• 1 Introduction

  The American Rule on costs in civil litigation – which holds that each party is responsible for paying its own costs and attorney’s fees¹xAlyeska Pipeline Serv. Co. v. Wilderness Soc’y, 421 U.S. 240, 247 (1975) (‘In the United States, the prevailing litigant is ordinarily not entitled to collect a reasonable attorneys’ fee from the loser.’). – is somewhat less prevalent in the United States than its name suggests. One American state has adopted the English (or ‘loser pays’) Rule.²xAlthough large, not a populous state: Alaska. See Alaska Stat. §09.60.010 (2012). Texas, a large and populous state, permits fee shifting in a range of civil matters. See Tex. Civ. Prac. & Rem. Code Ann. §38.001 (West 2005). Cf. S. Burbank et al., ‘Private Enforcement’, 17 Lewis & Clark Law Review 650­651, at 637 (2013) (discussing the general use of the American Rule in state courts). In both federal and state courts, prevailing parties are entitled to collect some of the costs of litigation.³xSee 28 U.S.C. §1920. In addition, more than 200 federal statutes provide for fee-shifting in favour of the prevailing party.⁴xFor a comprehensive list of federal fee-shifting provisions, see H. Cohen, Awards of Attorneys’ Fees by Federal Courts and Federal Agencies (2008). A mixture of statutes,⁵xSee, e.g., 28 U.S.C. §1927. rules⁶xSee, e.g., Fed. R. Civ. P. 11(c), 16(f), 26(g), 37(a)(5), 37(b)(2)(C). and common-law doctrines⁷xSee, e.g., Chambers v. NASCO, 501 U.S. 32 (1991) (upholding sanction against a party under the court’s inherent power). also permits fee-shifting in cases of frivolous claims or litigation.
  Nonetheless, the American Rule prevails in most civil litigation most of the time. Among other things, the American Rule expects both parties to bear their own costs during the disclosure-and-discovery process.⁸xIn the American system, ‘disclosure’ and ‘discovery’ are related but distinct processes for obtaining potential evidence. Each party must disclose, without a request from another party, certain basic information. See Fed. R. Civ. P. 26(a). Discovery, on the other hand, is party initiated. See Fed. R. Civ. P. 26(b). For the sake of simplicity, I refer to both disclosure and discovery as ‘discovery’, unless in a particular context the distinction matters. The party requesting information bears the fees and cost of making discovery requests and reviewing any material that is produced; and if the responding party objects to a request, the requesting party usually bears the fees and costs of seeking a court order compelling production of the discovery.⁹xIn limited circumstances, a requesting party who successfully obtains an order to compel discovery may obtain its costs and fees from the responding party. See Fed. R. Civ. P. 37(a)(5)(A). Conversely, the responding party bears the cost of locating potentially discoverable information, reviewing it to ensure that it is subject to discovery¹⁰xIn order for information to be subject to discovery, it must be relevant, proportional to the needs of the case, not privileged and not work product. See Fed. R. Civ. P. 26(b)(1), -(b)(3). In addition, a party can often prevent the disclosure or discovery of a trade secret or at least limit the scope of the disclosure. See Fed. R. Civ. P. 26(c)(1)(G). and supplying it to the requesting party; and if the responding party believes that the information should not be provided, it usually bears its own fees and costs in replying to any motion to compel that the requesting party files.¹¹xIn limited circumstances, a responding party who successfully resists a requesting party’s motion to compel discovery may obtain from the requesting party its costs and fees. See Fed. R. Civ. P. 37(a)(5)(B). Rather than waiting for a requesting party to file a motion to compel, a responding party can file a motion for a protective order, see Fed. R. Civ. P. 26(c), but here too, with limited exceptions, it bears its own costs and fees in making the motion, see Fed. R. Civ. P. 26(c)(3).
  The allocation of fees and costs during discovery matters because discovery is a major feature of American civil litigation¹²xSee P. Carrington, ‘Politics and Civil Procedure Rulemaking: Reflections on Experience’, 60 Duke Law Journal 597, 609 (2010) (stating that ‘the discovery process … was the central, distinguishing feature of civil procedure’ in federal court); M. Rosenberg, ‘Federal Rules of Civil Procedure in Action: Assessing Their Impact’, 137 University of Pennsylvania Law Review 2203, at 2197 (1989) (‘In the run of significant lawsuits, federal discovery has helped shift the center of gravity from the trial to the pretrial stages.’). and is likely its largest cost component.¹³xSee T.E. Willging et al., ‘An Empirical Study of Discovery and Disclosure Practices under the 1993 Federal Rule Amendments’, 39 Boston College Law Review 531, at 525 (1998) (describing sample results in which, of the $13,000 spent on litigation expenses in the median case, ‘about half’ went to discovery). See also J.S. Kakalik et al., ‘Discovery Management: Further Analysis of the Civil Justice Reform Act Evaluation Data’, 39 Boston College Law Review 637, at 613 (1998) (estimating that an average of 36% of attorney time spent on a lawsuit was devoted to discovery and discovery motions; further noting that the median time was 25%). For this reason, a 2015 amendment to the Federal Rules of Civil Procedure (which govern civil cases in federal court and often influence the procedural rules used in state courts) attracted considerable attention. This amendment to Rule 26 – which is the principal Rule that sets out the parameters of American discovery – made explicit a power that federal trial judges already enjoyed implicitly: the power to shift the responding party’s discovery costs to the requesting party.¹⁴xA proposed 1998 amendment to the Federal Rules of Civil Procedure would have created a similar explicit authority for judges to shift costs for document production; see Preliminary Draft of Proposed Amendments to the Federal Rules of Civil Procedure and Evidence, 181 F.R.D. 18, 88-89 (1998), but the amendment was ultimately rejected, see Rick Marcus, ‘Introduction to Proposals for Cost-Bearing Provisions in the Rules’ (dated 6 September 2013), Tab 5D in Agenda, Advisory Comm. on Civil Rules 216-219 (7-8 November 2013). The amendment added the following italicised text to Rule 26(c)(1)(B), allowing the court to ‘specify[] terms, including time and place or the allocation of expenses, for the disclosure or discovery’. In a 2006 amendment, a similar explicit power had been given to judges to shift discovery costs for electronically stored information;¹⁵xSee Fed. R. Civ. P. 26(b)(2)(B) (‘The court may specify conditions for the discovery.’) Rule 26(b)(2)(B) states a unique rule only for electronically stored information. Although the language (‘specify conditions’) does not specifically address cost shifting, it was well understood that cost shifting was the point of this language. See Fed. R. Civ. P. 26(b)(2) advisory committee’s note to 2006 amendment (‘The conditions may also include payment by the requesting party of part or all of the reasonable costs of obtaining information from sources that are not reasonably accessible.’). and before 2015, courts had interpreted the then-extant language of Rule 26(c)(1)(B) to authorise cost shifting even without the phrase ‘or the allocation of expenses’.¹⁶xSee Fed. R. Civ. P. 26(c)(1)(B) advisory committee’s note to 2015 amendment (noting that ‘courts already exercise this authority’ to allocate expenses of disclosure and discovery); Oppenheimer Fund, Inc. v. Sanders, 437 U.S. 340, 358 (1978) (noting in dicta ‘the district court’s discretion under Rule 26(c) to grant … orders conditioning discovery on the requesting party’s payment of the costs of discovery’); Zubulake v. UBS Warburg, LLC, 216 F.R.D. 280 (S.D.N.Y. 2003) (ordering partial cost shifting for producing electronically stored information). The point of the 2015 amendment, therefore, was principally to affirm that the power existed: ‘[e]xplicit recognition will forestall the temptation some parties may feel to contest this authority’.¹⁷xFed. R. Civ. P. 26(c)(1)(B) advisory committee’s note to 2015 amendment. At the same time, it was not contemplated that this power would, or should, be used broadly: ‘Recognizing the authority does not imply that cost-shifting should become a common practice. Courts and parties should continue to assume that a responding party ordinarily bears the costs of responding’.¹⁸xIbid.
  The 2015 amendment was not the first explicit grant of cost-shifting authority in the Federal Rules of Civil Procedure. In 2006, the Supreme Court promulgated a specific rule for electronically stored information (ESI), the last sentence of which provides: ‘The court may specify conditions for the discovery.’¹⁹xFed. R. Civ. P. 26(b)(2)(B). The ‘conditions’ were understood to ‘include payment by the requesting party of part or all of the reasonable costs of obtaining information that are not reasonably accessible’.²⁰xFed. R. Civ. P. 26(b)(2)(B) advisory committee’s note to 2006 amendment. A prior provision in Rule 26(b)(2)(B) had made ESI that was not reasonably accessible, non-disclosable and non-discoverable unless the requesting party demonstrated good cause. Rule 26(b)(2)(B) did not provide a definition of ‘reasonably accessible’. For the 2004 case that influenced the structure of Rule 26(b)(2)(B) and developed the concept of ‘reasonable accessibility’, see Zubulake v. UBS Warburg LLC, 217 F.R.D. 309, 317-20 (S.D.N.Y. 2004) (developing the concept of ‘reasonable accessibility’); see also ibid., at 322 (describing the seven factors that bear on whether to permit cost shifting of ESI discovery).

