UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA Elouise Pepion Cobell, et al.,: : Plaintiffs, : : v. : : Civil Action No. 96-1285 (JR) Dirk Kempthorne, Secretary of : the Interior, et al., : : Defendants. : FINDINGS OF FACT AND CONCLUSIONS OF LAW These findings and conclusions are the result of a 10­day bench trial in October 2007. The central purpose of the trial was to determine whether the Department of Interior has remedied or is remedying what Judge Lamberth found in Cobell v. Babbitt, 91 F. Supp. 2d 1, 58 (D.D.C. 1999) (Cobell V), aff’d, Cobell v. Babbitt, 240 F.3d 1081 (D.C. Cir. 2001) (Cobell VI), to be a breach of its duty under the Indian Trust Fund Management Reform Act of 1994 to produce an accounting for Individual Indian Money (IIM) account holders. In setting the matter for trial, I said that, although the details of the trial remained to be worked out, it was both appropriate and prudent to review the Interior Department’s historical accounting project in detail, and to do so in open court, where the government might present, and plaintiffs might test or challenge, its methodology and results up to the time of the hearing [Dkt. 3312]. The end product of such a proceeding was to include the answers to at least the following questions: • Have the defendants cured (or are they curing) the breaches of their fiduciary duty that were found in Cobell V? • Do the defendants’ historical statements of account . . . satisfy defendants’ duties “rooted in and outlined by the relevant statutes and treaties . . . [and] defined in traditional equitable terms”? Cobell VI, 240 F.3d at 1099. • Have the defendants unreasonably delayed the completion of the required accounting? • What further relief, if any, should be ordered? By the time the trial began, the issues for trial had been distilled to these four: First, it’s going to be about what you’re doing and what you’re not doing. . . . It’s going to be about both of those things. Second, what would it cost to do the things that they say that you should be doing and you’re not doing? Third, taking into account the cost, because that, I think, I’m required to do by the Court of Appeals, is what you’re doing adequate? Is it an adequate accounting? And fourth -- and this is what you don’t want to hear, but I think Mr. Gingold is entitled to at least a record on this point, fourth, what does it all add up to? Throughput versus what you can prove, what are the big numbers? H’rg Tr. 76:23 - 77:10 (6/18/07). The question of what further relief, if any, should be ordered was left to another day. These findings and conclusions, derived not only from the trial, but also from the extensive record that preceded it, support and explain my decision (i) that, although the defendants have attempted and continue to attempt to cure the breach of their fiduciary duty that was found in Cobell V and affirmed by Cobell VI, they have not succeeded in doing so; (ii) that the historical statements of account contemplated by defendants’ latest accounting plan will not satisfy defendants’ duties “rooted in and outlined by the relevant statutes and treaties . . . [and] defined in traditional equitable terms,” Cobell VI, 240 F.3d at 1099; and (iii) that the defendants have unreasonably delayed the completion of the required accounting. Indeed, it is now clear that completion of the required accounting is an impossible task. BACKGROUND To say that the histories of the IIM trust and of this lawsuit have been exhaustively chronicled in district court and appellate opinions is to stretch the limits of understatement. See, e.g., Cobell v. Babbitt, 30 F. Supp. 2d 24, 27-29 (D.D.C. 1998); Cobell v. Babbitt, 91 F. Supp. 2d 1, 6-12 (D.D.C. 1999); Cobell v. Norton, 240 F.3d 1081, 1086-94 (D.C. Cir. 2001); Cobell v. Norton, 226 F. Supp. 2d 1, 11-20 (D.D.C. 2002); Cobell v. Norton, 283 F. Supp. 2d 66, 72-86 (D.D.C. 2003). Those seeking CliffsNotes can even consult the Cobell v. Kempthorne Wikipedia entry (though the Court, of course, cannot vouch for its accuracy). At this date, there are 3,504 entries on the Cobell v. Kempthorne docket. Appellate panels hearing Cobell arguments have engaged ten of our Circuit judges, some of them more than once. Upon publication, this opinion will have the shorthand title Cobell XX. Nevertheless, those histories must be retold at least briefly in order to provide context for today’s opinion. Plaintiffs are a certified class of present and former IIM account holders numbering in excess of 300,000. Some account holders have more than one IIM account. Hundreds of thousands of IIM accounts exist, managed for the United States by its trustee- delegates, the Department of Interior and the Department of Treasury. Most of these IIM accounts exist to receive income the government collects for leasing or selling Indian-owned lands and then to distribute it to account holders when account balances reach certain thresholds (usually fifteen dollars). A small percentage of the funds flowing through the IIM trust are in “Judgment” and “Per Capita” accounts, which were created to hold funds derived from litigation settlements (Judgment accounts) and tribal revenues allocable to individual Native Americans (Per Capita accounts). By far the largest amount of trust funds flow through the “land-based” IIM accounts that contain lease, royalty, and land sale payments tied to individual land allotments. Individual Indian land allotments date to a period between the late 1800’s and 1934 when the federal government attempted to dismantle tribes and instill the Anglo-American concept of private ownership in Native Americans by carving reservation land into individually owned parcels of up to 160 acres (now known as “tracts” or “allotments”). See, e.g., Yakima v. Yakima Indian Nation, 502 U.S. 251, 254 (1992) (“The objectives of allotment were simple and clear cut: to extinguish tribal sovereignty, erase reservation boundaries, and force assimilation of Indians into the society at large.”), quoted in Cobell v. Norton, 240 F.3d 1081, 1087 (D.C. Cir. 2001). The government’s pursuit of the allotment policy occurred alongside its official abandonment of treaty-driven relationships with tribes in favor of “govern[ing] [tribes] by acts of Congress.” United States v. Kagama, 118 U.S. 375, 382 (1886); see Act of March 3, 1871, ch. 120, § 1, 16 Stat. 566 (1871) (codified as amended at 25 U.S.C. § 71). This policy shift and its corollary acts -- such as coercive assimilation -- were carried out without so much as the pretense of tribal consent. See, e.g., Lone Wolf v. Hitchcock, 187 U.S. 553, 565-68 (1903). The allotment policy was first codified in the Indian General Allotment Act (Dawes Act), ch. 119, 24 Stat. 388 (1887) (codified as amended at 25 U.S.C. § 331 et seq.), and was reflected in several subsequent allotment acts. In the Dawes Act, Congress granted unilateral authority to the executive branch to divide reservation land west of the Mississippi into plots for individual tribal members and families. It also allowed non-Indian settlement upon and exploitation of some reservation land, resulting in the alienation of millions of acres from tribal ownership. The statute required the federal government to hold the allotted land in trust for the individual allottees and their heirs for a period of 25 years -- a period subject to extension at the government’s discretion -- after which fee patents would issue to the allottees. See Cobell VI, 240 F.3d at 1087; Cobell v. Norton, 283 F. Supp. 2d 66, 74 (D.D.C. 2003). While held in trust, allotted lands were to be immune from state taxation. The expectation was that, during that time, Indians would establish self-sufficient farms and earn enough money to pay their own taxes. At the close of the 19th Century, Congress passed several acts allowing the government to lease allotments that had not been successfully cultivated. See, e.g., An Act Making Appropriations for Current and Contingent Expenses of the Indian Department and Fulfilling Treaty Stipulations with Various Indian Tribes for the Fiscal Year Ending June Thirtieth, Eighteen Hundred and Ninety-Five, and for Other Purposes, ch. 290, 28 Stat. 286 (1894).1 Any income 1This Act provided in part that [d]uring the trust period of twenty-five years, such part of the lands which have been allotted to members of the Yankton tribe of Indians in severalty, as the owner thereof can not cultivate or otherwise use advantageously, may be leased for one or more years at a time. But such leasing shall be subject to the approval of the Yankton Indian agent by and with the consent of the Commissioner of Indian Affairs; and provided that such leasing shall not in any generated from the land was to flow into IIM accounts