatory File for Somophyllin Oral Liquid contains all or nearly all documents in Fisons' possession that are reasonably related to plaintiffs' failure-to-warn allegations” (supra note 8, at 352Google Scholar, emphasis added). Later, when the first “smoking gun” document was leaked, the attorneys for the child's doctor moved for an order compelling discovery. The order compelling discovery was granted, and defendant's attorneys produced 10,000 documents, including the second “smoking gun” document, the following day (supra note 8, at 308).Google Scholar

72. WSPIEA v. Fisons, supra note 8, at 307.Google Scholar

73. We do not assume that a party may never properly benefit from or take advantage of information asymmetries. For example, in settlement negotiations, each attorney quite properly keeps confidential any special reasons that his or her client may have to settle the lawsuit Our discussion here is limited to the effect of information asymmetries in responding to discovery requests. It would take us far beyond the scope of this article to consider when it can be proper for a party to benefit from or take advantage of information asymmetries.

74. In our experience, when attorneys claim that documents of this kind are “not relevant” they are often actually making an even more controversial determination—for example, that what is obviously “relevant evidence” in the technical sense (Fed. R. Evid. 401)Google Scholar should be excluded because “its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury” (Fed. R. Evid. 402).Google Scholar Clearly, these determinaaons should never be made unilaterally by a party's attorney. They should always be made by a judge or special master.

75. The classic statement of this sort of procedural justice view is attributed to Samuel Johnson by his biographer, James Boswell:

Boswell: But what do you think of supporting a cause which you know to be bad?

Johnson: Sir, you do not know it to be good or bad till the Judge determine it.

James Boswell, Life of Johnson, quoted in Simon, , supra note 35, at 39.Google Scholar

76. This seems to have been the underlying rationale for the testimony of many of the fourteen experts, including at least one of the two professors of legal ethics (Boerner), at the hearing on discovery abuse that took place before the trial court in WSPIEA v. Fisons. See Taylor, , supra note 56, at 76–78.Google Scholar Our impression is that the Fair Fight Rationale is even more widespread among trial attorneys in criminal litigation, and its prevalence is easily explained in the same way that we explain its prevalence among civil litigators. A criminal attorney motivated by the Truth Rationale (TR) and considerations of substantive justice would be at a substantial competitive disadvantage in a competition with criminal attorneys motivated by the Fair Fight Rationale (FFR) and considerations of mere procedural justice, because an advocate of the FFR will regard it as a success when a guilty (in the substantive sense) client goes free (because found “not guilty”)

77. It may seem that the sort of argument we have made to explain adverse selection against TR attorneys would also lead to the expectation that there would be adverse selection against FFR attorneys if they were in competition with attorneys whose pursuit of material advantage were totally unconstrained. Attorneys who seek to maximize material advantage (MA attorneys) would tend to displace attorneys who advocate the Fair Fight Rationale, if there were situations in which the FFR would lead an attorney not to maximize his or her material advantage. This would be the case, if, for example, bribing judges or juries were a cost-free alternative, because FFR attorneys would not be willing to engage in bribery.

Although our impression is that FFR attorneys are not being selected against in general, they certainly are at a competitive disadvantage to MA attorneys in the legal representation of members of organized crime and drug cartels. In those cases, we believe that selective pressures definitely tend toward the extinction of FFR attorneys. This is a useful reminder that, although we believe that the Fair Fight Rationale is flawed, it is certainly preferable, from our point of view, to some of the alternative motivations that might displace it.

Moreover, we do not discuss in this article the sort of middle position of someone who accepts the Truth Rationale for the discovery system itself, but who believes that, properly understood, that rationale does not require an individual attorney to place oneself at a competitive disadvantage with respect to those who operate on the Fair Fight Rationale We refer to this hybrid as a TR/FFR attorney. A TR/FFR attorney will behave no differently from an FFR attorney. But we still believe that a TR/FFR attorney will be at a competitive disadvantage with respect to a purely FFR attorney, owing to the psychic costs involved in behaving as an FFR attorney. In the hypothetical case discussed in the text, an FFR attorney would be proud of having gotten away with nondisclosure of the “smoking gun” document and proud that he or she was able to win the case (if, as expected, successful nondisclosure led to her client's winning of the case). In contrast, a TR/FFR attorney would be disturbed at such a result. We believe that this psychic toll is often quite large, and that it is responsible for many attorneys' decisions to move into areas of practice that do not involve litigation.