• 2 Arguments for and against Shifting Discovery Costs

  American discovery is unique, for parties generally gather information with no judicial review of discovery requests or oversight of the discovery process itself.²¹xWhen initially promulgated in 1938, the Federal Rules of Civil Procedure required judicial approval of many forms of discovery. Virtually all restrictions were removed by amendments in 1946 and 1970. See Fed. R. Civ. P. 33 advisory committee’s note to 1946 amendment; Fed. R. Civ. P. 33 advisory committee’s note to 1970 amendment; Fed. R. Civ. P. 34 advisory committee’s note to 1970 amendment. Today, the only form of discovery requiring judicial approval is a request for a physical or mental examination; see Fed. R. Civ. P. 35(a)(1), although parties must also seek judicial approval for depositions, interrogatories and, in some districts, requests for admission that exceed the maximum permitted under the Rules; see Fed. R. Civ. P. 26(b)(2)(A) (authorising local rules that limit requests for production), 30(a)(2)(A)(i) (limiting depositions to ten per side), 33(a)(1) (limiting interrogatories to twenty-five per party). The court typically involves itself at the outset, working with the parties on a discovery plan and establishing deadlines for discovery to be completed.²²xSee Fed. R. Civ. P. 26(f) (describing the discovery-planning obligations of the parties), 16(b) (requiring the court to establish deadlines for discovery). It may, or may not, periodically check with the parties on the progress of discovery.²³xSee Fed. R. Civ. P. 16(a), -(c)(2)(F) (giving federal courts the authority to call pretrial conferences at which measures to control and schedule discovery). Otherwise, a court avoids intervention until a discovery dispute arises.²⁴xSee Fed. R. Civ. P. 26(c) (protective orders), 37 (motion to compel discovery). The system is also unique because the use of juries in some (though not all) civil cases means that the trial or other final adjudicatory hearing usually occurs only after the completion of all discovery.²⁵xSee Celotex Corp. v. Catrett, 477 U.S. 317, 322 (1986) (suggesting that summary judgment may be granted only ‘after an adequate time for discovery’). As a rule, federal courts establish the deadline for filing case-dispositive motions after the deadline for completing discovery. Cf. Fed. R. Civ. P. 16(b)(3)(A) (requiring that deadlines be set to ‘complete discovery’ and ‘file motions’); Smith v. OSF HealthCare Sys., 933 F.3d 859, 866 (7th Cir. 2019) (reversing summary judgment granted to the defendant before the discovery deadline when additional discovery might have aided the plaintiff to obtain relevant evidence).
  In light of these dynamics, there are excellent arguments why cost shifting of discovery expenses should be the norm in American civil litigation. Among them is the economic intuition that people tend to use a resource more wisely when they must pay for it rather than when it is free. A party that does not bear the full cost of discovery is therefore more likely to ask for too much information.²⁶xSee R.G. Bone, Civil Procedure: The Economics of Civil Procedure 230 (2003) (‘[A] cost-shifting rule deters excessive discovery by forcing a requesting party to internalize discovery costs.’). That possibility is particularly likely in cases of ‘asymmetric information’: situations in which one party possesses most of the relevant information and the other has little of value to discover. When the information in the parties’ possession is more or less in equipoise, neither party typically has an incentive to impose excessive costs on an opponent, owing to the fear that the opponent will respond in kind.²⁷xOf course, this logic does not pertain to information held in the possession of non-parties, but as a general matter, third parties hold less information, and discovery from them draws few complaints. But the logic of ‘mutual assured destruction’ does not hold when information is asymmetrically held and one side can impose significant discovery costs without consequence.²⁸xSee R. Cooter and D. Rubenfeld, ‘An Economic Model of Legal Discovery’, 23 Journal of Legal Studies 453, at 435 (arguing for a modified version of a ‘requester-pays’ rule and noting that ‘a cost-shifting rule completely eliminates impositional abuse’). See also J. Setear, ‘The Barrister and the Bomb: The Dynamics of Cooperation, Nuclear Deterrence, and Discovery Abuse’, 69 Boston University Law Review 616, at 569 (1989) (using game theory and the idea of nuclear deterrence to analyse ‘the effect that asymmetries in the availability of resources and opportunities for discovery have on the potential for discovery abuse’). A party that can inflict such costs can also often pressure (or perhaps even ‘extort’) an opponent to settle the case for an amount that does not reflect the true social value of the claim.
  A second reason for shifting discovery costs to the requesting party is to keep litigation expenses within proportionate bounds. From a social viewpoint, the cost of litigation should not exceed the expected recovery.²⁹xCf. Bone, above n. 26, at 217. (‘In economic terms, an additional investment in discovery is “excessive” whenever the social costs of the investment exceed the social benefits.’) In his 2009 report on discovery costs in the United Kingdom, Lord Justice Jackson advocated for the same principle, arguing that even costs necessary to prosecute or defend a claim can be disproportional in light of the value of the claim and other factors; he proposed that such necessary but disproportional discovery be disallowed. See Review of Civil Litigation Costs: Final Report 37 (2009) (‘If the level of costs incurred is out of proportion to the circumstances of the case, they cannot become proportionate simply because they were “necessary” in order to bring or defend the claim.’) This principle was implemented in the Civil Procedure Rules. See CPR 44.3(2)(a) (allowing only those costs assessed on the standard basis ‘which are proportionate to the matters in issue’); id. (stating that ‘[c]osts which are disproportionate in amount may be disallowed or reduced even if they were reasonably or necessarily incurred’). Cf. CPR 44.3(5) (listing factors bearing on proportionality that mostly mirror those recommended in Lord Jackson’s report). Under the American rule, it might appear that parties will make such economically rational choices regarding litigation expenditures: after all (leaving the issue of asymmetrical information aside), they bear the costs of their litigation choices, and rational economic actors will spend no more on litigation than they can expect the litigation to yield. But that intuition is wrong in numerous scenarios. For instance, the parties might act irrationally, might be risk takers or might miscalculate either the likelihood of recovery or the value of the lawsuit. Moreover, using real-options analysis, it can be shown that, in some scenarios, a rational economic actor will spend an amount on discovery that exceeds the expected return; stated more broadly, a party may have an incentive to continue spending on discovery even though, from a social viewpoint, the next dollar spent on discovery does not return at least a dollar in expected litigation value.³⁰xSee J. Grundfest and P. Huang, ‘The Unexpected Value of Litigation: A Real Options Perspective’, 58 Standford Law Review 1293-1298, at 1267 (2006) (adapting real-option theory, which is often used in deciding whether to invest in research and development, to the decision whether to invest in litigation). For applications of the concept to American civil procedure and to discovery specifically, see J. Tidmarsh, ‘The Litigation Budget’, 68 Vanderbilt Law Review 862-64, at 855 (2015); J. Tidmarsh, ‘Opting Out of Discovery’, 71 Vanderbilt Law Review 1801, 1827 n.115, 1836 (2018). Third, in some instances, a repeat player in litigation might overspend on discovery because it values the litigation differently from the opponent. For example, a manufacturer facing one thousand lawsuits concerning a defective product has an incentive to spend a great deal more in the first lawsuit than the case is worth, because winning that lawsuit will discourage future litigants.³¹xAt a practical level, this reality has led some products-liability defendants to engage in ‘scorched earth’ discovery tactics that impose such great costs on the opponent that the litigation becomes too expensive to maintain. The classic example of the use of these tactics to suppress plaintiffs’ claims is the tobacco litigation. For a short history of this litigation, see R. Rabin, ‘A Sociolegal History of the Tobacco Tort Litigation’, 44 Stanford Law Review 867, at 853 (1992) (describing how the tobacco defendants were ‘able to wear down the tobacco litigants through a seemingly inexhaustible expenditure of resources’). Asymmetrical incentives to litigate have sometimes been used as an argument to justify class actions, which aggregate the parties on both sides to create a level set of incentives. See D. Rosenberg, ‘Mandatory Litigation Class Actions: The Only Option for Mass Tort Cases’, 115 Harvard Law Review 848-53, at 831 (2002). Or a wealthy litigant may calculate that a poorly financed opponent cannot afford even economically rational litigation expenditures and may therefore engage in litigation and discovery practices that drive the opponent to abandon the case or accept a lopsided settlement.³²xSee Miller UK Ltd. v. Caterpillar, Inc., 17 F. Supp. 3d 711, 718 (N.D. Ill. 2014) (‘Where a defendant enjoys substantial economic superiority, it can, if it chooses, embark on a scorched earth policy and overwhelm its opponent. … But even where a case is not conducted with an ulterior purpose, the costs inherent in major litigation can be crippling, and a plaintiff, lacking the resources to sustain a long fight, may be forced to abandon the case or settle on distinctly disadvantageous terms.’).
  For these and other reasons – including constitutional concerns about forcing a party to pay for disclosing information detrimental to its financial well-being – some scholars have proposed the mandatory shifting of discovery costs to the requesting party.³³xSee, e.g., Cooter and Rubenfeld, above n. 28; M. Redish and C. McNamara, ‘Back to the Future: Discovery Cost Allocation and Modern Procedural Theory’, 79 George Washington Law Review 773, 822 (2011) (‘An evaluation of fundamental moral, economic, democratic, and constitutional principles reveals that our current method of discovery cost allocation rests on a theoretical foundation that is at best shaky, and at worst completely illusory.’); M. Redish, ‘Discovery Cost Allocation, Due Process, and the Constitution’s Role in Civil Litigation’, 71 Vanderbilt Law Review 1847 (2018) (arguing that shifting discovery costs may be mandated by due process). See generally E. Elliott, ‘How We Got Here: A Brief History of Requester-Pays and Other Incentive Systems to Supplement Judicial Management of Discovery’, 71 Vanderbilt Law Review 1785 (2018) (reviewing prior proposals to require cost shifting). The law of American discovery has not yet gone so far, but the 2015 amendment demonstrates that the idea has traction.
  Shifting discovery expenses to the requesting party also has drawbacks. One concern is the experience of countries that routinely shift costs: in such a system, there is often limited incentive to control costs.³⁴xFor a telling analysis of lawyers’ economic incentives to complicate and protract litigation, see A.A.S. Zuckerman, ‘Lord Woolf’s Access to Justice: Plus ça change …’, 59 Modern Law Review 775–778, at 773 (1996). This incentive is exacerbated in a loser-pays regime, because ‘a litigant who believes that an increase in the amount spent on litigation will increase his chances of success has good reason for progressively raising his stakes’. Ibid., at 779. Of course, shifting only discovery costs limits this incentive; but given how large a component of overall litigation expenditures discovery costs are, see above n. 13, the basic point holds. In the American context, a second principal concern is access to justice. In cases in which information is held symmetrically, cost shifting is usually unnecessary because costs for both sides will wash out: all that cost shifting does is to introduce a cumbersome accounting obligation.³⁵xTo avoid this obligation, it seems not unlikely that parties in symmetrical-information cases would agree to waive discovery-cost reimbursement. Cf. Abernathy v. E. Ill. R.R., 940 F.3d 982, 994 (7th Cir. 2019) (noting that each party bore the expense of its own expert in providing pretrial testimony to the opposing party, even though the parties could have sought reimbursement under Rule 26(b)(4)(E)). On the other hand, cases involving asymmetrical information often involve ‘little guys’ with little information suing large enterprises with much information. Shifting discovery costs may therefore impose an expense on a less well-financed ‘little guy’, deterring them from turning to courts to challenge the behaviour of wealthy and powerful entities.
  The Federal Rules’ approach to shifting discovery costs raises a second concern: it is ad hoc and discretionary. After conducting searches of the Westlaw federal-court database using two different sets of search terms, I collected and reviewed approximately 150 federal cases – all decided after the 2015 amendment to Rule 26(c)(1)(B) became effective – that discussed shifting costs in discovery. Many discussed the principle of cost shifting only in the abstract; other cases considered cost-shifting either as a sanction or as a form of redress for the tardy introduction of new claims. Only thirty-four cases considered the shifting of discovery expenses as a general matter. Of that number, twelve shifted discovery costs,³⁶xSee, e.g., Lawson v. Spirit AeroSystems, Inc., No. 18-1100-EFM-ADM, 2020 WL 3288058 (D. Kan. 18 June 2020); CFPB v. Ocwen Fin. Corp., Case No. 9:17-CV-80495-Marra/Matthewman, 2018 WL 6843629 (S.D. Fla. 21 December 2018). while twenty-two declined to do so.³⁷xSee, e.g., Brogan v. Fred Beans Motors of Doylestown, Inc., Civil Action No. 17-5628, 2020 WL 5017500 (E.D. Pa. 25 August 2020), Travelers Prop. Cas. Co. of Am. v. Mountaineer Gas Co., Civil Action No. 2:15-cv-07959, 2018 WL 899078 (S.D. W. Va. 15 February 2018); Oxbow Carbon & Minerals LLC v. Union Pac. R.R., 322 F.R.D. 1 (D.D.C. 2017). Even when cost shifting occurred, the court often ordered only partial sharing of expenses or else sharing of expenses only for certain components of discovery.³⁸xSee, e.g., McClurg v. Mallinckrodt, Inc., Case No. 4:12-CV-00361-AGF, 2016 WL 7178745 (E.D. Mo. 9 December 2016) (shifting 18% of defendants’ document-collection expenses to plaintiffs).
  My review was not exhaustive, and reliance on reported decisions may not accurately capture patterns in American courts.³⁹xSee D. Engstrom, ‘The Twiqbal Puzzle and Empirical Study of Civil Procedure’, 65 Stanford Law Review 1214, at 1203 (2013) (discussing the ‘serious concerns’, especially sampling bias, with using only Westlaw and Lexis databases to conduct empirical research on district-court behaviour); E. McCuskey, ‘Submerged Precedent’, 16 Nevada Law Journal 522, at 515 (2016) (‘It is well-documented that district-court opinions selected for the print reporter volumes (the Federal Supplement and Federal Rules Decisions) may not be representative of decision-making, and that therefore reliance solely on reported decisions to study judicial behavior risks biased results.’). Nonetheless, this review confirmed my impression, formed in prior research,⁴⁰xI conducted the research principally in 2016 and in 2019, while reviewing hundreds of recent decisions to update discovery chapters in a co-authored casebook. that federal courts typically decline to use their cost-shifting powers even after the 2015 amendment’s encouragement to do so; moreover, courts that decline to shift costs frequently invoke the American rule as one reason for their decision.⁴¹xSee, e.g., Oxbow, 322 F.R.D. at 11 (‘Oxbow has failed to rebut the presumption imposed by the Federal Rules of Civil Procedure that it should bear the cost of complying with Defendants’ proposed discovery.’). Such an ad hoc system can add to cost and delay, as parties litigate the satellite issue of cost-sharing, and can lead to injustice, as similarly situated parties receive different treatment in different courtrooms.⁴²xThere are 677 federal district-court judges in the United States, as well as another 588 federal magistrate judges who are often tasked by district-court judges to resolve discovery matters. See ‘Judicial Business 2020, U.S. District Courts’ tbl.5 (Admin. Office of U.S. Courts) www.uscourts.gov/statistics-reports/us-district-courts-judicial-business-2020 (last visited 21 June 2022); ‘Judicial Business 2019, Status of Magistrate Judge Positions and Appointments’ tbl.13 (Admin. Office of U.S. Courts), https://www.uscourts.gov/statistics-reports/status-magistrate-judge-positions-and-appointments-judicial-business-2021 (last visited 21 June 2022). Should a magistrate judge’s decision on a discovery matter be appealed to the district judge, the judge reviews the decision on a deferential basis. See 28 U.S.C. §636(b)(1)(A) (permitting reconsideration of a magistrate judge’s order only when the order ‘is clearly erroneous or contrary to law’). Likewise, great deference is given to the discovery decisions of district-court judges on appeal. See Khoury v. Miami–Dade Cty. Sch. Bd., 4 F.4th 1118, 1125 (11th Cir. 2021) (‘We review the District Court’s ruling on discovery matters only for abuse of discretion.’). With such a diffusion of authority, it is a near certainty that different judges will make different decisions on similar facts.
  A third difficulty with cost shifting in the American context is that it is far from clear how significant a problem the issue of ‘impositional’ or otherwise excessive discovery is in practice.⁴³xIn a third or more of federal cases, no discovery occurs, and discovery in most other cases is not burdensome. See Kakalik et al., above n. 13, 636, at 613 (1998) (reporting that no discovery occurred in 38% of cases in a sample and that ‘[d]iscovery is not a pervasive litigation cost problem for the majority of cases’); Willging et al., above n. 13, at 530-31 (reporting that, in a sample of cases likely to involve discovery, only 85% of attorneys reported that discovery occurred and that the median cost of discovery was $13,000 per client, or about 3% of the stakes in the case). Data from state courts show that discovery costs constitute only a small fraction of the total recovery from litigation. E. Helland et al., ‘Contingent Fee Litigation in New York City’, 70 Vanderbilt Law Review 1971, 1988 (2017) (describing a random sample of New York tort cases in which the median for all expenses, other than attorney’s fees, was 3% of recoveries and the average was 5%). In the main, lawyers are satisfied with the discovery process, believing that it reveals the right amount of information at a fair price. See Kakalik et al., above n. 13, at 636 (‘Subjective information from our interviews with lawyers also suggests that the median or typical case is not “the problem.”|’); E.G. Lee III and T.E. Willging, National, Case-Based Civil Rules Survey 27, 34 tbl.4, 37 tbl.5, 43 tbl.10 (2009) (reporting survey results in which a clear majority of both plaintiffs’ and defence lawyers believed that discovery revealed ‘just the right amount’ of information, the median costs for discovery by plaintiffs’ lawyers was $15,000 and by defence lawyers $20,000, and the costs of discovery in relation to the stakes of the litigation were 1.6% for plaintiffs and 3.3% for defendants). Contra, Interim Report on the Joint Project of the American College of Trial Lawyers Task Force on Discovery and the Institute for the Advancement of the American Legal System 3-5 (2008) (describing survey results that ‘confirm[] that there are serious problems in many parts of the civil justice system, especially the rules governing discovery’). The prospect (some would say myth) of parties inflicting needless discovery costs on opponents to gain tactical advantage has driven much of the procedural reform in the United States in the past forty years – from expanded case management to limits on the scope of discovery to more stringent requirements for pleading.⁴⁴xFor a fuller review of these reforms, see Tidmarsh (2018), above n. 30, at 1813-1814. The 2015 amendment that permits the shifting of discovery expenses sits at the end of a long line of unsuccessful (or at best partially successful) efforts to contain litigation costs. But the available data – as opposed to lawyers’ anecdotes – has never clearly supported the need to do anything at all about discovery costs.
  That concern leads to a final difficulty with cost shifting: its tunnel vision. Cost shifting is often proposed as a stand-alone solution to the problem of expensive discovery. But other responses are possible. For instance, effective case management might make cost shifting unnecessary, and American procedure has been strongly committed to case management since 1983.⁴⁵xSee Fed. R. Civ. P. 16. If we adopt cost shifting, should case management be relaxed or even abandoned? Discovery is just one part – albeit a critical part – of the machinery of civil justice; and as with any complex system, tinkering with one feature (cost allocation) of this one part will have repercussions elsewhere. Adopting an incentives-based approach to controlling discovery, such as cost shifting, may make more direct controls over the discovery process, such as case management, superfluous or even overkill; it may skew outcomes too much in favour of those who possess information and too much against those who do not. I am not trying to resolve the age-old debate about incentives versus command-and-control rules,⁴⁶xCf. E. Elliott, ‘Managerial Judging and the Evolution of Procedure’, 53 University of Chicago Law Review 326-34, at 306 (1986) (arguing that incentives-based approaches are more efficient than command-and-control procedural rules). but instead to highlight that an incentives-based approach like shifting discovery costs does not exist in a vacuum.
  It is important to put arguments for cost shifting in discovery in proper context. In countries that employ loser-pays fee shifting, the concern over shifting discovery costs plays out differently: some or all of the expenses of both sides’ discovery may be borne by the losing party in any event. In the American-rule context, however, shifting discovery costs may benefit the losing party as readily as the winning party because the shifting of costs is not tied to prevailing in the litigation. We can debate the merits of the American rule that each side bears its own costs, but as long as the rule prevails, there must be a compelling justification to depart from it just for one aspect of litigation (discovery) where cost shifting is not even keyed to winning or losing.
  In view of the American rule, a general, mandatory shifting of discovery costs to the requesting party faces a steep climb in American litigation. If stand-alone mandatory cost shifting can, in fact, deliver on its promise to hold down the costs of discovery to a reasonable level, if the savings from stand-alone mandatory cost shifting exceed the detrimental side effects (such as less access to justice), and if stand-alone mandatory cost shifting does so in a more effective way than alternatives, the necessary justification to require cost shifting would exist. But these are three big ‘ifs’. Let me tackle them briefly.