established by the government. Native American landowners could not (and today still cannot) sell or lease allotted land without the government’s consent. Congress soon realized that the allotment policy was responsible for innumerable problems, see, e.g., 1915 Congressional Report to the Joint Commission of the Congress of the United States, PX-681, not least of which was the phenomenon known as ‘fractionation’. Fractionation occurs when Indian allotments are divided and divided again by inheritance through succeeding generations, diluting the ownership interests of allottees and causing enormous administrative difficulties for the BIA. The Indian Reorganization Act (Wheeler-Howard Act), ch. 576, 48 Stat. 984 (1934), (codified as amended at 25 U.S.C. § 461 et. seq.), was supposed to reconsolidate Indian lands and reverse the allotment process, but the land reclamation effort prescribed case interfere with the cultivation of the allotted lands by the owner thereof to the full extent of the ability of such owner to improve and cultivate his holdings. The intent of this provision is to compel every owner of allotted lands to cultivate the same to the full extent of his ability to do so, before he shall have the privilege of leasing any part thereof, and then he shall have the right to lease only such surplus of his holdings as he is wholly unable to cultivate advantageously. 28 Stat. 286, 316. by that statute was never properly funded and never materialized. Instead, the Act succeeded only in ending the creation of new allotments and, for allotted lands already held in trust, extending the trust period indefinitely. Cobell v. Norton, 283 F. Supp. 2d 66, 75 (D.D.C. 2003). As of 1990, some eleven million acres were held in trust for the heirs of allottees, by now several generations removed.2 Many trust allotments are owned in common by hundreds or even thousands of beneficiaries -­each with undivided (“fractionated”) interests in the whole parcel. The statute that gave rise to this litigation was enacted in 1994 as the Indian Trust Fund Management Reform Act, Pub. L. No. 103-412, 108 Stat. 4239 (codified at 25 U.S.C. § 4001 et seq.) (hereinafter “the 1994 Act”). The 1994 Act reflected many years of congressional frustration over Interior’s handling of the IIM trust. It commanded Interior, among other things, to provide an historical accounting to trust beneficiaries. 25 U.S.C. § 4011(a). Two years after the enactment of the 1994 Act, concerned that the required accounting had been neither accomplished nor even begun, the plaintiffs filed this suit. They alleged that the defendants were in breach of their 2Seventy years earlier, in 1920, 35.897 million acres ofallotted land were owned by individual Indians. FRANCIS PAUL PRUCHA, 2 THE GREAT FATHER: THE UNITED STATES GOVERNMENT ANDAMERICAN INDIANS 865 (1984). fiduciary duties, and they prayed for an accounting and various other forms of declaratory, injunctive, and other equitable relief. Judge Lamberth certified the plaintiff class on February 4, 1997 [Dkt. 27]. On November 5, 1998, he ruled that plaintiffs were not entitled to mandamus relief, but otherwise rejected the government’s motion to dismiss the complaint. Cobell v. Babbitt, 30 F. Supp. 2d 24 (D.D.C. 1998) (Cobell I). On June 7, 1999, he denied defendants’ motion for summary judgment [Dkt. 317], and, six months later, he declared defendants to be in breach of their statutory trust obligations, dismissed plaintiffs’ common law claims, and remanded the case to the agency with an injunction to bring its actions into conformity with its fiduciary duties and to submit quarterly reports on its progress (which quarterly reports it has been submitting ever since). Cobell v. Babbitt, 91 F. Supp. 2d 1 (D.D.C. 1999) (Cobell V). The Court of Appeals affirmed the declaration of breach, but vacated several of Judge Lamberth’s more specific findings. Cobell v. Norton, 240 F.3d 1081 (D.C. Cir. 2001) (Cobell VI). Judge Lamberth then divided the proceedings into two phases: one addressing the “to-be plan,” or the agency’s plans to remedy its breach going forward, and another addressing defendants’ historical accounting work. After a 44-day trial, he issued an historical accounting opinion, a “fixing the system” opinion, and a structural injunction. Cobell v. Norton, 283 F. Supp. 2d. 1 (D.D.C. 2003) (Cobell X). The historical accounting portion of the structural injunction was reversed by the D.C. Circuit on December 10, 2004, Cobell v. Norton, 392 F.3d 461 (D.C. Cir. 2004) (Cobell XIII); reinstated by Judge Lamberth after Congress failed to meet its self-imposed deadline for achieving the legislative settlement on which the Court of Appeals had relied in deciding Cobell XIII, Cobell v. Norton, 357 F. Supp. 2d 298 (D.D.C. 2005) (Cobell XIV); stayed by the Court of Appeals, Cobell v. Norton, 2005 U.S. App. LEXIS 5788 (D.C. Cir. Apr. 7, 2005); and eventually reversed, Cobell v. Norton, 428 F.3d 1070 (D.C. Cir. 2005) (Cobell XVII). On July 11, 2006, the Court of Appeals issued its eighth and ninth published opinions in the case, and directed that the case be reassigned. Cobell v. Kempthorne, 455 F.3d 301 (D.C. Cir. 2006) (Cobell XVIII); Cobell v. Kempthorne, 455 F.3d 317 (D.C. Cir. 2006) (Cobell XIX). On December 12, 2006, the case was assigned to me [Dkt. 3278]. During Judge Lamberth’s heroic stewardship of this case, he was beset by a host of important but ancillary issues: vulnerabilities within the Interior Department’s information technology systems, civil contempt proceedings concerning several government employees, retaliatory action within the agency against agency employees testifying on plaintiffs’ behalf, objections to communication between the agency and the plaintiff class, and the appointment and removal of court monitors and special masters, to name a few. The nineteen published opinions in this case have yielded no definitive, undisturbed ruling on the core question that looms over this dispute, which is: What is the scope or nature of the accounting that is required by the 1994 Act? This opinion seeks to answer that question. The outline set forth below emerges from the testimony taken in October 2007 on the trial questions identified above, and from the enormous record that has been compiled over the past eleven years. I. THE GOVERNMENT’S ACCOUNTING EFFORT 1. Evolution of the current historical accounting project. 2. Activities the government has planned -- and now plans -- to discharge its accounting duty. 3. Activities the government has declared beyond the scope of its plan. 4. Cost to do what is not being done. 5. “Throughput” -- amount of money received into the IIM trust; amount distributed to IIM beneficiaries. II. COMPLIANCE WITH THE FIDUCIARY DUTY TO ACCOUNT 1. Does this court have jurisdiction? 2. Taking cost into account, will Interior’s 2007 Historical Accounting Plan result in an adequate accounting that is compliant with the 1994 Act, prior Cobell opinions, and other precedent? I. THE GOVERNMENT’S ACCOUNTING EFFORT I.1. Evolution of the current historical accounting project No real accounting, historical or otherwise, has ever been done of the IIM trust. What is now called the historical accounting project would have been unthinkable before the advent, some time in the 1980s, of automated systems that began to offer the possibility of merging the records and procedures that had been used for nearly 100 years in the various regions and agencies of the Bureau of Indian Affairs. This section describes the legacy computer systems that first held out some promise of doing an historical accounting, some of the problems that were encountered soon after the 1994 Act mandated an historical accounting, the exaggerated promises of Interior’s 2003 historical accounting plan, various ancillary IIM-related projects that Interior has undertaken, and the establishment in 2004 of the American Indian Records Repository, which has brought some degree of order out of the chaos that was IIM trust record- keeping. I.1.A. Systems used in administering the IIM trust The government uses an array of data and financial 3 systems, referred to mostly by acronyms, to collect, manage, and distribute IIM trust funds. The two major types of systems 3See Appendix A. relevant to historical accounting contain land ownership data and financial data. i. Integrated Records Management System (IRMS) A major premise of the government’s accounting effort is that the transition from paper to electrons took the accuracy, completeness and reliability of IIM trust data to a level that far surpassed the “paper ledger era.” The agencies refer to February 1985 as the beginning of the new “electronic ledger era.” The first electronic ledger was the Integrated Records Management System (IRMS). IRMS contained four platforms for the input and access of lease data, ownership data, account data, and transactional data by BIA employees. Tr. 2041:1-23 (Christie); Tr. 343:22-344:5 (Ramirez).4 The “IIM platform” within IRMS contained most of the accounting data. Tr. 343:23-25 (Ramirez). The government considers IRMS the beginning of the “electronic ledger era,” an era when data became putatively more uniformly handled and more easily accessible. Some BIA locations began using IRMS long before February 1985, however -- some as early as the 1970s. Tr. 288:13-20 (Ramirez). The first regional office to use IRMS was Billings, Montana (one of the twelve regional BIA offices, now called the Rocky Mountain regional office). 