78. Fed. R. Evid. 40

79. Taylor, , supra note 56, at 78.Google Scholar

80. Sorenson, Charles W. Jr., Disclosure Under Federal Rules of Civil Procedure 26(a)—‘Much Ado About Nothing?’, 46 Hastings L.J. 679, 784 (03 1995).Google Scholar

81. Taylor, , supra note 56, at 77.Google Scholar

82. When the attorneys for the child's doctor in WSPIEA v. Fisons first brought their motion for discovery sanctions in the trial court, the attorney against whom the sanctions were sought was reported to have said that he felt personally attacked by the motion. See judge refuses to impose fines in asthma drug case, Seattle Post-Intelligencer, 01 31, 1991Google Scholar, at B1, B3. The attorney who represented him and his firm on the motion for sanctions, “considered one of the most prominent attorneys in Seattle,” called the rule under which the motion was brought “a disgrace.” (A tragic asthma case, Seattle Post-Intelligencer, Jan. 17, 1991, at B1.) We believe that this is yet a further indication of the power of the forces of selection that we have described: Litigation practice selects for (because it rewards) attorneys with biased standards for what is, and is not, disgraceful.

83. See, e.g., Sorenson, supra note 80.

84. As defined in supra note 77, a TR/FFR attorney would behave as an FFR attorney and not comply, but, unlike an FFR attorney, would also feel guilty for not complying with them.

85. WSPIEA v. Fisons, supra note 8, at 344.Google Scholar

86. Id. at 5, 76–78.

87. WSPIEA v. Fisons, supra note 8, at 352.Google Scholar The court added: “Having read the record herein, we cannot conceive of any request that could have been made to this drug company that would have produced the smoking gun documents. Unless the [other party] had been somehow specifically able to request the June 30, 1981, “Dear Doctor” letter, it is unlikely that the letter would have been discovered. Indeed, the drug company claims the letter was not an official “dear doctor” letter and therefore was not required to be produced”(id. at 354).

88. Id. at 353.

89. Id. at 343.

90. Id. at 343.

91. Id. at 345.

92. One respect in which our hypothetical case differs from the facts in WSPIEA v. Fisons is that we suppose that the plaintiffs attorney moved to compel production of non-Somophyllin documents before the “smoking gun” document was leaked. As we explain below, we do not mean to imply that the defendants were under no duty to disclose the “smoking gun” documents in the absence of a motion to compel discovery, but only to explain why, under the Old Henry standards, it is not at all unlikely that the motion to compel production of non-Somophyllin documents would have been denied and that the defendants would have been granted a protective order for non-Somophyllin documents.

93. “Rather it is the misleading nature of the drug company's responses that is contrary to the purposes of discovery and which is most damaging to the fairness of the litigation process”(WSPIEA v. Fisons, supra note 8, at 346Google Scholar, footnotes omitted). See also id. at 352.

94. WSPIEA v. Fisons, supra note 8, at 308.Google Scholar

95. Although our example involves a biased interpretation of the rules by a defense attorney, we do not mean to imply that plaintiffs attorneys are not also rewarded for biased interpretations of the rules that favor their clients. The problem is a symmetric one, although in many types of lawsuit (e.g., product liability) the effects of defense attorney bias on discovery are more significant, simply because almost all of the relevant evidence is in the possession of the defendant

96. Because our concern in this article is the problem of known individual documents with negauve evidence, we limit our proposal to cases of that kind. However, we believe that a general proposal should also address the related problem that arises when a large covering class of documents includes known subclasses that are believed to be much more likely to contain negative evidence than the covering class as a whole. Because a solution to this “known subclass” problem would introduce further complications, we do not address it here.

97. Obviously, there will be situations in which some information disclosed to the judge or special master cannot be disclosed to the other parties. This is clearest in the case of in camera review of documents.

98. If our three proposals were adopted in some form, we would expect that they would be augmented by a strong duty to supplement one's responses, so that a party and its attorney would not be rewarded for waiting to investigate the case until after they had responded to the other party's discovery requests. In the text, we do not take up the issue of the duty to supplement because we feel that it has been well-addressed by the 1993 amendments to the federal discovery rules. See Fed. R. Civ. P. 26(e).Google Scholar

99. Although we criticize some of the 1993 amendments, we do not wish to imply that they do not make any improvements over the pre-1993 rules. There are some clear improvements—for example, the strengthened duty to supplement responses under Fed. R. Civ. P. 26Google Scholar (e). Moreover, we do not even mean to claim that the mandatory disclosure provisions of the 1993 amendments do not offer some improvements over the pre-1993 rules. We limit our discussion to document production, where we do not expect much improvement for the reasons that we discuss in the text.