• 3 The Hidden Economic Costs of Cost Shifting

  The first and second assumptions are related: together they assume that the savings from mandatory cost shifting will outweigh the costs. Because these assumptions are essentially economic in nature, I approach the question from that perspective. From an economic point of view, litigation generates two costs: the direct cost of litigation (attorneys’ fees, expert witnesses, discovery, judicial time, and so on) and error costs.⁴⁷xSee R.A. Posner, Economic Analysis of Law §21.1 (7th ed. 2007). Error costs are often neglected, but they are real: if a court fails to resolve disputes accurately, the negative social effects could be substantial. For instance, we can devise a very cheap process for resolving disputes – such as flipping a coin – but the errors that would result would be socially disastrous: why would anyone engage in productive activity that amasses wealth if I can take that wealth away with a ginned-up claim and a lucky flip of the coin?
  As a general matter, the two costs of litigation are inversely related (at least if our procedural rules are rationally designed to ferret out the truth): the more that we spend directly on litigating a dispute, the fewer the errors; while the less we spend, the more the errors. Procedural rules should minimise the sum of direct litigation costs and error costs.⁴⁸xThis statement is not precisely accurate, and I will refine it shortly. See below n. 53 and accompanying text. As an initial matter, however, procedural scholars usually proceed from the assumption that keeping the sum of direct litigation costs and error costs to their minimum is the best way to keep overall social costs at their minimum. Cost shifting in discovery is designed to give the parties an incentive to lower one direct litigation cost: the cost of discovery. It assumes that, in a cost-shifting regime, parties will limit their discovery expenditures to a cost-justified level, where another dollar spent on discovery will not yield at least another dollar in expected economic benefit from the settlement or judgment.
  Like many incentives-based approaches, the concept of cost shifting also assumes that the parties have full information with which to make the requisite marginal-cost calculation. In the real world, however, parties do not know the value of the information that they might obtain in discovery, so they cannot know whether another dollar spent on discovery is worthwhile until it has been spent. In such a world, whether mandatory cost shifting will hold discovery costs to a cost-justified level – especially in instances of asymmetrical information that form the central case for cost shifting – is unclear. Uncertain of the value of the unknown, parties may engage in too little discovery, and cases will be settled or adjudicated inaccurately owing to insufficient information – a result that constitutes an error cost. Such inaccuracy is an error cost that may offset, or at least cut into, the savings in direct litigation costs that cost shifting promises. Or perhaps parties will engage in too much discovery – a result that constitutes an excessive direct litigation cost.
  Although proponents of cost shifting might want you to believe otherwise, the question is not as simplistic as whether cost shifting reduces direct litigation costs. Without knowing the value of the information that discovery might yield, it is impossible for the parties in a cost-shifting world to know whether the amount of discovery for which they must now pay has landed near the marginal-cost sweet spot. Over time lawyers may develop sufficient experience or algorithms that dictate whether asking for some types of discovery is (or is not) ‘worth it’, but that result remains speculative at present.⁴⁹xOf course, the same criticism can be levelled against a system of discovery in which costs are not shifted. But, as I have described, there is very little data suggesting that the cost of discovery in the present non-cost-shifting world is problematic. See above n. 43-44 and accompanying text. The burden of proof for making a substantial change lies with the proponents of cost shifting.
  Of course, it is not even clear how much mandatory cost shifting will reduce discovery expenses. More likely, shifting discovery costs will accelerate a trend that has emerged in other countries but that has been slower to develop in the United States: third-party funding. In asymmetrical-information cases, which often pit Davids against Goliaths, the ‘little guy’ may be unable to afford the full burden of paying for the discovery requested. The party’s lawyer may be able to front the costs, or the lawyer may need to find someone else willing to do so. Third-party funders are the likeliest source. But third-party funding is not viewed as an unadulterated good in the United States. Although often marketed as a way to expand access to justice by providing the resources necessary to conduct litigation effectively, third-party funding has raised concerns about how it skews resources to high-dollar-value claims. It may also generate conflicts of interest between funders and plaintiffs. In particular, the fear is that funders, desirous of achieving a higher rate of return, may pressure plaintiffs to settle claims quickly and for an inadequate amount. This result generates a type of error cost.⁵⁰xFor an overview of third-party litigation funding in the United States and some of its concerns, see M. Shapiro, ‘Distributing Civil Justice’, 109 Georgetown Law Journal 1509-1512, at 1473 (2021). It is fair to say that judges and legislatures have so far treated the arrival of third-party funding with a fair amount of scepticism. Ibid., at 1510. For a more optimistic appraisal of third-party litigation financing, see S. Bedi and W. Marra, ‘The Shadows of Litigation Finance’, 74 Vanderbilt Law Review 563 (2021) (arguing that the prospect of litigation funding positively affects parties’ pre-litigation behaviour, especially in contractual settings, and also has positive post-litigation effects).
  Concerns over third-party funding have led both lawmakers and rule-makers to consider changes that would require disclosure of the identity of third-party funders and any funding agreements. See, e.g., Litigation Funding Transparency Act of 2021, S. 840, 117th Cong. (2021) (requiring disclosure of funding sources in class action and multidistrict litigation); Agenda, Advisory Committee on Civil Rules 371-86 (5 October 2021) (describing history of efforts to regulate third-party funding by rule or legislation and recommending against immediate action on the issue). No proposals to regulate third-party funding have yet been enacted at the federal level; and in any event, disclosure obligations would likely have little effect on the issues described in the text.
  To the extent that third-party funding, with its attendant concerns, is unavailable, shifting discovery costs seems likely to suppress litigation in David-versus-Goliath scenarios. ‘Access to justice’ is an abstract notion, and it often seems to be an ‘apples to oranges’ response to the economic intuitions that underlie cost shifting. But ‘access to justice’ can be translated into economic terms as a type of error cost. Typically, when we think about error costs, we think about the errors in the case being adjudicated: the reason that a coin flip is problematic is that the likelihood of getting the wrong result in the specific case is high. But error also exists when cases with merit are not filed at all. Granted, the errors are hidden from view in a way that errors in litigated cases are not. But from a social perspective, the failure to bring a meritorious case is as much a cost as the failure accurately to resolve a litigated case.
  Furthermore, leaving the economic analysis to the side, the ‘day in court’ ideal is a treasured foundation for American justice. As Mauro Cappelletti and Bryant Garth have recognised, ‘access to justice’ is ‘not easily defined’, but it embodies