4Transcript citations to the October 2007 bench trial willbe referenced only by page and line number and witness name.Transcript citations to other proceedings will also include theabbreviated title of the proceeding and the date of thetestimony. Tr. 1743:24-1744:2 (Infield). Interior did not require all BIA regions to use IRMS until the late 1980s. Tr. 1743:17-21 (Infield). Because IRMS was implemented on a rolling basis, it is impossible to draw a hard line between the paper ledger era and electronic ledger era. Up until 1991, moreover, Interior deleted from IRMS data that were more than six years old, so as not to overwhelm their rudimentary computer systems. Data recorded before 1985 has been erased. Tr. 1744:22-1746:11 (Infield). And the post­1985 data on the IRMS system is incomplete and difficult to 5 analyze, since some agencies and regional offices did not convert to IRMS until after 1985, Tr. 693:1-6 (Herman), at least one agency (Osage) handled money collected from oil and gas leases in a unique way, Tr. 1737:4-1741:6 (Infield), bookkeeping within IRMS varied across regions, Tr. 572:19-573:15 (Herman), and even offices that did use IRMS did not always keep complete records on the system. Tr. 789:22-790:13 (Herman). ii. Trust Funds Accounting System (TFAS) In August 1998, Interior began its conversion from IRMS to the Trust Funds Accounting System (TFAS). The conversion was complete by March 2000. Tr. 450:18-24 (Herman). The upgrade to 5Underneath the BIA’s twelve regional offices are 93“agencies” or “agency offices” responsible for, among otherthings, processing IIM transactions. Feb. 2002 OHTA AccountingRecords Conference Materials at 213, AR-058. TFAS gave BIA the ability to send quarterly IIM statements, Tr. 866:15-867:9 (Winter), which it began to do in 2000. Those statements reflect opening and ending balances as well as posted transactions. Tr. 867:1-4 (Winter). Consequently, the agency uses that year as the end point of its HSA project. TFAS also contains data regarding the IIM funds that are pooled for investment purposes. Tr. 872:3-12 (Winter). Though transactional information has been recorded in many different ways over the life of the trust, the HSA project focuses primarily on information contained in paper records, and later in IRMS and TFAS. iii. Land Records Information System (LRIS) and Trust Asset and Accounting Management System (TAAMS) The Land Records Information System (LRIS) contains historical land ownership information that is used in Interior’s historical accounting project, National Opinion Research Center (NORC) Analysis of LRIS Tract History Reports, AR-405 at 5, but at least some agencies determined that the ownership information in LRIS was out of date, and that LRIS data could not be relied upon for distributing trust funds. Tr. 1317:17-1321:25 (Redthunder). Interior began converting from LRIS to a “more robust land title system” called the Trust Asset and Accounting Management System (TAAMS) in 2005. Tr. 69:6-13 (Cason). As of September 2007, all BIA agency offices had at least partially converted to TAAMS. Tr. 868:6-11 (Winter). TAAMS contains two different systems: realty and title. Tr. 903:15-904:2 (Winter). The TAAMS title system produces invoices for lease holders, who then send their payments along with the invoice to a lockbox. Tr. 868:6-11 (Winter). The invoice and lockbox procedures are very new: they accompanied the TAAMS title roll-out in 2005. Tr. 903:15-904:12 (Winter). It is not clear whether all agencies have fully implemented the TAAMS title system at this date, but those that have now have an accounts receivable system for the first time in the history of the IIM trust. Tr. 902:23-904:12 (Winter). This accounts receivable system is referred to as the Trust Funds Receivable System (TFR), and permits the agency to follow up on leases to be sure that payments are made. Tr. 902:23-903:14 (Winter). Prior to the TAAMS conversion, using something like an honor system, Interior simply relied on lease holders to submit accurate, timely payments. iv. Other data systems Other systems, while not prominently featured in the historical accounting project, contribute to the functioning of the IIM trust. For example, the Treasury Department has systems that record IIM trust funds and balances. The Minerals Management Service within Interior (MMS) has a data system for processing payments received from lessees and sending payment and lease-level information (not broken down by beneficiary or allotment) to Treasury and the Bureau of Indian Affairs. BIA processes oil and gas monies from MMS in a system called the Royalty Distribution Reporting System (RDRS) in order to allocate oil and gas revenue among land owners. Tr. 919:6-920:1 (Winter). BIA relied on Osage computer systems to process quarterly annuity payments and some oil and gas income. Tr. 1736:2-32 (Infield). In the past, BIA used several different systems to distribute trust revenues, such as the ownership module within IRMS, the GLAD distribution system in the Great Plains Agency, and the MAD system used in the Aberdeen and Standing Rock Agencies. Tr. 1756:12-18 (Infield); Tr. 1321:20-1322:2 (Redthunder). I.1.B. Inconsistencies between agency and regional offices The BIA is divided into twelve regions, under which a total of 93 agency offices operate. Feb. 2002 OHTA Accounting Records Conference Materials at 213, AR-058. As mentioned above, the use of electronic records systems varied among the twelve regional offices, and even among agency offices within a single region. Former Special Trustee Paul Homan recalled a contractor describing the information systems used by the regional offices as “twelve islands of information without a ferry in between.” Tr. 1552:21-23 (Homan). The regional offices, and agencies within those regions, each had their own ways of using -- or not using -- IRMS. Most agencies used the MMS system to process oil, gas, and mineral payments, but Osage was again an exception to this rule. Tr. 1737:4-24 (Infield). I.1.C. Destruction of documents Until very recently, Interior had no system for the preservation of trust records. Over the life of the IIM trust, trust documents were routinely destroyed in accordance with record retention schedules at National Archives and Records Administration (NARA) centers. Tr. 2054:10-2056:18 (Christie). Most documents and data were retained for only six years pursuant to standard NARA policy, until 1995, when Interior instructed NARA to cease destroying trust records according to the six year schedule. Trust records from 1989 forward have not been subject to the document destruction schedules. The loss of trust records before 1989 has been mitigated -- to what extent is a matter of dispute -- by the fact that, in many cases, Interior created several copies of IIM records. See Defendants’ Corrected Proposed Findings of Fact and Conclusions of Law [3459-1] at 8 (table listing types of IIM transactional documents, number of copies produced, and locations stored). Interior’s expert contends that the existence of multiple copies of trust records stored in different locations renders the loss of trust records immaterial, as the loss or destruction rarely impacted all copies of a given document. Tr. 1264:8-11, Tr. 1282:3-7 (Angel). Defendants have located and centralized 43 miles of Indian records potentially relevant to the accounting at the National Archives and the American Indian Records Repository (AIRR) in Lenexa, Kansas, Tr. 1198:14-21 (Angel), and they have access to another 10,000 cubic feet of MMS and United States Geological Survey (USGS) records potentially useful to those conducting the HSA accounting. Tr. 1181:5-10 (Angel). Problems related to the disorganized or poor condition of records were noted early in this litigation and have been addressed by defendants’ contractors at the AIRR. Interior’s Office of Historical Trust Accounting (OHTA) continues to work with NARA to collect data for the