100. Amendments to the Federal Rules of Civil Procedure, 146 F.R.D. 501, 510 (1993)Google Scholar

101. Lasso, , Gladiators Be Gone, supra note 11, at 481–82.Google Scholar

102. Fed. R. Civ. P. 26(a)(1)(B).

103. Fed. R. Civ. P. 26(a)(1)(B), Advisory Committee Note.

104. It is perhaps not obvious why a “good faith” standard is not sufficient to generate disclosure of the “smoking gun” document. We do believe that if attorneys actually acted in good faith, the problems discussed in this article would not exist But that is to assume that the “good faith” standard is applied by “New Henry” attorneys who genuinely act in good faith. The situation is very different if the “good faith” standard is applied by “Old Henry” attorneys. Old Henry attorneys would simply treat the “good faith” standard as one more rule to be given a unilateral, biased interpretation for the benefit of their clients. The problem with such a rule is that it is exceedingly difficult to prove a violation of it. Consider, for example, the Fisons case. If the attorney who made the decision not to disclose the “smoking gun” documents testifies that he acted in “good faith,” it is not easy to see how the other side can rebut that testimony, because the other side will usually not know anything about the discussions and deliberations that led up to the decision not to disclose it Thus, while in principle, it would be desirable to require “good faith” from attorneys, in practice such a requirement places only minimal constraints on their behavior.

105. Fed. R. Civ. P. 26(b) (5).Google Scholar

106. Beckner, R. Bruce, Advance Sheet: Cooking up Document Production—or How Not to Get Burned by Smoking Guns, 20 Litigation 49–51 (Spring 1994)Google Scholar. This journal is published by the Litigation Section of the American Bar Association.

107. Id. at 51.

108. The 1992 Arizona rules require automatic production without the need for a written request (within a specified time period) of all documents “which the party believes may be relevant to the subject matter of the action, and those which appear reasonably calculated to lead to the discovery of admissible evidence” (Ariz. R. Civ. P. 26.1).

109. Birmingham, William T. & Onofry, Charles D., Mandatory Disclosure of Information: One State's Experience, 36 For the Def. 7, 9 (07 1994) (emphasis added).Google Scholar

110. Id. at 10.

111. Birmingham and Onofry believe the disclosure is easily avoided under the more limited 1993 federal mandatory disclosure rule: “Under the federal rule, one might rely on the fact that the overdriving claim is not one which is “alleged with particularity in the pleadings”, and, therefore, there is no obligation to disclose”(id. at 10). Thus, the federal mandatory disclosure rules have been effectively eviscerated.

112. Id. at 11.

113. Chanen, Jill Schachner, States Considering Discovery Reform, 81 A.B.A. J. 20 (04 1995).Google Scholar

114. Fed. R. Civ. P. 26(a)(1)(B).

115. Fed. R. Civ. P. 26(a)(1)(B), Advisory Committee Note.

116. Id.

117. For failures to respond to discovery requests or for evasive or incomplete responses, Rule 37 gives the court authority to make such orders “as are just,” including any of the following:

(A) An order that the matters regarding which the order was made or any other designated facts shall be taken to be established for the purposes of the action in accordance widi the claim of the party obtaining the order;

(B) An order refusing to allow the disobedient part to support or oppose designated claims or defenses, or prohibiting that party from introducing designated matters in evidence;

(C) An order striking out pleadings or parts thereof; or staying further proceedings until the order is obeyed, or dismissing the action or proceeding or any part thereof, or rendering a judgment by default against the disobedient party” (Fed R. Civ. P. 37 (b) (2)).