  two basic purposes of the legal system – the system by which people may vindicate their rights and/or resolve their disputes under the general auspices of the state. First, the system must be equally accessible to all; second, it must lead to results that are individually and socially just.⁵¹xM. Cappelletti and B. Garth, ‘Access to Justice: The Newest Wave in the Worldwide Movement to Make Rights Effective’, 27 Buffalo Law Review 181, 182 (1978); see also ibid., at 185 (‘Effective access to justice can thus be seen as the most basic requirement – the most basic “human right” – of a modern, egalitarian legal system which purports to guarantee, and not merely proclaim, the legal rights of all.’).

  In the United States, the openness of courts to hear disputes from all citizens without distinction or favour, and thus to level the playing field between the powerful and the ordinary, is a pillar of the American sense of equality before law. Indeed, the principal argument that undergirds the American rule on costs is that the rule promotes access to American courts.⁵²xSee E. Labaton, ‘Courts on Trial Symposium Closing Remarks’, 40 Arizona Law Review 1113, at 1111 (1998) (‘Access to American courts is available for two reasons: the contingent fee and the American rule.’); see also A. Goodhart, ‘Costs’, 38 Yale Law Journal 874, at 849 (1929) (‘Apart from purely historical reasons , the American rules as to costs may also be due in part to a vague feeling that they favor the poor man, and are therefore democratic, while the English system helps the wealthy litigant.’); cf. ibid., at 877 (noting that allowing a judge ‘wide discretion [to tax costs] as is inherent in the English system would be contrary to the general American conception of a judiciary bound by fixed rules’). The damage to that ideal could also be seen as a cost.
  Another potential cost of cost shifting is its effect on litigants’ primary behaviour (i.e. how they act before the claim arises). From a social-welfare viewpoint, the goal of a legal system is simply not to minimise the sum of direct litigation costs and error costs. Rather, the law should minimise the sum of three costs: the cost of harm, the cost of preventing harm and transaction costs, which include direct litigation costs and error costs.⁵³xSee G. Calabresi, The Cost of Accidents (1970), at 26-31. This overarching cost-minimisation goal does not require that the sum of direct litigation and error costs be kept to their minimum. Indeed, the fear of incurring large discovery costs may induce actors to spend more to prevent harm, and the resulting reduction in the cost of harm may exceed the cost of discovery or errors.⁵⁴xFor example, assume that, with precautions costing $400, the defendant’s conduct causes $800 in harm to eight plaintiffs (or $100 apiece); and the defendant’s discovery costs in ensuing litigation with the plaintiffs are $240 (or $30 apiece). The defendant could spend $510 in safety precautions, and only seven plaintiffs would suffer harm (for a total of $700). Discovery costs would be reduced to $210. From a social viewpoint, it would be better if the defendant took the extra $110 in safety precautions, since the total cost is $1,420, compared with total costs for the lower-precaution alternative of $1,440. If the defendant bears the cost of discovery, it will take the extra precaution, because an expenditure of $110 saves the defendant $130 in liability and litigation costs. But if the defendant does not bear the cost of discovery, the defendant will not take the extra precaution, for it would be spending an additional $110 and receiving a benefit of only $100. Obviously, this example is stylised, and its result does not pertain across all permutations. There are more realistic examples, which factor into the likelihood of suit, that better prove the point, but they would take more time to unpack than is justified for present purposes. The point of the given example is only to illustrate that there are circumstances in which the potential exposure to litigation costs can induce a party to undertake a socially appropriate level of care that it might not undertake if it did not bear its own discovery costs. The party has less incentive to reduce harm, however, if it knows that other parties bear its discovery costs.
  I am not suggesting that mandatory shifting of discovery costs raises costs without corresponding benefits. The potential of cost shifting to limit impositional discovery and to avoid skewing settlements in favour of requesting parties in asymmetrical-discovery litigation is perhaps its strongest feature. My point is to identify potentially negative consequences of cost shifting that may offset or substantially reduce its cost-saving potential – concerns that proponents of cost shifting sometimes ignore or minimise. Of course, my critiques, as well as arguments favouring cost shifting, must ultimately rise or fall on the basis of the evidence. So far, no empirical studies of the effects of cost shifting on the proportional disclosure of information have been conducted in the United States.⁵⁵xCongress or the federal judiciary has sometimes developed pilot programmes in some federal district courts to generate data on the efficacy of proposed procedural reforms. See 28 U.S.C. §651(b) (requiring each federal district court to authorise the use of one or more alternative dispute resolution processes); J. Sutton and D. Webb, ‘Bold and Persistent Reform: The 2015 Amendments to the Federal Rules of Civil Procedure and the 2017 Pilot Projects’, 101 Judicature 12 (Autumn 2017) (touting the benefits of pilot projects on discovery reform). Piloting programmes and gathering data on their successes and failures seem appropriate before undertaking major reformation of a procedure as foundational to the American system as discovery. The experience of countries that generally shift costs to the winning party can be examined for relevant clues, although differences in the extent of discovery and the pesky American rule would make it difficult to draw firm conclusions.