AIRR. Defendants have been surprised to discover that the records needed to perform the accounting are indeed available and accessible, Interior’s 2007 HSA Plan Part 1, AR-565 at 33-02-08, and they contend that plaintiffs’ complaints about the existence and condition of trust records are essentially based on outdated information. Plaintiffs acknowledge, as they must, that an enormous volume of records has been collected at Lenexa, but their response, essentially, is “Compared to what?” They suggest that the volume of records is meaningless without some assessment of the total number of trust documents produced over the history of the trust. They point out that in 1998, the BIA estimated that a total of approximately 1.425 billion pages of IIM trust records existed. (Government experts later determined that that number had been overestimated by some 1.205 billion pages, partially because they had not adequately accounted for destroyed documents. DOI memo stamped June 2002, AR-80 at 54-21-32; Tr. 1266:13-1268:15 (Angel)). Plaintiffs’ witnesses testified that a great deal of data was missing from the IRMS database, Tr. 1746:9-1747:14 (Infield), that the manual entry of information into BIA’s Royalty Distribution and Reporting System (RDRS) was unreliable, Tr. 1749:24-1751:13 (Infield), and that some copies of documents necessary for determining the quantity of resources exploited from allotted land -- such as oil and timber -- were never recorded electronically and had been routinely destroyed. Tr. 2043:19-2049:17 (Christie). It is likely that, in some cases, all copies of timber scaling tickets, distribution settlement worksheets, and run tickets were destroyed, since relatively few copies of these documents were created. Tr. 2044:3-2045:6 (Christie). Joe Christie, a 27-year veteran of the Interior Department, opined that distribution worksheets are essential if the historical accounting project seeks to “verify who was considered the owners [of an allotment], the amount of funds that were collected, and who [the funds were] distributed to. It’s the only document, by the way, that does that.” Tr. 2042:5-8 (Christie). Plaintiffs’ witness Mona Infield, an Interior employee, also indicated that lease information was frequently destroyed. Tr. 1750:18-1751:13 (Infield). Before 1990, cleared disbursement checks that had been issued to IIM beneficiaries were destroyed by Treasury. Motions Hearing Tr. 171:14-172:4 (Locks, 11/24/98) (no checks available pre-dating Oct. 1990); Oct. 19, 1994 Letter from Sandra Chambers (FMS) to James Parris (DOI), PX-3340 (records prior to 1987 not available); Nov. 22, 1994 Letter from Leo Warring (FMS) to Jim Parris (DOI), PX-3342 (records unavailable prior to 1988). Interior has an abysmal record of failing to prioritize the maintenance and preservation of trust documents. Many court opinions, audits, and congressional committee reports have catalogued that record. As late as 1995, the Deputy Commissioner of Indian Affairs discovered that agencies were destroying “financial and lease documents . . . required in the reconciliation process.” 3/07/95 Memo from Hilda Manuel to Area Directors and Agency Superintendents, PX-0350. After decades of neglect, it is impossible to imagine that all documents necessary to perform a complete historical accounting are presently accessible to Interior. Nevertheless, the agency has made an impressive (and very expensive) effort in recent years to find, scan, and preserve whatever documents still exist. If the government has failed to prove that all documents necessary for the completion of its historical accounting have been or will be 6 found, plaintiffs have also failed to establish that the problem of lost or destroyed documents renders the historical accounting project entirely pointless. I.1.D. Predecessor historical accounting plans The 1994 Act did not tell the Interior Department how, or how thoroughly, to perform the historical accounting that it mandated. Associate Deputy Secretary James Cason said that because the agency had no model to follow, it had to engage in an “iterative process where we learn and redesign, learn and redesign.” Tr. 65:2-9 (Cason). The Department’s initial compliance effort was the work of the “Tiger Team;” a group of Officer of Trust Fund Management (OTFM), BIA, and MMS employees assembled soon after the passage of the 1994 Act to study trust accounting issues. Tr. 1753:16­1754:4 (Infield). In August 1995, the team issued a report entitled “IIM Related Systems Improvement Project: The Tiger Team,” which was to be implemented by the end of 1996. August 1995 Draft Tiger Team Report, PX-607 at 157. The Tiger Team report noted uncertainty about the accuracy of data entered in 6Defendants’ contractors were still looking for 35,000 outof 80,000 requested documents at the date of trial, Tr. 466:4-11(Herman), Assistant Deputy Secretary Haspel acknowledged that,“until we open all those boxes,” the extent to which historicaltrust records may be missing remains unknown, Tr. 1138:4-11(Haspel), and defendants’ historian Edward Angel testified thathe could not say whether there are sufficient records to accountfor any particular year of the IIM trust. Tr. 1261:5-10 (Angel). IRMS, and it criticized the internal controls and inadequate automated systems relied upon in managing the IIM trust. Id. at 2-8. The Team’s efforts appear to have focused more on reforms going forward than on historical accounting. Two years after the issuance of the Tiger Team report, the first Special Trustee for American Indians Paul Homan finalized a Strategic Plan to Implement the Reforms Required by the American Indian Trust Fund Management Reform Act of 1994, PX­ 615. Homan submitted the report in April 1997 to the Secretary, who incorporated many of its recommendations into the 1998 High Level Implementation Plan (HLIP) for the Trust Management Improvement Project (TMIP), filed with the Court in August 1998. High Level Implementation Plan, PX-4154. The HLIP identified thirteen projects Interior planned to undertake to improve trust management and to correct shortcomings in data maintenance, records management, training, policies and procedures, system weaknesses and internal controls. Id. at 3-8. The 1998 HLIP called for the completion of the Internal Control Sub-project by June 1999, and of the whole HLIP within three years. As with the Tiger Team report, this project was focused on trust reform, though it devoted significant attention to historical weaknesses, particularly in the area of internal controls. A revised HLIP was filed in this litigation on March 1, 2000 [Dkt. 438]. The expected dates of completion were extended, but many of the objectives remained the same, such as the intention of reviewing trust records to determine accurate title and resource management information. The updated HLIP remained focused on documenting and rectifying historical weaknesses, inconsistencies, and failures in the Department’s trust management operations. In November 2001, after the Court of Appeals affirmed Judge Lamberth’s finding that defendants were in breach of their trust duties, the government filed an additional plan with the court titled “DOI Trust Reform: Interim Report and Roadmap for TAAMS and BIA Data Cleanup” [Dkt. 990]. That report, prepared by Electronic Data Systems, Inc., identified significant problems with missing data, inconsistent records, and conflicting information, and it recommended reforms that were to be completed within three years. Id. at 21-30. On July 2, 2002, after requests from House and Senate Appropriations Committees made during the FY 2001 and 2002 budgeting processes, Interior submitted a report to Congress that detailed its historical accounting plan. Report to Congress on the Historical Accounting of Individual Indian Money Accounts, AR-561; Tr. 58:5-9 (Cason). The report informed Congress that Interior planned to conduct a transaction-by-transaction reconciliation of all funds in the IIM accounts through December 31, 2000, and that it expected the historical accounting project to cost $2.425 billion to complete. AR-561 at 25-02-15­ 16. The report indicated that the historical accounting would result in transaction histories for all accounts that resembled the account reconstructions done for the named plaintiffs in this case. Id. at 25-02-30. The historical statements of account envisioned in that report were to include ownership information for each account: allotment number and ownership interest. Id. at 25-02-87. Statements would be provided for the estates of deceased beneficiaries whose IIM accounts and allotment interests had been probated, for closed accounts, and for transactions dating back apparently from the inception of