118. Fed. R. Civ. P. 37(b) (2) (E).

119. Lawyers Face Fines in Medical Case, Seattle Times, 09 21, 1993, at B1.Google Scholar

120. Wspiea V. Fisons, supra note 8, at 356 (footnotes omitted).Google Scholar

121. Id. at 356.

122. Id.

123. Bogle & Gates Agrees to Pay Sanction for Misconduct in Suit, Seattle Post-Intelligencer, 01 29, 1994, A1, A6Google Scholar. Apparently the $325,000 was paid to reimburse the Washington State Physicians Insurance Exchange & Association for its attorneys' fees on the case. According to newspaper reports, the defendant and its attorneys did not disclose how much of the fine would be paid by each (Fines Say It's Not OK to Withhold Evidence, Seattle Times, Jan. 30, 1994, at B2. Thus, it is impossible to determine how much, if any, of the fine was paid by the defendant's attorneys.

124. Bogle & Gates Agrees to Pay Sanction for Misconduct in Suit, supra note 123, at A6.Google Scholar

125. We had written these words long before it was reported that Bogle & Gates, the firm that was sanctioned for discovery abuse in the Fisons case (supra note 8), had been sanctioned again for discovery abuse. In November 1995, it was reported that Bogle & Gates had been sanctioned by U.S. District Court Judge Robert Bryan for discovery responses that denied the existence of certain types of documents, which subsequently were disclosed during deposition testimony. The sanctioning took place in May 1995, but was not publicly reported until the following November. Dismaying Discovery, Puget Sound Bus. J. 1, 52 (Nov. 17–23, 1995). Judge Bryan was quoted as scolding Bogle at the hearing: “I am … disappointed and a little disgusted—well, quite a little disgusted that we are in this situation” (at 52). The amount of the latest sanction was not reported. We believe that the negative publicity that the Bogle firm has received may be a more potent sanction than the monetary payments. We include the potential for sanctioning by negative publicity in our discussion of attorney discipline on page 149.

126. As before, we make simplifying assumptions to keep the mathematics elementary. A more realistic assumption would be that there is a probability density function defined over possible monetary sanctions, with expectation of $1 million.

127. There are two main simplifying assumptions: (1) Utility is linear in dollars; (2) there are no other consequences that matter to the defendant, or that such collateral consequences balance out—for example, the harm to the defendant's reputation from having nondisclosure detected (when discounted by the probability of being detected) might be much less than the discounted potential additional loss from future lawsuits if the plaintiff in this lawsuit is successful.

128. To calculate expected utility of the alternatives (P and −P) it is necessary to make utility assignments to the following four possible outcomes, listed in order of diminishing utility: (a) Win at Trial and Don't Pay Sanctions (W&−S), with utility 0; (b) Win at Trial and Pay Sanctions (W&S), with utility of −1; (c) Lose at Trial and Don't Pay Sanctions (−W&−S), with utility of −10; (d) Lose at Trial and Pay Sanctions (− W&S), with utility of −11. Given this information, the expected utility of the two alternatives, Produce (P) or Not to Produce (−P), can be calculated as follows:

It will be rational to Produce the “smoking gun” document just in case the Expected Utility of P is greater than or equal to the Expected Utility of −P, that is, just in case:

That is, just in case:

129. Of course, we would not expect an attorney and client to actually sit down and calculate the expected utility of the alternatives. But we believe that a successful sanctioning system should successfully deter attorneys from providing their clients with the following type of advice: “Of course, I cannot advise you to withhold evidence, and any evidence that you show me I will be obligated to produce. But if there were any ‘smoking gun’ documents—I'm speaking hypothetically now—you might find that when you compared the cost of revealing them to the other side with the potential costs of the other side's happening upon them you might find it to your advantage to keep them from me, and perhaps even to destroy them”. We believe that a genuine departure from the Old Henry standards requires a system that effectively deters advice of this kind.

130. The proposal in the text applies to attorneys who bill on an hourly basis. If the attorney's fee is a contingent one, then the sanctions cannot be based on the attorney's actual fee for the case, because that fee may be quite low, as a result of the other side's finding out about the nonproduced documents. In such cases, we believe that the sanctions should be based on what the attorney s fee would have been had her client been successful in the case. Where sanctions are awarded, we agree with the approach of the Fisons court that the party making the discovery motion should be reimbursed for reasonable expenses due to the nondisclosure, including reasonable attorneys fees in bringing the motion; and the remaining funds should be deposited in a court-approved fund for charity (WSPIEA v. Fisons, supra note 8, at 356).Google Scholar

131. Again, our upper limit is a conservative one. It is estimated that Fisons' fees to its attorneys in WSPIEA v. Fisons exceeded $2 million (Taylor, supra note 56)Google Scholar

132. Even in this case, it seems to us that the rational course for the attorney is not to discuss the possibility of withholding documents (and thus risk potential sanctions), but rather for the attorney simply to raise her fees. It is only if the future legal work itself depends on the withholding of the documents (e.g., if by withholding the documents, the client will be motivated to defend future lawsuits actively rather than simply settling without a fight) that it seems to us it would be rational for the attorney to suggest to her client that they withhold documents.