• 4 Alternatives to Mandatory Cost Shifting

  Before it can be adopted, mandatory cost shifting as a stand-alone remedy for excessive discovery must also prove itself better than other alternatives that address the same problem. Other options exist. Let me highlight two: one is a command-and-control approach and the other an incentive-based approach. As I discuss, both ideas have difficulties but both strike me as more promising than mandatory cost shifting.
  Recent developments in English procedure suggest the command-and-control alternative. Courts could require parties to set, and then live within, a budget for the litigation. The budget includes an allotment for discovery expenses, which must not be exceeded. English courts adopted the concept of costs budgets, which applies principally to ‘multi-track’ cases,⁵⁶xSee CPR 3.12-3.18. There are exceptions; multi-track cases exceeding £10 million are exempt from costs budgeting, see CPR 3.12(1)(a)-(b), as are cases involving minors, see CPR 3.12(1)(d), and those that are subject to ‘fixed costs or scale costs’, see CPR 3.12(1)(d). The court can also exempt a multi-track case from costs budgeting by order. CPR 3.12(1)(e). Conversely, costs budgeting may also apply in other cases when the court so orders. See CPR 3.12(1A).
  The Civil Procedure Rules divide cases into three tracks: ‘multi-track cases’, ‘small claims cases’, and ‘fast track cases’. ‘Multi-track cases’ are defined as cases that are neither ‘small claims cases’ nor ‘fast-track cases’. See CPR 26.6(6) (‘The multi-track is the normal track for any claim for which the small claims track or the fast track is not the normal track.’). In general, ‘small claims’ cases are those personal-injury cases whose value does not exceed £10,000 (with some additional exceptions), certain landlord-tenant disputes with a value less than £1,000, and other cases whose value does not exceed £10,000. CPR 26.6(1)-(3). In general, ‘fast-track’ cases are those cases whose value does not exceed £25,000, the trial will likely last no more than one day and expert witnesses are limited. See CPR 26.6(4)-(5). Other factors, such as the number of likely parties, the complexity of the issues, and the importance of the case to non-parties, can influence the court’s decision in assigning a case to a given track. See CPR 26.8. in 2013. Insofar as limiting discovery expenses is concerned, a functionally equivalent idea would be a judicially imposed cap on discovery expenditures.⁵⁷xA costs budget is a broader concept because it would also impose caps on other pretrial and trial expenditures. England has also adopted a form of this ‘costs capping’. A party can apply for an order that limits costs from the date of the order.⁵⁸xOriginally set out in Rule 44 in 2009, costs capping was moved to CPR 3.19-21 in 2013, at the time that costs budgeting was installed in the Civil Procedure Rules. In theory available in all multi-track and other cases,⁵⁹xCosts capping is unlikely to arise in small-claims-track or fast-track litigation; in small-claims cases, each party bears most of its own costs, see CPR 27.14, while in fast-track cases, fixed costs are the norm, see CPR 45.37-40. Thus, the main terrain for costs capping is the multi-track case, including those multi-track cases ordinarily exempt from costs budgeting. See above n. 56. a costs-capping order is in practice entered ‘only in exceptional circumstances’.⁶⁰xPractice Direction 3F ¶1.1. Leaving aside the details of when and how English courts control costs, the basic intuition is spot on: if the concern is the disproportionality of discovery costs, the best way to control those costs is to limit the ability of the parties to spend money on discovery.
  There are, however, practical problems with costs budgets and costs capping as a means to limit discovery costs. The first is the court’s ability to know where to set the cap: the court might have a reasonable sense of the cost of discovery, but then the ever-present bogie man of all efforts to limit discovery to proportional levels rears its head: the judge, like the parties, is unlikely to have accurate information on either the value of the case or the extent to which discoverable information will affect that value. These are critical variables in trying to limit expenditures to a proportional amount.⁶¹xSee, e.g., Crystal Lakes v. Bath & Body Works, LLC, No. 2:16-CV-2989-MCE-GGH, 2018 WL 533915, at *2 n.1 (E.D. Cal. 23 January 2018) (noting that, without further clarification about the scope of the plaintiff’s claims, ‘it is difficult to determine the proportionality issue on his discovery demands’). See also 8 C.A. Wright et al., Federal Practice and Procedure §2008.1 (2010) (stating that ‘the proportionality concept … seemed to require great familiarity with the case’, thus somewhat frustrating its implementation). As a result, courts may fall back on possibly erroneous assumptions about the merits (or demerits) of broad discovery and the social value (or lack of value) of particular cases. At a minimum, the case-specific nature of the inquiry is likely to result in discretionary judgments that will result in like cases being treated unalike in different courtrooms across the United States.
  The second problem is establishing a mechanism to ensure that parties remain within their budgets. In England, where the loser-pays rule allows the winning party to collect its costs from the losing party, the incentive to remain within the budget or cap is created by limiting the winning party’s award of costs to those contained in the budget or cap; in other words, the winning party may bear any costs spent in excess of the budget or cap.⁶²xCPR 3.18(b) (noting that a court, when assessing costs on the standard basis, will ‘not depart from such approved or agreed budgeted costs unless satisfied that there is good reason to do so’). That incentive is far from perfect: winning parties can spend in excess of the budget or cap, and they have an economic reason to do so when they can increase their expected recovery by more than the amount of the excessive expenditure.⁶³xSee J. Tidmarsh, ‘Realising the Promise of Costs Budgets: An Economic Analysis’, 35 Civil Justice Q 219 (2016). But even this limited incentive to keep costs within bounds does not work in the United States, where each side bears its own costs. Courts possess no award-of-costs incentive to control the parties’ expenditures. Other incentives are possible. Judges can require lawyers to advise their clients that they need not pay their lawyer more than the budgeted fee; but, equivalent to the problem in England, this advice will not constrain a party who believes that spending another dollar on discovery will favourably change the expected recovery by more than a dollar. Perhaps a better control would be incentives based: require a party who exceeds the budgeted amount to indemnify other parties who must respond to excessive discovery.⁶⁴xSee Tidmarsh (2018), above n. 30, at 888-89. This incentive is, of course, a form of cost shifting, but it is targeted to control behaviour that is, by definition, impositional (because the party’s behaviour imposes costs in excess of the judicial cap).
  In short, capping costs through an order or a budget is an excellent alternative in theory. The practical difficulties of setting the budget and enforcing the obligation to stay on budget, however, make the direct control of costs more problematic.⁶⁵xOther practical, legal and constitutional issues would also need to be addressed before a realistic costs-capping measure could be implemented. See ibid., at 901-17. But these problems are akin to and not significantly worse than the problems of mandatory cost shifting.
  Barring the ready ability to control costs directly, a second alternative is to construct a procedural system that reduces reliance on the discovery process. The most radical approach would ban discovery entirely, thus reverting to common-law procedure of two centuries ago.⁶⁶xI know of no scholar whose proposals go quite so far, but a few American writers have advocated the abolition of nearly all discovery. See G. Shepard, ‘Failed Experiment: Twombly, Iqbal, and Why Broad Discovery Should be Further Eliminated’, 49 Indiana Law Review 466, at 465 (2016) (‘Broad discovery should be eliminated. It is a seventy-year experiment that has failed. The rest of the world recognizes this …’). As disputes have become more complex and accuracy more critical to their resolution, return to a system that even nineteenth-century lawyers recognised as too draconian is likely to lead to disaster as information necessary to decide cases collapses and errors in judgment rise: there is a reason that the common law abandoned its ban on discovery.⁶⁷xSee Common Law Procedure Act 1854, 17 & 18 Vict. c. 125 (Eng.) (authorizing common-law courts to conduct discovery); K. Funk, ‘Equity Without Chancery: The Fusion of Law and Equity in the Field Code of Civil Procedure, New York 1846-1876’, 36 Journal of Legal History 152, 174-79 (2015) (describing the breadth of discovery in New York before and after the adoption of the Field Code in 1848); E. Sunderland, ‘Scope and Method of Discovery Before Trial’, 42 Yale Law Journal 869-70, at 863 (1933) (describing the scope of discovery in American and Canadian jurisdictions prior to the Federal Rules of Civil Procedure). Less radically, judges could be placed in charge of the discovery process, as they are in numerous other countries;⁶⁸xSee, e.g., J. Langbein, ‘The German Advantage in Civil Procedure’, 52 University of Chicago Law Review 827-29, at 823 (1985) (discussing the German approach to fact-gathering by the judge). but a lack of judicial resources⁶⁹xAs mentioned, there are 677 authorised federal trial-level judges; and at any given time, a number of those positions remain unfilled. See above note 43. As of September, 2020, the number of cases pending in federal court was 511,666. See ‘Federal Judicial Caseload Statistics 2020’ (Admin. Office of U.S. Courts), www.uscourts.gov/statistics-reports/federal-judicial-caseload-statistics-2020 (last visited 1 October 2021). That number works out to more than 755 cases per authorised judgeship. and an American legal culture wedded to a more adversarial process make this approach unimaginable in practice. Nor does expanding existing requirements to disclose information to parties on a mandatory basis⁷⁰xSee Fed. R. Civ. P. 26(a) (requiring the mandatory disclosure of information supporting a party’s claims or defences, expert witnesses and witnesses and documents to be used at trial). solve the problem: no system of rules can fully encompass the parties’ disclosure obligations, and, in any event, disputes over the parties’ right to information and other parties’ obligation to provide it would simply shift from discovery to disclosure.⁷¹xFor a well-crafted plan to expand the parties’ mandatory-disclosure obligations, while still permitting some modest follow-on discovery, see D. Rosenberg et al., ‘A Plan for Reforming Federal Pleading, Discovery, and Pretrial Merits Review’, 71 Vanderbilt Law Review 2059 (2018).
  A third approach is to build a system of incentives to induce parties to reduce their reliance on discovery. Of course, mandatory cost shifting as a stand-alone measure is one example of a system of incentives. But other systems may be better, and I have proposed a different set of incentives: carrots for parties who waive discovery rights and sticks for those who do not.⁷²xSee Tidmarsh (2018), above n. 30.
  The principal carrot is to exempt parties who choose not to engage in discovery from two pretrial motions – the motion to dismiss and the motion for summary judgment that figure prominently in American procedure. The motion to dismiss accepts the plaintiff’s factual allegations in the complaint (the document that commences an American lawsuit) and asks for dismissal when the plaintiff has no plausible claim to relief.⁷³xSee Fed. R. Civ. P. 12(b)(6); Bell Atl. Corp. v. Twombly, 550 U.S. 544, 557-58 (2007) (imposing a plausibility pleading standard); see also Ashcroft v. Iqbal, 556 U.S. 662, 678-79 (2009) (clarifying the scope of the plausibility standard). The motion for summary judgment seeks judgment when the material obtained in discovery and other admissible evidence shows that no genuine dispute exists regarding the material facts so that judgment is appropriate.⁷⁴xSee Fed. R. Civ. P. 56(a); Anderson v. Liberty Lobby, Inc., 477 U.S. 242 (1986) (discussing the summary-judgment standard); Celotex Corp. v. Catrett, 477 U.S. 317 (1986) (same). The effect of removing these motions from the picture is to bring a case to trial immediately, a result that some parties may value more highly than the right to conduct discovery.⁷⁵xA second carrot, when both parties waive discovery, is to waive jury trial in cases that would be triable to a jury when both parties opt to conduct discovery. Admittedly, some parties may regard the waiver of jury-trial rights as a stick rather than a carrot to forgo discovery, but nothing turns on whether it is characterised as a carrot or a stick. Eliminating jury trial would allow a court to adopt certain case-management tools that could make an American trial run more along the lines of civil-law trials and thus avoid needless discovery on issues that the court never reaches. See Langbein, above n. 68, at 830 (describing how ‘[t]he implications for procedural economy are large’ under the episodic nature of German trial),
  The principal stick in my proposal is to shift discovery costs (with some exceptions) to a party that opts to engage in discovery when the opposing party opts out of discovery. In particular, courts should retain some discretion not to shift costs when doing so would seriously disadvantage the ‘little guy’ pursuing litigation against a major institution.⁷⁶xThe types of factors that might inform a court’s discretion not to shift costs would be essentially those given in Lord Jackson’s 2009 report and presently found in Rule 44.3(5) of the Civil Procedure Rules (including the amount in controversy, the value of non-monetary relief, the complexity of the case, and broader considerations about the case’s public importance), as well as the essentially overlapping factors identified as relevant to the proportionality inquiry under Federal Rules of Civil Procedure 26(b)(1) (including the amount in controversy, the importance of the issues at stake in the case, the parties’ resources and their relative access to information). Of course, there are evident objections to such discretion, including, as always, the potential for disparate treatment of litigants and satellite litigation over the need to shift costs. Given that discretionary cost shifting is currently the norm and that enough law about cost shifting has emerged to keep discretion within reasonable boundaries,⁷⁷xSee above n. 39-42 and accompanying text. the cost of operating a system of discretionary cost shifting in the ‘little guy’ situation is not much greater than under present law. Indeed, if the primary goal of the proposal is met, some parties will opt out of discovery, reducing the number of instances in which cost shifting might arise and thus also reducing the number of cases in which disparate treatment might arise.
  An opt-out proposal aligns courts much more with arbitration processes that have eroded the important role of public adjudication in the resolution of social disputes.⁷⁸xFor the classic opposition to the wide-scale resolution of disputes through private ordering systems, see O. Fiss, ‘Against Settlement’, 93 Yale Law Journal 1073 (1984). One of arbitration’s most cited advantages is that the parties do not incur discovery expenses.⁷⁹xSee Gilmer v. Interstate/Johnson Lane Corp., 500 U.S. 20, 31 (1991) (‘Although [discovery] procedures might not be as extensive [in arbitration] as in the federal courts, by agreeing to arbitrate, a party trades the procedures and opportunity for review of the courtroom for the simplicity, informality, and expedition of arbitration.’ (internal quotation marks omitted)); Dodson Int’l Parts, Inc. v. Williams Int’l Co., 12 F.4th 1212, 1230 (10th Cir. 2021) (‘[O]ne cannot expect full discovery in arbitration proceedings, as extensive discovery could undermine much of the advantage of arbitration.’). By offering an arbitration-like alternative, an opt-out system could reclaim for the courts a more central role in resolving disputes.
  It also bears emphasis that the point of this system is to eliminate discovery in some cases entirely. Assume that ten cases would each have $12,000 in discovery costs and that mandatory cost shifting would reduce the per-case spending on discovery by $1,000. Under a system that allows parties to opt out of discovery, further assume that the parties do so in one case. From a social-welfare viewpoint, even if the judge does not shift discovery costs in any of the other nine cases, the system is money ahead: mandatory cost shifting would result in a $10,000 reduction ($120,000 in total down to $110,000) in discovery costs, while opting out of discovery would result in a $12,000 reduction.⁸⁰xIn this hypothetical case, if cost shifting occurred in just a third of the cases that did not opt out of discovery, the savings would be even larger: the reduction in discovery costs would now rise to $15,000 ($12,000 in the case that opted out of discovery plus $1,000 apiece in the cases in which discovery costs shifted). The broad point is that it can be more socially beneficial to adopt structural changes that eliminate discovery in some cases rather than introduce changes that seek to reduce discovery in each case.