the trust (no beginning date is listed in the report). Id. at 25-02-40. The historical accounting process would involve the reconciliation of leases, land ownership interests, and the collection and disbursement of all IIM funds. Id. at 25-02-92. That report to Congress was prepared in response to House Appropriations Committee concerns about the costs associated with the historical accounting. Id. at 25-02-21. After its submission, the House Committee on Resources addressed a December 9, 2002 letter to then Interior Secretary Gale Norton, describing the report as “troubling in several areas,” and requesting that the Secretary “promptly consider ways to reduce the costs and the length of time necessary for an accounting . . . [including] alternative accounting methods.” AR-184. A month later, on January 6, 2003, the Interior Department issued an Historical Accounting Plan for Individual Indian Money Accounts (the “2003 HSA Plan”), PX-507, that was strikingly different from the accounting project envisioned in the 2002 Report to Congress. The scope of the accounting was narrowed to exclude accounts closed before October 25, 1994, transactions occurring after December 31, 2000, and transactions in closed accounts or in the accounts of deceased beneficiaries. Id. at 8-9. While the accounting described to Congress in July 2002 would have cost approximately $2.4 billion, the effort described in the 2003 HSA Plan was projected to cost $335 million over five years. Id. at 1. The implementation of the 2003 Plan began, though not quite according to plan, with the commencement of the Litigation Support Accounting Project in the fall of 2003. Tr. 95:20-98:2 (Cason). The 2003 Plan -- with its abandonment of a total transaction-by-transaction approach to land-based accounts in favor of a mixture of transaction-by-transaction and statistical sampling reconciliations -- reflected Interior’s acquiescence to the House Committee’s request that it “consider all available options regarding the use of alternative accounting methods.” AR-184; 2003 Historical Accounting Plan, PX-507 at 17-18. Specifically, the Department indicated that, within the population of transactions occurring during or after 1985, all transactions over $5,000 would be reconciled, 10% of transactions between $500 - $5,000 would be reconciled, and .31% of transactions under $500 would be reconciled. PX-507 at 63. In total, 160,000 transactions less than $5,000 would be sampled, and the sampled transactions would be pulled from every agency within Interior. For transactions occurring before 1985 (the “paper ledger era”), all transactions of $5,000 or greater would be reconciled, and transactions below $5,000 would be sampled. The sampling plan for these transactions was not yet fully developed at the time of the 2003 HSA Plan’s publication. Id. at 69. The government’s 2003 estimate was that this stripped-down historical accounting would be completed in about four years: Judgment and Per Capita accounts were to be reconciled by June 30, 2004, land-based accounts by September 30, 2006, IIM systems tests by September 30, 2006, and Special Deposit Accounts by December 31, 2006.7 Id. at 32. The historical statements of account prepared for IIM account holders pursuant to the 2003 HSA Plan were to include Interior’s assessment of the accuracy of the account transaction history and sufficient information for IIM 7 All IIM funds in Special Deposit Accounts -- temporaryholding accounts used to deposit funds before they are creditedto specific accounts -- were to be transferred into theappropriate IIM accounts as part of the historical accountingprocess. beneficiaries to “ascertain whether Interior has faithfully carried out its IIM Trust Fund duties.” Id. at 4. On the same day it issued its 2003 HSA Plan, Interior filed a court-ordered Fiduciary Obligations Compliance Plan, PX­ 508. The purpose of this plan was to bring the Department into compliance with the 1994 Act. This plan recognized shortcomings in the HLIP and acknowledged the importance of verifying the accuracy of account balances, a step the government noted was crucial “no matter how carefully future transactions may be recorded.” Id. at 5. Defendants next issued a Comprehensive Trust Management Plan, which conceded failures to implement prior trust reform plans and announced a new strategic plan for doing so in the future [Dkt. 2050]. The government filed its “To-Be” Trust Business Model and its Fiduciary Trust Model on March 15, 2005 [Dkt. 2882, attachments 2, 3]. According to the Special Trustee, implementation of the Fiduciary Trust Model would ensure future compliance with fiduciary obligations owed to IIM beneficiaries. I.1.E. IIM-related projects before the 2007 plan i. Paragraph 19 project In 1999, after Secretaries Babbitt and Rubin and Director Gover had been found in contempt for their non­compliance with orders of this Court, Cobell II, 37 F. Supp. 2d 6, 17 (D.D.C. 1999), the government hired Arthur Andersen to help both the Treasury and Interior Departments comply with Paragraph 19 of the First Order for Production of Information [Dkt. 16]. Under Paragraph 19, defendants were to produce all “documents, records or tangible things which embody, refer to, or relate to the IIM accounts of the named Plaintiffs or their predecessors in interest.” Id. at ¶19. Treasury produced over 2,000 documents pursuant to Paragraph 19, and Andersen ultimately concluded that Treasury’s Paragraph 19 search had been “thorough [and] well-executed [at or above] industry practices.” Trial 1.5 Tr. 51:15-52:9 (Brunner 6/6/03). Interior searched approximately 80 facilities for documents responsive to Paragraph 19 and produced around 160,000 documents. Trial 1.5 Tr. 54:6-55:8, Tr. 66:5-7 (Brunner 6/6/03). A total of 37 accounts were analyzed during this project. The documents collected dated back to 1914. Interior’s 2007 HSA Plan Part 2, AR-566 at 33-03-12. The records produced in response to Paragraph 19 were reviewed by Ernst & Young partner Joseph Rosenbaum in 2001. Trial 1.5 Tr. 53:3-10 (Rosenbaum 6/9/03). Rosenbaum’s report analyzed a virtual ledger of transactions reflecting the documents Interior had collected in response to Paragraph 19. Rosenbaum determined that the documents necessary for assembling transaction histories for the named plaintiffs and their predecessors were available, and that TFAS balances from December 31, 2000 were sufficiently supported by supplemental documentation. Trial 1.5 Tr. 56:15-22 (Rosenbaum 6/9/03). Supporting documentation was discovered for 86 percent of the 12,617 transactions reviewed, representing 93 percent of the total dollar value of those transactions, or approximately $1.12 million. Trial 1.5 Tr. 75:5-76:19, Tr. 77:6-19 (Rosenbaum 6/9/03). Only small variances were noted. The total cost of the project, however, was around $20 million. AR-566 at 33-03-12. Interior concluded that, although the documents necessary to complete adequate accountings are available, the accounting process is extremely expensive, often dwarfing the dollar amounts reflected in beneficiaries’ accounts. Rosenbaum also performed an “expected versus actual” analysis by comparing information about leases, transactions, and ownership interests. Trial 1.5 Tr. 54:17-55:10 (Rosenbaum 6/9/03). He identified and analyzed the majority of leases associated with the named plaintiffs that related to farming and oil and gas extraction, Trial 1.5 Tr. 57:20-58:3 (Rosenbaum 6/9/03), finding a net of only $32.04 in unexplained differences between transaction ledger entries and leases. During the October 2007 bench trial, Associate Deputy Secretary James Cason testified that Interior understood the results of the Paragraph 19 project as indicating that, although there were errors in the accounts, the errors were relatively few, the errors tended to be small, and the errors were on both sides of the ledger. Tr. 62:12-21 (Cason). The Paragraph 19 project ignored direct pay transactions, Trial 1.5 Tr. 44:13-23 (Rosenbaum), escheated interests, Trial 1.5 Tr. 7:18-23 (Rosenbaum), and other types of transactions. The analysis of disbursements typically ended when funds were disbursed from IIM accounts, not when beneficiaries received the funds. Trial 1.5 Tr. 84:25-85:3 (Rosenbaum 6/10/03). ii. Mass cancellation project Historically, Treasury checks were of unlimited payability -- in other words, there were no temporal limits on when they could be cashed. That changed when Congress passed the Competitive Equality Banking Act of 1987 (CEBA), Pub. L. No. 100-86, 101 Stat. 552 (1987), which provided that, as of October 1, 1989, Treasury checks would be of “limited payability” and could be cashed for only one year from the date of issuance. Treasury