133. Bogle & Gates Agrees to Pay Sanction for Misconduct in Suit, supra note 123, at A6.Google Scholar

134. Taylor, , supra note 56, at 77.Google Scholar

135. Id. at 77.

136. Fed. R. Civ. P. 37(c) (1).

137. This distinction is not appreciated by Angela Lang, who writes: “The failure to make or supplement the required disclosures can result in the inability to use the information at trial, which nullifies the incentive for underdisclosure” (Mandatory Disclosure Can Improve the Discovery System, supra note 13, at 671)Google Scholar. This is true for known positive evidence, but it is certainly not true for known negative evidence.

138. Fed. R. Civ. P. 37(b) (2) (C).

139. Fed. R. Civ. P. Rule 26 (b) (3).

140. In our brief exposition of the work product doctrine, we follow Holmes, James, The Disruption of Mandatory Disclosure with the Work Product Doctrine, 73 Tex. L. Rev. 177–207 (1994)Google Scholar. See his article for a more thorough explanation of the doctrine, with citations.

141. See id. for a more general argument that the protection of intangible opinion work product conflicts with the goals of discovery itself and should not be permitted to frustrate those goals. Holmes's argument concerns the mandatory disclosure provisions of the 1993 amendments to the federal discovery rules, but his discussion equally applies to our proposal. He concludes: “Intangible opinion work product protection does not further the purposes of the adversary system; it merely disrupts the efficient information exchanges of mandatory disclosure (id. at 207).

142. See, e.g., Lasso, , supra note 11, at 511–13.Google Scholar

143. If TR attorneys are at a competitive disadvantage, as we claim, one might wonder how there could be any TR attorneys left in civil litigation. I lowever, our theory does not imply that there would be no TR attorneys, only that they would be at a competitive disadvantage with respect to FFR attorneys. It is possible to compensate for competitive disadvantages of one kind with compeuuve advantages of other kinds. Thus, our theory predicts that, because TR attorneys are at a competitive disadvantage in the current system, those who survive will be more able, on average, than the average FFR attorney.

144. Sorenson, , supra note 80, at 783.Google Scholar

145. A mendments to the Federal Rules of Civil Procedure, supra note 100.

146. A.B.A., Model Rules of Professional Conduct (1995).

147. Id. at 5.

148. Id. at 5.

149. Id. at 6 (emphasis added).

150. Id. at 66.

151. Sorenson, (supra note 80, at 782–83, n. 368)Google Scholar points out that there has been a shift away from the zealous advocacy concept in the recent Model Rules (supra note 146) as compared with the older A.B.A. Model Code of Professional Responsibility (1980). Noticeably absent trom either the Model Rules or the commentary is the equivalent of the former principle of resolving doubts as to legality in favor of the client (Model Code, id., EC 7–3). On the topic of this shift, see also Maute, Judith L., Sporting Theory of Justice: Taming Adversary Zeal With a Logical Sanctions Doctrine, 20 Conn. L. Rev. 7, 19–21 (1986).Google Scholar

152. See Freedman, Monroe, Lawyers' Ethics in an Adversary System 3–8 (1975).Google Scholar

153. Luban, , supra note 35, at 75.Google Scholar

154. See Model Rules of Professional Conduct, supra note 146, at 6.Google Scholar

155. A similar point is made by Sorenson concerning Fed. R. Civ. P. 26(a): “Rule 26(a) clearly signals the reversal of two of the central systematic tenets of traditional zealous advocacy lnthe context of discovery. When potential conflicts arise between a lawyer's ethical duty as an officer of the court to disclose facts and a lawyer's ethical duty to represent the client's interest, the duty to the court and the administration of justice rather than client's interest take primacy” (Sorenson, , supra note 80, at 789–90)Google Scholar

156. Amendments to the Federal Rules of Civil Procedure, supra note 100, at 511.Google Scholar

157. For discussion of role morality in the context of lawyers' duties, see Luban, , supra note 35, ch. 7.Google Scholar

158. Lasso, , supra note 11, at 481.Google Scholar