• 5 Conclusion

  My suggestion that cost shifting could be adopted as part of an opt-out proposal might raise your eyebrows, given the rather sceptical attitude that I previously evinced about cost shifting. Thus, two final points are in order. First, in an opt-out system, shifting costs is not a stand-alone proposal. In this short article I cannot pursue the proof of this point down all the rabbit holes of American procedure, but the incentives such as those that I develop integrate discovery into larger themes in the American civil-justice system, such as disappearing trials, the expansion of case management, the rise of settlement and alternative forms of dispute resolution, and growing pretrial motion practice. Shifting discovery costs works better in connection with a broader set of proposals that considers the issue of discovery and its costs in light of the entire civil-justice system and its needs.

  Second, such incentives-based proposals should be understood as a second-best solution. Cost budgeting (or cost capping) strikes me as a preferable option if solutions to the difficulties of its practical application can be found. Moreover, the ultimate proof of any alternative – whether mandatory shifting of discovery costs, cost budgeting or a proposal to reduce reliance on discovery – lies on the ground. There is a need to experiment with multiple ideas, to gather the data from each experiment and then to chart a course forward. The relentless nature of procedural reform demands no less.

Noten

• * I thank two anonymous referees for their helpful suggestions, and I am also grateful to Masood Ahmed for sources that he helpfully supplied and for numerous exchanges on the issue of discovery.

• 1 Alyeska Pipeline Serv. Co. v. Wilderness Soc’y, 421 U.S. 240, 247 (1975) (‘In the United States, the prevailing litigant is ordinarily not entitled to collect a reasonable attorneys’ fee from the loser.’).

• 2 Although large, not a populous state: Alaska. See Alaska Stat. §09.60.010 (2012). Texas, a large and populous state, permits fee shifting in a range of civil matters. See Tex. Civ. Prac. & Rem. Code Ann. §38.001 (West 2005). Cf. S. Burbank et al., ‘Private Enforcement’, 17 Lewis & Clark Law Review 650­651, at 637 (2013) (discussing the general use of the American Rule in state courts).

• 3 See 28 U.S.C. §1920.

• 4 For a comprehensive list of federal fee-shifting provisions, see H. Cohen, Awards of Attorneys’ Fees by Federal Courts and Federal Agencies (2008).

• 5 See, e.g., 28 U.S.C. §1927.

• 6 See, e.g., Fed. R. Civ. P. 11(c), 16(f), 26(g), 37(a)(5), 37(b)(2)(C).

• 7 See, e.g., Chambers v. NASCO, 501 U.S. 32 (1991) (upholding sanction against a party under the court’s inherent power).

• 8 In the American system, ‘disclosure’ and ‘discovery’ are related but distinct processes for obtaining potential evidence. Each party must disclose, without a request from another party, certain basic information. See Fed. R. Civ. P. 26(a). Discovery, on the other hand, is party initiated. See Fed. R. Civ. P. 26(b). For the sake of simplicity, I refer to both disclosure and discovery as ‘discovery’, unless in a particular context the distinction matters.

• 9 In limited circumstances, a requesting party who successfully obtains an order to compel discovery may obtain its costs and fees from the responding party. See Fed. R. Civ. P. 37(a)(5)(A).

• 10 In order for information to be subject to discovery, it must be relevant, proportional to the needs of the case, not privileged and not work product. See Fed. R. Civ. P. 26(b)(1), -(b)(3). In addition, a party can often prevent the disclosure or discovery of a trade secret or at least limit the scope of the disclosure. See Fed. R. Civ. P. 26(c)(1)(G).

• 11 In limited circumstances, a responding party who successfully resists a requesting party’s motion to compel discovery may obtain from the requesting party its costs and fees. See Fed. R. Civ. P. 37(a)(5)(B). Rather than waiting for a requesting party to file a motion to compel, a responding party can file a motion for a protective order, see Fed. R. Civ. P. 26(c), but here too, with limited exceptions, it bears its own costs and fees in making the motion, see Fed. R. Civ. P. 26(c)(3).

• 12 See P. Carrington, ‘Politics and Civil Procedure Rulemaking: Reflections on Experience’, 60 Duke Law Journal 597, 609 (2010) (stating that ‘the discovery process … was the central, distinguishing feature of civil procedure’ in federal court); M. Rosenberg, ‘Federal Rules of Civil Procedure in Action: Assessing Their Impact’, 137 University of Pennsylvania Law Review 2203, at 2197 (1989) (‘In the run of significant lawsuits, federal discovery has helped shift the center of gravity from the trial to the pretrial stages.’).

• 13 See T.E. Willging et al., ‘An Empirical Study of Discovery and Disclosure Practices under the 1993 Federal Rule Amendments’, 39 Boston College Law Review 531, at 525 (1998) (describing sample results in which, of the $13,000 spent on litigation expenses in the median case, ‘about half’ went to discovery). See also J.S. Kakalik et al., ‘Discovery Management: Further Analysis of the Civil Justice Reform Act Evaluation Data’, 39 Boston College Law Review 637, at 613 (1998) (estimating that an average of 36% of attorney time spent on a lawsuit was devoted to discovery and discovery motions; further noting that the median time was 25%).

• 14 A proposed 1998 amendment to the Federal Rules of Civil Procedure would have created a similar explicit authority for judges to shift costs for document production; see Preliminary Draft of Proposed Amendments to the Federal Rules of Civil Procedure and Evidence, 181 F.R.D. 18, 88-89 (1998), but the amendment was ultimately rejected, see Rick Marcus, ‘Introduction to Proposals for Cost-Bearing Provisions in the Rules’ (dated 6 September 2013), Tab 5D in Agenda, Advisory Comm. on Civil Rules 216-219 (7-8 November 2013).

• 15 See Fed. R. Civ. P. 26(b)(2)(B) (‘The court may specify conditions for the discovery.’) Rule 26(b)(2)(B) states a unique rule only for electronically stored information. Although the language (‘specify conditions’) does not specifically address cost shifting, it was well understood that cost shifting was the point of this language. See Fed. R. Civ. P. 26(b)(2) advisory committee’s note to 2006 amendment (‘The conditions may also include payment by the requesting party of part or all of the reasonable costs of obtaining information from sources that are not reasonably accessible.’).

• 16 See Fed. R. Civ. P. 26(c)(1)(B) advisory committee’s note to 2015 amendment (noting that ‘courts already exercise this authority’ to allocate expenses of disclosure and discovery); Oppenheimer Fund, Inc. v. Sanders, 437 U.S. 340, 358 (1978) (noting in dicta ‘the district court’s discretion under Rule 26(c) to grant … orders conditioning discovery on the requesting party’s payment of the costs of discovery’); Zubulake v. UBS Warburg, LLC, 216 F.R.D. 280 (S.D.N.Y. 2003) (ordering partial cost shifting for producing electronically stored information).

• 17 Fed. R. Civ. P. 26(c)(1)(B) advisory committee’s note to 2015 amendment.

• 18 Ibid.

• 19 Fed. R. Civ. P. 26(b)(2)(B).

• 20 Fed. R. Civ. P. 26(b)(2)(B) advisory committee’s note to 2006 amendment. A prior provision in Rule 26(b)(2)(B) had made ESI that was not reasonably accessible, non-disclosable and non-discoverable unless the requesting party demonstrated good cause. Rule 26(b)(2)(B) did not provide a definition of ‘reasonably accessible’. For the 2004 case that influenced the structure of Rule 26(b)(2)(B) and developed the concept of ‘reasonable accessibility’, see Zubulake v. UBS Warburg LLC, 217 F.R.D. 309, 317-20 (S.D.N.Y. 2004) (developing the concept of ‘reasonable accessibility’); see also ibid., at 322 (describing the seven factors that bear on whether to permit cost shifting of ESI discovery).

• 21 When initially promulgated in 1938, the Federal Rules of Civil Procedure required judicial approval of many forms of discovery. Virtually all restrictions were removed by amendments in 1946 and 1970. See Fed. R. Civ. P. 33 advisory committee’s note to 1946 amendment; Fed. R. Civ. P. 33 advisory committee’s note to 1970 amendment; Fed. R. Civ. P. 34 advisory committee’s note to 1970 amendment. Today, the only form of discovery requiring judicial approval is a request for a physical or mental examination; see Fed. R. Civ. P. 35(a)(1), although parties must also seek judicial approval for depositions, interrogatories and, in some districts, requests for admission that exceed the maximum permitted under the Rules; see Fed. R. Civ. P. 26(b)(2)(A) (authorising local rules that limit requests for production), 30(a)(2)(A)(i) (limiting depositions to ten per side), 33(a)(1) (limiting interrogatories to twenty-five per party).

• 22 See Fed. R. Civ. P. 26(f) (describing the discovery-planning obligations of the parties), 16(b) (requiring the court to establish deadlines for discovery).

• 23 See Fed. R. Civ. P. 16(a), -(c)(2)(F) (giving federal courts the authority to call pretrial conferences at which measures to control and schedule discovery).

• 24 See Fed. R. Civ. P. 26(c) (protective orders), 37 (motion to compel discovery).

• 25 See Celotex Corp. v. Catrett, 477 U.S. 317, 322 (1986) (suggesting that summary judgment may be granted only ‘after an adequate time for discovery’). As a rule, federal courts establish the deadline for filing case-dispositive motions after the deadline for completing discovery. Cf. Fed. R. Civ. P. 16(b)(3)(A) (requiring that deadlines be set to ‘complete discovery’ and ‘file motions’); Smith v. OSF HealthCare Sys., 933 F.3d 859, 866 (7th Cir. 2019) (reversing summary judgment granted to the defendant before the discovery deadline when additional discovery might have aided the plaintiff to obtain relevant evidence).

• 26 See R.G. Bone, Civil Procedure: The Economics of Civil Procedure 230 (2003) (‘[A] cost-shifting rule deters excessive discovery by forcing a requesting party to internalize discovery costs.’).

• 27 Of course, this logic does not pertain to information held in the possession of non-parties, but as a general matter, third parties hold less information, and discovery from them draws few complaints.

• 28 See R. Cooter and D. Rubenfeld, ‘An Economic Model of Legal Discovery’, 23 Journal of Legal Studies 453, at 435 (arguing for a modified version of a ‘requester-pays’ rule and noting that ‘a cost-shifting rule completely eliminates impositional abuse’). See also J. Setear, ‘The Barrister and the Bomb: The Dynamics of Cooperation, Nuclear Deterrence, and Discovery Abuse’, 69 Boston University Law Review 616, at 569 (1989) (using game theory and the idea of nuclear deterrence to analyse ‘the effect that asymmetries in the availability of resources and opportunities for discovery have on the potential for discovery abuse’).

• 29 Cf. Bone, above n. 26, at 217. (‘In economic terms, an additional investment in discovery is “excessive” whenever the social costs of the investment exceed the social benefits.’) In his 2009 report on discovery costs in the United Kingdom, Lord Justice Jackson advocated for the same principle, arguing that even costs necessary to prosecute or defend a claim can be disproportional in light of the value of the claim and other factors; he proposed that such necessary but disproportional discovery be disallowed. See Review of Civil Litigation Costs: Final Report 37 (2009) (‘If the level of costs incurred is out of proportion to the circumstances of the case, they cannot become proportionate simply because they were “necessary” in order to bring or defend the claim.’) This principle was implemented in the Civil Procedure Rules. See CPR 44.3(2)(a) (allowing only those costs assessed on the standard basis ‘which are proportionate to the matters in issue’); id. (stating that ‘[c]osts which are disproportionate in amount may be disallowed or reduced even if they were reasonably or necessarily incurred’). Cf. CPR 44.3(5) (listing factors bearing on proportionality that mostly mirror those recommended in Lord Jackson’s report).

• 30 See J. Grundfest and P. Huang, ‘The Unexpected Value of Litigation: A Real Options Perspective’, 58 Standford Law Review 1293-1298, at 1267 (2006) (adapting real-option theory, which is often used in deciding whether to invest in research and development, to the decision whether to invest in litigation). For applications of the concept to American civil procedure and to discovery specifically, see J. Tidmarsh, ‘The Litigation Budget’, 68 Vanderbilt Law Review 862-64, at 855 (2015); J. Tidmarsh, ‘Opting Out of Discovery’, 71 Vanderbilt Law Review 1801, 1827 n.115, 1836 (2018).

• 31 At a practical level, this reality has led some products-liability defendants to engage in ‘scorched earth’ discovery tactics that impose such great costs on the opponent that the litigation becomes too expensive to maintain. The classic example of the use of these tactics to suppress plaintiffs’ claims is the tobacco litigation. For a short history of this litigation, see R. Rabin, ‘A Sociolegal History of the Tobacco Tort Litigation’, 44 Stanford Law Review 867, at 853 (1992) (describing how the tobacco defendants were ‘able to wear down the tobacco litigants through a seemingly inexhaustible expenditure of resources’). Asymmetrical incentives to litigate have sometimes been used as an argument to justify class actions, which aggregate the parties on both sides to create a level set of incentives. See D. Rosenberg, ‘Mandatory Litigation Class Actions: The Only Option for Mass Tort Cases’, 115 Harvard Law Review 848-53, at 831 (2002).

• 32 See Miller UK Ltd. v. Caterpillar, Inc., 17 F. Supp. 3d 711, 718 (N.D. Ill. 2014) (‘Where a defendant enjoys substantial economic superiority, it can, if it chooses, embark on a scorched earth policy and overwhelm its opponent. … But even where a case is not conducted with an ulterior purpose, the costs inherent in major litigation can be crippling, and a plaintiff, lacking the resources to sustain a long fight, may be forced to abandon the case or settle on distinctly disadvantageous terms.’).

• 33 See, e.g., Cooter and Rubenfeld, above n. 28; M. Redish and C. McNamara, ‘Back to the Future: Discovery Cost Allocation and Modern Procedural Theory’, 79 George Washington Law Review 773, 822 (2011) (‘An evaluation of fundamental moral, economic, democratic, and constitutional principles reveals that our current method of discovery cost allocation rests on a theoretical foundation that is at best shaky, and at worst completely illusory.’); M. Redish, ‘Discovery Cost Allocation, Due Process, and the Constitution’s Role in Civil Litigation’, 71 Vanderbilt Law Review 1847 (2018) (arguing that shifting discovery costs may be mandated by due process). See generally E. Elliott, ‘How We Got Here: A Brief History of Requester-Pays and Other Incentive Systems to Supplement Judicial Management of Discovery’, 71 Vanderbilt Law Review 1785 (2018) (reviewing prior proposals to require cost shifting).

• 34 For a telling analysis of lawyers’ economic incentives to complicate and protract litigation, see A.A.S. Zuckerman, ‘Lord Woolf’s Access to Justice: Plus ça change …’, 59 Modern Law Review 775–778, at 773 (1996). This incentive is exacerbated in a loser-pays regime, because ‘a litigant who believes that an increase in the amount spent on litigation will increase his chances of success has good reason for progressively raising his stakes’. Ibid., at 779. Of course, shifting only discovery costs limits this incentive; but given how large a component of overall litigation expenditures discovery costs are, see above n. 13, the basic point holds.

• 35 To avoid this obligation, it seems not unlikely that parties in symmetrical-information cases would agree to waive discovery-cost reimbursement. Cf. Abernathy v. E. Ill. R.R., 940 F.3d 982, 994 (7th Cir. 2019) (noting that each party bore the expense of its own expert in providing pretrial testimony to the opposing party, even though the parties could have sought reimbursement under Rule 26(b)(4)(E)).

• 36 See, e.g., Lawson v. Spirit AeroSystems, Inc., No. 18-1100-EFM-ADM, 2020 WL 3288058 (D. Kan. 18 June 2020); CFPB v. Ocwen Fin. Corp., Case No. 9:17-CV-80495-Marra/Matthewman, 2018 WL 6843629 (S.D. Fla. 21 December 2018).

• 37 See, e.g., Brogan v. Fred Beans Motors of Doylestown, Inc., Civil Action No. 17-5628, 2020 WL 5017500 (E.D. Pa. 25 August 2020), Travelers Prop. Cas. Co. of Am. v. Mountaineer Gas Co., Civil Action No. 2:15-cv-07959, 2018 WL 899078 (S.D. W. Va. 15 February 2018); Oxbow Carbon & Minerals LLC v. Union Pac. R.R., 322 F.R.D. 1 (D.D.C. 2017).

• 38 See, e.g., McClurg v. Mallinckrodt, Inc., Case No. 4:12-CV-00361-AGF, 2016 WL 7178745 (E.D. Mo. 9 December 2016) (shifting 18% of defendants’ document-collection expenses to plaintiffs).

• 39 See D. Engstrom, ‘The Twiqbal Puzzle and Empirical Study of Civil Procedure’, 65 Stanford Law Review 1214, at 1203 (2013) (discussing the ‘serious concerns’, especially sampling bias, with using only Westlaw and Lexis databases to conduct empirical research on district-court behaviour); E. McCuskey, ‘Submerged Precedent’, 16 Nevada Law Journal 522, at 515 (2016) (‘It is well-documented that district-court opinions selected for the print reporter volumes (the Federal Supplement and Federal Rules Decisions) may not be representative of decision-making, and that therefore reliance solely on reported decisions to study judicial behavior risks biased results.’).

• 40 I conducted the research principally in 2016 and in 2019, while reviewing hundreds of recent decisions to update discovery chapters in a co-authored casebook.

• 41 See, e.g., Oxbow, 322 F.R.D. at 11 (‘Oxbow has failed to rebut the presumption imposed by the Federal Rules of Civil Procedure that it should bear the cost of complying with Defendants’ proposed discovery.’).

• 42 There are 677 federal district-court judges in the United States, as well as another 588 federal magistrate judges who are often tasked by district-court judges to resolve discovery matters. See ‘Judicial Business 2020, U.S. District Courts’ tbl.5 (Admin. Office of U.S. Courts) www.uscourts.gov/statistics-reports/us-district-courts-judicial-business-2020 (last visited 21 June 2022); ‘Judicial Business 2019, Status of Magistrate Judge Positions and Appointments’ tbl.13 (Admin. Office of U.S. Courts), https://www.uscourts.gov/statistics-reports/status-magistrate-judge-positions-and-appointments-judicial-business-2021 (last visited 21 June 2022). Should a magistrate judge’s decision on a discovery matter be appealed to the district judge, the judge reviews the decision on a deferential basis. See 28 U.S.C. §636(b)(1)(A) (permitting reconsideration of a magistrate judge’s order only when the order ‘is clearly erroneous or contrary to law’). Likewise, great deference is given to the discovery decisions of district-court judges on appeal. See Khoury v. Miami–Dade Cty. Sch. Bd., 4 F.4th 1118, 1125 (11th Cir. 2021) (‘We review the District Court’s ruling on discovery matters only for abuse of discretion.’). With such a diffusion of authority, it is a near certainty that different judges will make different decisions on similar facts.

• 43 In a third or more of federal cases, no discovery occurs, and discovery in most other cases is not burdensome. See Kakalik et al., above n. 13, 636, at 613 (1998) (reporting that no discovery occurred in 38% of cases in a sample and that ‘[d]iscovery is not a pervasive litigation cost problem for the majority of cases’); Willging et al., above n. 13, at 530-31 (reporting that, in a sample of cases likely to involve discovery, only 85% of attorneys reported that discovery occurred and that the median cost of discovery was $13,000 per client, or about 3% of the stakes in the case). Data from state courts show that discovery costs constitute only a small fraction of the total recovery from litigation. E. Helland et al., ‘Contingent Fee Litigation in New York City’, 70 Vanderbilt Law Review 1971, 1988 (2017) (describing a random sample of New York tort cases in which the median for all expenses, other than attorney’s fees, was 3% of recoveries and the average was 5%). In the main, lawyers are satisfied with the discovery process, believing that it reveals the right amount of information at a fair price. See Kakalik et al., above n. 13, at 636 (‘Subjective information from our interviews with lawyers also suggests that the median or typical case is not “the problem.”|’); E.G. Lee III and T.E. Willging, National, Case-Based Civil Rules Survey 27, 34 tbl.4, 37 tbl.5, 43 tbl.10 (2009) (reporting survey results in which a clear majority of both plaintiffs’ and defence lawyers believed that discovery revealed ‘just the right amount’ of information, the median costs for discovery by plaintiffs’ lawyers was $15,000 and by defence lawyers $20,000, and the costs of discovery in relation to the stakes of the litigation were 1.6% for plaintiffs and 3.3% for defendants). Contra, Interim Report on the Joint Project of the American College of Trial Lawyers Task Force on Discovery and the Institute for the Advancement of the American Legal System 3-5 (2008) (describing survey results that ‘confirm[] that there are serious problems in many parts of the civil justice system, especially the rules governing discovery’).

• 44 For a fuller review of these reforms, see Tidmarsh (2018), above n. 30, at 1813-1814.

• 45 See Fed. R. Civ. P. 16.

• 46 Cf. E. Elliott, ‘Managerial Judging and the Evolution of Procedure’, 53 University of Chicago Law Review 326-34, at 306 (1986) (arguing that incentives-based approaches are more efficient than command-and-control procedural rules).

• 47 See R.A. Posner, Economic Analysis of Law §21.1 (7th ed. 2007).

• 48 This statement is not precisely accurate, and I will refine it shortly. See below n. 53 and accompanying text. As an initial matter, however, procedural scholars usually proceed from the assumption that keeping the sum of direct litigation costs and error costs to their minimum is the best way to keep overall social costs at their minimum.

• 49 Of course, the same criticism can be levelled against a system of discovery in which costs are not shifted. But, as I have described, there is very little data suggesting that the cost of discovery in the present non-cost-shifting world is problematic. See above n. 43-44 and accompanying text. The burden of proof for making a substantial change lies with the proponents of cost shifting.

• 50 For an overview of third-party litigation funding in the United States and some of its concerns, see M. Shapiro, ‘Distributing Civil Justice’, 109 Georgetown Law Journal 1509-1512, at 1473 (2021). It is fair to say that judges and legislatures have so far treated the arrival of third-party funding with a fair amount of scepticism. Ibid., at 1510. For a more optimistic appraisal of third-party litigation financing, see S. Bedi and W. Marra, ‘The Shadows of Litigation Finance’, 74 Vanderbilt Law Review 563 (2021) (arguing that the prospect of litigation funding positively affects parties’ pre-litigation behaviour, especially in contractual settings, and also has positive post-litigation effects).
  Concerns over third-party funding have led both lawmakers and rule-makers to consider changes that would require disclosure of the identity of third-party funders and any funding agreements. See, e.g., Litigation Funding Transparency Act of 2021, S. 840, 117th Cong. (2021) (requiring disclosure of funding sources in class action and multidistrict litigation); Agenda, Advisory Committee on Civil Rules 371-86 (5 October 2021) (describing history of efforts to regulate third-party funding by rule or legislation and recommending against immediate action on the issue). No proposals to regulate third-party funding have yet been enacted at the federal level; and in any event, disclosure obligations would likely have little effect on the issues described in the text.

• 51 M. Cappelletti and B. Garth, ‘Access to Justice: The Newest Wave in the Worldwide Movement to Make Rights Effective’, 27 Buffalo Law Review 181, 182 (1978); see also ibid., at 185 (‘Effective access to justice can thus be seen as the most basic requirement – the most basic “human right” – of a modern, egalitarian legal system which purports to guarantee, and not merely proclaim, the legal rights of all.’).

• 52 See E. Labaton, ‘Courts on Trial Symposium Closing Remarks’, 40 Arizona Law Review 1113, at 1111 (1998) (‘Access to American courts is available for two reasons: the contingent fee and the American rule.’); see also A. Goodhart, ‘Costs’, 38 Yale Law Journal 874, at 849 (1929) (‘Apart from purely historical reasons , the American rules as to costs may also be due in part to a vague feeling that they favor the poor man, and are therefore democratic, while the English system helps the wealthy litigant.’); cf. ibid., at 877 (noting that allowing a judge ‘wide discretion [to tax costs] as is inherent in the English system would be contrary to the general American conception of a judiciary bound by fixed rules’).

• 53 See G. Calabresi, The Cost of Accidents (1970), at 26-31.

• 54 For example, assume that, with precautions costing $400, the defendant’s conduct causes $800 in harm to eight plaintiffs (or $100 apiece); and the defendant’s discovery costs in ensuing litigation with the plaintiffs are $240 (or $30 apiece). The defendant could spend $510 in safety precautions, and only seven plaintiffs would suffer harm (for a total of $700). Discovery costs would be reduced to $210. From a social viewpoint, it would be better if the defendant took the extra $110 in safety precautions, since the total cost is $1,420, compared with total costs for the lower-precaution alternative of $1,440. If the defendant bears the cost of discovery, it will take the extra precaution, because an expenditure of $110 saves the defendant $130 in liability and litigation costs. But if the defendant does not bear the cost of discovery, the defendant will not take the extra precaution, for it would be spending an additional $110 and receiving a benefit of only $100. Obviously, this example is stylised, and its result does not pertain across all permutations. There are more realistic examples, which factor into the likelihood of suit, that better prove the point, but they would take more time to unpack than is justified for present purposes. The point of the given example is only to illustrate that there are circumstances in which the potential exposure to litigation costs can induce a party to undertake a socially appropriate level of care that it might not undertake if it did not bear its own discovery costs.

• 55 Congress or the federal judiciary has sometimes developed pilot programmes in some federal district courts to generate data on the efficacy of proposed procedural reforms. See 28 U.S.C. §651(b) (requiring each federal district court to authorise the use of one or more alternative dispute resolution processes); J. Sutton and D. Webb, ‘Bold and Persistent Reform: The 2015 Amendments to the Federal Rules of Civil Procedure and the 2017 Pilot Projects’, 101 Judicature 12 (Autumn 2017) (touting the benefits of pilot projects on discovery reform). Piloting programmes and gathering data on their successes and failures seem appropriate before undertaking major reformation of a procedure as foundational to the American system as discovery.

• 56 See CPR 3.12-3.18. There are exceptions; multi-track cases exceeding £10 million are exempt from costs budgeting, see CPR 3.12(1)(a)-(b), as are cases involving minors, see CPR 3.12(1)(d), and those that are subject to ‘fixed costs or scale costs’, see CPR 3.12(1)(d). The court can also exempt a multi-track case from costs budgeting by order. CPR 3.12(1)(e). Conversely, costs budgeting may also apply in other cases when the court so orders. See CPR 3.12(1A).
  The Civil Procedure Rules divide cases into three tracks: ‘multi-track cases’, ‘small claims cases’, and ‘fast track cases’. ‘Multi-track cases’ are defined as cases that are neither ‘small claims cases’ nor ‘fast-track cases’. See CPR 26.6(6) (‘The multi-track is the normal track for any claim for which the small claims track or the fast track is not the normal track.’). In general, ‘small claims’ cases are those personal-injury cases whose value does not exceed £10,000 (with some additional exceptions), certain landlord-tenant disputes with a value less than £1,000, and other cases whose value does not exceed £10,000. CPR 26.6(1)-(3). In general, ‘fast-track’ cases are those cases whose value does not exceed £25,000, the trial will likely last no more than one day and expert witnesses are limited. See CPR 26.6(4)-(5). Other factors, such as the number of likely parties, the complexity of the issues, and the importance of the case to non-parties, can influence the court’s decision in assigning a case to a given track. See CPR 26.8.

• 57 A costs budget is a broader concept because it would also impose caps on other pretrial and trial expenditures.

• 58 Originally set out in Rule 44 in 2009, costs capping was moved to CPR 3.19-21 in 2013, at the time that costs budgeting was installed in the Civil Procedure Rules.

• 59 Costs capping is unlikely to arise in small-claims-track or fast-track litigation; in small-claims cases, each party bears most of its own costs, see CPR 27.14, while in fast-track cases, fixed costs are the norm, see CPR 45.37-40. Thus, the main terrain for costs capping is the multi-track case, including those multi-track cases ordinarily exempt from costs budgeting. See above n. 56.

• 60 Practice Direction 3F ¶1.1.

• 61 See, e.g., Crystal Lakes v. Bath & Body Works, LLC, No. 2:16-CV-2989-MCE-GGH, 2018 WL 533915, at *2 n.1 (E.D. Cal. 23 January 2018) (noting that, without further clarification about the scope of the plaintiff’s claims, ‘it is difficult to determine the proportionality issue on his discovery demands’). See also 8 C.A. Wright et al., Federal Practice and Procedure §2008.1 (2010) (stating that ‘the proportionality concept … seemed to require great familiarity with the case’, thus somewhat frustrating its implementation).

• 62 CPR 3.18(b) (noting that a court, when assessing costs on the standard basis, will ‘not depart from such approved or agreed budgeted costs unless satisfied that there is good reason to do so’).

• 63 See J. Tidmarsh, ‘Realising the Promise of Costs Budgets: An Economic Analysis’, 35 Civil Justice Q 219 (2016).

• 64 See Tidmarsh (2018), above n. 30, at 888-89.

• 65 Other practical, legal and constitutional issues would also need to be addressed before a realistic costs-capping measure could be implemented. See ibid., at 901-17.

• 66 I know of no scholar whose proposals go quite so far, but a few American writers have advocated the abolition of nearly all discovery. See G. Shepard, ‘Failed Experiment: Twombly, Iqbal, and Why Broad Discovery Should be Further Eliminated’, 49 Indiana Law Review 466, at 465 (2016) (‘Broad discovery should be eliminated. It is a seventy-year experiment that has failed. The rest of the world recognizes this …’).

• 67 See Common Law Procedure Act 1854, 17 & 18 Vict. c. 125 (Eng.) (authorizing common-law courts to conduct discovery); K. Funk, ‘Equity Without Chancery: The Fusion of Law and Equity in the Field Code of Civil Procedure, New York 1846-1876’, 36 Journal of Legal History 152, 174-79 (2015) (describing the breadth of discovery in New York before and after the adoption of the Field Code in 1848); E. Sunderland, ‘Scope and Method of Discovery Before Trial’, 42 Yale Law Journal 869-70, at 863 (1933) (describing the scope of discovery in American and Canadian jurisdictions prior to the Federal Rules of Civil Procedure).

• 68 See, e.g., J. Langbein, ‘The German Advantage in Civil Procedure’, 52 University of Chicago Law Review 827-29, at 823 (1985) (discussing the German approach to fact-gathering by the judge).

• 69 As mentioned, there are 677 authorised federal trial-level judges; and at any given time, a number of those positions remain unfilled. See above note 43. As of September, 2020, the number of cases pending in federal court was 511,666. See ‘Federal Judicial Caseload Statistics 2020’ (Admin. Office of U.S. Courts), www.uscourts.gov/statistics-reports/federal-judicial-caseload-statistics-2020 (last visited 1 October 2021). That number works out to more than 755 cases per authorised judgeship.

• 70 See Fed. R. Civ. P. 26(a) (requiring the mandatory disclosure of information supporting a party’s claims or defences, expert witnesses and witnesses and documents to be used at trial).

• 71 For a well-crafted plan to expand the parties’ mandatory-disclosure obligations, while still permitting some modest follow-on discovery, see D. Rosenberg et al., ‘A Plan for Reforming Federal Pleading, Discovery, and Pretrial Merits Review’, 71 Vanderbilt Law Review 2059 (2018).

• 72 See Tidmarsh (2018), above n. 30.

• 73 See Fed. R. Civ. P. 12(b)(6); Bell Atl. Corp. v. Twombly, 550 U.S. 544, 557-58 (2007) (imposing a plausibility pleading standard); see also Ashcroft v. Iqbal, 556 U.S. 662, 678-79 (2009) (clarifying the scope of the plausibility standard).

• 74 See Fed. R. Civ. P. 56(a); Anderson v. Liberty Lobby, Inc., 477 U.S. 242 (1986) (discussing the summary-judgment standard); Celotex Corp. v. Catrett, 477 U.S. 317 (1986) (same).

• 75 A second carrot, when both parties waive discovery, is to waive jury trial in cases that would be triable to a jury when both parties opt to conduct discovery. Admittedly, some parties may regard the waiver of jury-trial rights as a stick rather than a carrot to forgo discovery, but nothing turns on whether it is characterised as a carrot or a stick. Eliminating jury trial would allow a court to adopt certain case-management tools that could make an American trial run more along the lines of civil-law trials and thus avoid needless discovery on issues that the court never reaches. See Langbein, above n. 68, at 830 (describing how ‘[t]he implications for procedural economy are large’ under the episodic nature of German trial),

• 76 The types of factors that might inform a court’s discretion not to shift costs would be essentially those given in Lord Jackson’s 2009 report and presently found in Rule 44.3(5) of the Civil Procedure Rules (including the amount in controversy, the value of non-monetary relief, the complexity of the case, and broader considerations about the case’s public importance), as well as the essentially overlapping factors identified as relevant to the proportionality inquiry under Federal Rules of Civil Procedure 26(b)(1) (including the amount in controversy, the importance of the issues at stake in the case, the parties’ resources and their relative access to information).

• 77 See above n. 39-42 and accompanying text.

• 78 For the classic opposition to the wide-scale resolution of disputes through private ordering systems, see O. Fiss, ‘Against Settlement’, 93 Yale Law Journal 1073 (1984).

• 79 See Gilmer v. Interstate/Johnson Lane Corp., 500 U.S. 20, 31 (1991) (‘Although [discovery] procedures might not be as extensive [in arbitration] as in the federal courts, by agreeing to arbitrate, a party trades the procedures and opportunity for review of the courtroom for the simplicity, informality, and expedition of arbitration.’ (internal quotation marks omitted)); Dodson Int’l Parts, Inc. v. Williams Int’l Co., 12 F.4th 1212, 1230 (10th Cir. 2021) (‘[O]ne cannot expect full discovery in arbitration proceedings, as extensive discovery could undermine much of the advantage of arbitration.’).

• 80 In this hypothetical case, if cost shifting occurred in just a third of the cases that did not opt out of discovery, the savings would be even larger: the reduction in discovery costs would now rise to $15,000 ($12,000 in the case that opted out of discovery plus $1,000 apiece in the cases in which discovery costs shifted).