Bulletin No. 90-03, DX-231 at 1, 18; Tr. 323:17-324:4 (Ramirez). CEBA also mandated the “mass cancellation” of all checks issued by Treasury that were at least one year old by April 1, 1991. DX-231 at 1, 3; Tr. 323:17-324:4, Tr. 325:13-20 (Ramirez). At that time, some 10 million Treasury checks from as early as 1954 remained outstanding; approximately 60,000 of these checks were IIM checks with a combined value of approximately $1.9 million. Mass Cancellation Project, Analysis of Treasury Listing, DX-225 at 8; Tr. 324:8-328:16 (Ramirez); BIA Office of Trust Funds Management (OTFM) Instructions on Completing Mass Cancellation Project, DX-217 at 1-2. Certain provisions of CEBA were problematic as applied to IIM trust funds. For example, CEBA broadly prohibited re- crediting to agencies the funds associated with the cancelled checks. DX-231 at 4. Generally, an agency will deposit funds at Treasury to cover the payments of checks issued pursuant to its authority. Under CEBA, all funds that had been deposited to cover mass cancelled checks were to remain in the Treasury general account rather than revert back to the agencies. At the time, legislators apparently were not focused on the fact that some funds -- like IIM and tribal trust funds -- were not tax dollars, but were monies that were only being held by Treasury after their collection and before their disbursement. To address this problem, BIA put Katherine Ramirez in charge of a “Mass Cancellation Project,” Tr. 323:11-15, Tr. 332:8-20 (Ramirez), whose mission was to identify IIM checks that had been mass cancelled and to re-credit their amounts to the proper accounts. Tr. 323:11-15, Tr. 331:8-14 (Ramirez); July 30, 1992 Letter from Mary Sandoval to Donald Gray, DX-207 at 1. Interior had to obtain a special appropriation for the purpose of re-crediting IIM accounts, and, in 1992, Congress appropriated $3 million for this purpose. See Department of the Interior and Related Agencies Appropriations Act, 1993, Pub. L. No. 102-381, 106 Stat. 1374, 1391 (1992). The mass cancellation project discovered that approximately 22,000 of the approximately 60,000 mass-cancelled IIM checks were zero dollar instruments,8 and attempted to trace the 38,554 remaining checks to specific IIM accounts. Tr. 349:3-350:22, Tr. 352:18-353:4 (Ramirez). BIA traced $616,736.31 -- almost a third of the value of the 38,554 non-zero dollar, mass cancelled IIM checks -- to specific accounts or to voided checks, 1993 OTFM CEBA Report, DX-221 at 2; Tr. 353:5-8, Tr. 354:11-356:7 (Ramirez), leaving approximately $1.3 million in mass cancelled IIM checks unresolved. DX-221 at 2. Of that $616,726.31, BIA re-credited approximately $278,000 to IIM accounts; $338,000 was apparently attributable to voided checks. Tr. 356:12-357:9 (Ramirez). At trial, no evidence was presented as to Interior’s post-1993 progress in resolving the remaining approximately $1.3 million worth of mass cancelled IIM checks. Ramirez testified 8Ramirez testified that the zero dollar instruments were not checks issued to beneficiaries in the amount of zero dollars, butcheck stock that had been voided, for any number of reasons.Tr. 349:3-351:15 (Ramirez). For example, the check stock mayhave contained typos, or may have been destroyed in the printer,or could not be used because it did not contain the limited payability instruction (“void after one year”). In those cases,Treasury would not be informed that a particular check number wasused, so Treasury officials would input a zero dollar amount.Tr. 349:6-350:24 (Ramirez). that Interior continues to maintain a fund totaling approximately $500,000 to pay claims on mass cancelled checks, although no claims against the fund have been made recently. Tr. 364:4-365:11 (Ramirez). It appears that reimbursement for mass cancelled checks occurs now only if an IIM beneficiary or BIA agency presents a mass-cancelled trust check to the agency for payment, and that unless and until such a check is presented, the trust funds associated with the mass cancelled check are not disbursed. Examining unrestored funds from mass cancelled checks is not part of Interior’s 2007 HSA project. iii. 20-year tribal reconciliation project The original concept of the reconciliation project was to reconcile both the IIM trust and the tribal trust. After a preliminary assessment revealed significant problems relating to missing IIM records, however, Arthur Andersen informed Interior that reconciling the IIM trust would be infeasible under the $12 million contract awarded by the government. May 1996 GAO Report to the Senate Committee on Indian Affairs Re: The Tribal Reconciliation Project, PX-710 at 3. Instead, Arthur Andersen endeavored to reconcile the tribal trust between the dates of July 1, 1972 and September 29, 1992. Tr. 2074:9-10 (Christie). The 20-year reconciliation project began in December 1992 and was terminated before completion in 1995, at a total cost of approximately $21 million. Tr. 2073:8-17, Tr. 2084:6-20 (Christie); PX-710 at 2. The tribal reconciliation project was not an audit, but a contract governed by “agreed upon procedures” -- in other words, a contract in which the client defines the scope and nature of the project. Interior’s tribal reconciliation project manager Joe Christie testified that the project accomplished its goal of reconciling tribal trust fund transactions “to the extent possible . . . within the time frame we were given.” Tr. 2076:14-19 (Christie). However, because they encountered “lots of missing documents,” Tr. 2076:21 (Christie), they were typically unable to put together reconcilable packages for Arthur Andersen’s analysis that met the highest standard set forth in the agreed-upon procedures (referred to as the “C” standard). Indeed, the “vast majority” of the packages assembled by BIA workers did not meet this standard. Tr. 2076:21-2080:2 (Christie). When the reconciliation packages were ultimately presented to the relevant tribes, the tribes were told that they could accept or reject packages reconciled at levels other than the “C” level. Tr. 2080:3-9 (Christie). The results of the reconciliation project were mixed and inconclusive. Approximately 320 reconciliation packages were provided to tribes after Interior terminated the project. At least some supporting documentation was located for 86% of the non-investment transactions reviewed, but the government could not verify the remaining 14% of those transactions -­representing $2.4 billion -- and abandoned an effort to test investment transactions after being stymied by instances of missing records. PX-710 at 5-6. The 1996 GAO Report on the Tribal Trust Fund Reconciliation Project reveals that BIA failed to certify that the reconciliation was performed in compliance with the contract’s agreed upon procedures (a modest certification standard). PX-710 at 3. Notably, the reconciliation did not “address the completeness of records,” nor calculate “receipts and disbursements that should have been recorded,” and it revealed that “BIA did not know the [complete] universe of leases.” Id. at 5-6. This project did not involve an analysis of the IIM trust beyond Arthur Andersen’s initial conclusion that the problem of missing documentation within the universe of IIM trust records was more severe than within the universe of tribal trust records. iv. TIME project In 2000, Interior hired DataCom Sciences, Inc., to “examine the accuracy of the current ownership document information in LRIS by comparing mandatory elements.” DataCom Time Project Report, PX-4352 at 4. This became known as the “TIME project.” DataCom received a random sample of 93 of the 239,311 tracts within the LRIS system at the Rocky Mountain Land Title Records Office, PX-4352 at 2, and analyzed 541 documents reflecting ownership information stored in LRIS at the time of the project. Operating on the assumption that the paper records used in the comparative analysis were accurate, Contempt II Tr. 3363:13-3364:2 (Nessi 2/01/02), DataCom determined that over 30% of the 541 LRIS documents analyzed contained errors of different types. PX-4352 at 2. Interior later hired NORC to conduct a pilot study of the accuracy of probate entries contained in the Tract History Reports housed within LRIS. NORC Analysis of LRIS Tract History Reports, AR-405. After reviewing 99 probated IIM accounts, NORC reported no material errors and no evidence of an error rate as high as 33%, but the NORC report was narrower in scope than the TIME report and indicated that it was not to be used to extrapolate an “error rate” among probate entries. AR-405 at 4. The differences between the NORC and DataCom findings appear attributable -- at least in part -- to different definitions of “error”. NORC criticizes DataCom for over- identifying errors, because DataCom’s definition of error was tied to ‘conformance to requirements’ rather than ‘fitness for use.’ A misspelling would be a ‘conformance to requirements’ error, even if it did not affect the entry’s fitness for use in the administration of the trust. NORC Analysis of LRIS Tract History Reports, AR-405 at 50-02-06-08; Tr. 1079:3-9 (Scheuren). The DataCom report indicates only that “inconsistencies” between hard copy documents and LRIS entries were treated as errors. The NORC pilot study operated with a more inquiring definition of “error”. For example, documents missing from LRIS due to the probate backlog were not considered errors, nor were incorrect IIM numbers. AR-405 at 50-2-6. Neither definition of “error” can be judged “right” or “wrong.” They are merely more and less sensitive to mistakes within the LRIS system. The error rate calculations of both approaches can be useful, if for different purposes. The record will not support plaintiffs’ submission that the high error rate found by DataCom is indicative of material errors throughout current LRIS ownership information. It is reasonable to conclude, however, that further analysis of LRIS ownership records should precede total reliance on that database for purposes of auditing or accounting for ownership interests. v. Straw Man project Dr. Scheuren drafted a document concerning a proposed “Straw Man” approach to the historical accounting during his early days as an Interior contractor, when he was “just learning about this work.” Tr. 1080:4-21 (Scheuren); Undated document titled “Adaptive Testing Approach for Phase 1 Straw Man Design,” AR-167; Undated Scheuren document titled “Straw Man Pretest,” AR-170; see also Tr. 1080:4-12 (Scheuren); Exchange containing 07/08/02 email from Scheuren to Edwards mentioning straw man proposal, AR-304 at 14-02-04. The “Straw Man Pretest” was prepared after meetings between NORC and Interior and included several questions for Interior to consider as it planned its data gathering efforts. AR-170; AR-304. The documents describing the “Straw Man Pretest” focus on the benefits of adaptive sampling -- an approach that is heavily utilized in both the 2003 and 2007 HSA plans. In one document, the government notes that “[t]he main benefit emphasized in an adaptive approach is obviously reduced expense,” and identifies other benefits such as scalability, speed, suitability for targeting of efforts, likelihood that results can be reported as “a series of successes,” use of pilots to benefit from lessons learned by prior pilots, and increased accuracy. AR-167 at 57-29-01-02. Adaptive sampling “links up nicely with the legitimate payment of a cash settlement, since inherently an adaptive approach admits error, even though it may well manage to keep it less than would have been the case in a full accounting.” Id. at 57-29-02. The “Straw Man” approach appears to reflect Interior’s initial reconsideration of the 100% transaction-by­transaction approach proposed in the 2002 Report to Congress: “the adaptive strategy . . . is completely compatible with the . . . [July 2, 2002] Historical Accounting Report. The only difference is that we do not necessarily have to test every transaction.” Id. vi. Recent audits of the IIM trust To the court’s knowledge, there has never been an unqualified independent audit of the IIM trust. “Qualified” audits are issued when the auditor is unable to comply with generally accepted auditing standards. In the case of the IIM trust, auditors have issued qualified audits after discovering cash balances that conflicted with those reported by the U.S. Treasury, internal control deficiencies, records management problems, and irreconcilable ledgers, among other things. See, e.g., AR-343; AR-347; AR-350; AR-352; AR-355; AR-369; AR-374; PX­575 at 3, 10-12. The first IIM trust audit performed by an independent auditor was the Arthur Andersen audit conducted in 1988. PX-575 at 3. That audit documented internal controls weaknesses attributable to inadequate training, lack of experienced supervisors, understaffing, and out-of-date accounting manuals, and concluded that the “accounting systems [were] unreliable.” PX-575 at 10-12. These concerns have been reflected in more recent audits as well. See PX-695; AR-633; AR­377; AR-374. The most recent independent audit -- prepared by KPMG for the year ending on September 30, 2006 -- noted that OST relied upon unreliable BIA data and unresolved financial reporting from prior periods, that OST’s processing of trust data relies upon the accuracy of information from BIA, MMS, and other bureaus and offices, and that “current management is burdened with the ongoing impact of decades of accumulated discrepancies in the accounting records.” AR-343 at 60-02-34. I.1.F. Establishment of the AIRR in Lenexa, Kansas In 2004, the government opened the American Indian Records Repository (AIRR) in an underground limestone mine in Lenexa, Kansas. AR-563 at 11-12; Tr. 378:19-379:17 (Ramirez). The AIRR is within NARA’s Federal Records Center, which is one of many government offices and private businesses occupying space in the 90 acre mine. Tr. 368:3-8; Tr. 370:13-14 (Ramirez). Retired Indian records are sent to Lenexa for storage and as a resource for research. The government now reports that 165,825 boxes of Indian records are stored at the AIRR (not all of which contain IIM trust records). OTR Activity Report for December 2007 (Jan. 15, 2008) [Dkt. 3479]. Two Interior offices have outposts at the AIRR: the Office of Historical Trust Accounting (OHTA) and the Office of Trust Records (OTR). Tr. 367:1-8 (Ramirez). OTR is within the Office of the Special Trustee for American Indians (OST). The AIRR is under the overall management of OTR, but OHTA requires access to trust records that are stored on-site, so OHTA sublets space from OTR and employs on-site contractors who search for (“searchers”) and code (“coders”) documents. Id. OTR also has an off-site annex facility, where boxes of records destined for the AIRR are initially received from BIA, OST, and regional offices across the United States. These sealed boxes are indexed, inventoried, and labeled according to strict procedures before they are sent into the cave and officially transferred from the custody of the DOI into the custody of NARA. Tr. 372:12-373:7 (Ramirez). Kathy Ramirez -- the former manager of the mass cancellation project -- manages OHTA’s contract staff at the AIRR. Tr. 366:25-367:1 (Ramirez). Two primary systems are used for indexing, storing, searching for, coding, and scanning specific documents at the AIRR: the Box Index Search System (BISS) and Account Reconciliation Tool (ART). After they have been delivered and labeled, the boxes’ numbers and indexing information are entered into the BISS by annex contractors before the records are transferred to NARA for storage. Tr. 372:18-373:7, Tr. 377:5-378:9, Tr. 385:19-386:19 (Ramirez). OHTA’s accounting contractors (Clifton Gunderson) use the BISS system to identify records that will be potentially useful in conducting historical accounting work. Tr. 381:3-25 (Ramirez). After identifying potentially relevant boxes on the BISS, these “searchers” submit requests to NARA, and NARA delivers the requested boxes to OHTA’s searchers. Tr. 382:1-383:9 (Ramirez). Once the searchers identify relevant documents within the requested boxes, another group of contractors (the “coders”) images and codes the responsive documents and loads them into the ART system where they are accessible to OHTA contractors performing historical accounting work. Tr. 383:20-384:11, Tr. 385:3-11, Tr. 387:2-389:15 (Ramirez). Quality control measures are observed throughout the process. It is evident from the photographs presented and testimony given at trial that the AIRR is a state of the art, climate-controlled, organized, and sizable facility suitable to the storage and research obligations of the Interior Department. Storage bays containing Indian trust records were constructed in accordance with the highest standards, the facility maintains low temperature and low humidity, and particulate matter and ultraviolet light are controlled. Tr. 370:17-19 (Ramirez). While the facility reflects a significant improvement in the conditions and treatment of the records that have survived the cramped, disorganized, flooded, rat-infested storage facilities in which some of them were found, it is unclear from the evidence presented at trial what percentage of total IIM records that should have been maintained over the life of the trust have actually been recovered and shipped to the AIRR. Document requests still present vexing challenges for the considerable number of “searchers” working in the AIRR. At the time of trial, of the 80,000 requests FTI has submitted to AIRR in connection with its DCV work, approximately 35,000 requests remained pending, and some requests can take over a year to fill. Tr. 505:3-14 (Herman). As recently as June 2007, the government estimated that it could cost up to $37.4 million to search the AIRR for documents related to plaintiffs’ request for the trust records of less than one hundred specific beneficiaries. [Dkt. 3340] at 3-4. The development of the AIRR has streamlined the analysis of trust records to some extent, but difficulties in locating documents and the inability to certify that all relevant documents have been consolidated in the facility prevent a finding that the records management problems chronicled in earlier opinions have been fully rectified. I.2. The 2007 Historical Accounting Plan The historical accounting plan announced by Interior in May 2007 -- the 2007 HSA Plan -- is dramatically different from the far more ambitious plan announced four years earlier. It reflects Interior’s decisions, sometimes made for cost reasons and sometimes because of its legal interpretation of the statute, to rely much more heavily on sampling and statistics. Interior will reconcile only a small sample of IIM transactions (which it believes will yield statistically significant results), test its existing electronic data for completeness, test the process by which revenues collected from allotments are transferred into IIM accounts (again on a sample basis), and then begin issuing historical statements of account to IIM holders -- with Interior’s assurance that they are correct. I.2.A. Major changes to the Historical Accounting Project reflected in the 2007 HSA Plan The Interior Department’s 2007 HSA Plan “builds upon and replaces” the 2003 HSA Plan. Interior’s 2007 Historical Accounting Plan, AR-566 at 33-03-03. Many significant changes are reflected in the Department’s newest plan. i. Cost and Schedule When it adopted the 2003 HSA Plan, Interior estimated that the IIM historical accounting project would cost $335 million to perform and would be completed within five years. Between 2003 - 2007, however, not only did Interior receive only $127.1 million in appropriations for its IIM historical accounting work, but it also discovered that the accounting process it had envisioned would be both more costly and more time-consuming than it had anticipated. Tr. 67:24-68:7 (Cason). Even with the dramatic reduction in the number of land-based transactions that will be sampled under the 2007 Plan, Interior expects that the 2007 Plan work will not be completed until the end of 2011. That estimate is contingent upon congressional appropriations for IIM historical accounting totaling $144 between 2007 and 2011. ii. Transactions and Accounts to be Reconciled The 2003 plan relied on statistical sampling of lower value transactions in land-based accounts, but the number of transactions to be reconciled under the new plan is considerably smaller. A chart may help to illustrate the differences: 2003 HSA Plan 2007 HSA Plan Reconciliations of transactions from land-based accounts post-1985(the “ElectronicLedger Era”) < transactions $ $5,000: 100%transaction-bytransaction approach(73,000transactions)< transactions btw $500-$5,000:80,000transactions sampled< transactions < $500: 80,000transactions sampled samples must bedrawn from everyagency < projected total . 233,000transactions reconciled – AR-566 at 33-03-22 - 33-03-23 < transactions $ $100,000: 100%transaction-bytransaction approach(2,099transactions)< transactions < $100,000: 4,500transactions sampled all agenciessubject tosampling, but norequirement thatsamples be drawnfrom each agency < projected total =6,599transactions reconciled – AR-566 at 33-03-22 - 33-03-23 Reconciliations of transactions from land-based accounts pre-1985(the “Paper LedgerEra”) < transactions $ $5,000: 100%transaction-bytransaction approach(100,000transactions)< transactions < $5,000: 160,000transactions sampled < projected total . 260,000transactions reconciled AR-566 at 33-03-22 < transactions of all dollar amounts subject tostatistical sampling; unknownnumber of sampledtransactions < projected total =unknown AR-566 at 33-03-23 ­33-03-24 Reconciliations of 100% transaction-by­ 100% transaction-by­ transactions from transaction approach transaction approach Judgment and Per 86% complete; Capita accounts further reconciliations deferred indefinitely < projected total . all transactions in 96,823 accountsreconciled AR-566 at 33-03-24 < projected total =unknown (reconciliationsof all transactions in at least 83,226accounts complete)AR-566 at 33-03-24 The bulk of the reconciliation work that remains to be done under Interior’s 2007 HSA plan involves pre-1985 transactions in land-based accounts. Because most of the materials necessary for this work are in paper form, the process is considerably more time consuming. Even though only approximately 65,000 of the total 267,949 land-based accounts Interior intends to reconcile were opened prior to 1985, analyzing them is by far the most expensive and labor intensive effort. At the time of the trial, preparations for “paper ledger era” land-based account reconciliations had begun, but the work had not commenced in earnest. Most of the other reconciliation work contemplated in the 2007 HSA Plan is complete. Interior has finished its analysis of 6,599 post-1985 transactions within land-based accounts, and has reconciled all transactions within 86% of the Judgment and Per Capita accounts. It has suspended reconciliation work on Judgment and Per Capita accounts in order to focus on the more time-consuming “paper ledger era” land-based accounts, and has deferred consideration of whether to complete transaction-by-transaction reconciliations of the remaining 14% of Judgment and Per Capita accounts. Interior plans to produce HSAs for all Judgment and Per Capita account holders, but it may conclude that reconciliation of transactions within the remaining approximately 13,600 accounts is unnecessary. In other words, its reconciliation work on Judgment and Per Capita accounts may also be complete. iii. Special Deposit Accounts Special Deposit Accounts (SDAs) are temporary accounts used to hold trust funds until they are identified for transfer to specific accounts. AR-566 at 33-03-27. The 2003 Plan revealed that $67.9 million in residual balances were in SDAs awaiting distribution. Id. As part of the historical accounting project described in the 2003 Plan, Interior planned to identify the proper owners of those funds and distribute the funds accordingly. Id. Interior reports that it has since distributed over $51 million in residual SDA balances at a cost of $48 million. Id. No documentation supporting those numbers is contained in the Administrative Record submitted for the October 2007 bench trial, but testimony at trial echoed the total figures reported in the 2007 Plan. Tr. 158:12-19 (Cason). Interior’s current position is that “SDA work . . . is not truly part of the historical accounting,” as money distributed from SDAs are “part of current accounting distributions.” AR-566 at 33-03-27. Consequently, while OHTA will continue to work on distributing funds in SDAs as part of its trust reform efforts, the resolution of these accounts is no longer an element of the HSA project. Id. iv. Asset statements The 2003 HSA plan reflected Interior’s intent to include, as part of its historical statements of all land-based IIM accounts, statements of land assets owned as of December 31, 2000. AR-566 at 33-03-28. In the 2007 HSA plan, Interior has changed its mind. Now statements of land assets will not be provided to beneficiaries in HSAs. Id. Instead, Interior indicated that it would begin to send combined statements that would include current land ownership, encumbrance, and financial information to IIM account holders in late 2007, after the TAAMS lease management module is fully implemented and the TFAS/TAAMS interface enables the agency to produce such statements. Id. This change reflects Interior’s judgment that producing land asset ownership statements as of Decemeber 31, 2000 would be costly and complicated, and would “add little value” to the historical statements of account. Id. I.2.B. Rationale for changes to the historical accounting project Several interrelated factors caused Interior to revise the plan it had adopted in 2003. Original cost and time estimates were off by several multiples, and Congress failed to appropriate the funds Interior had requested and expected. Interior initially estimated that reconciliation work would cost around $100 per transaction, but experience in recent years has revealed that reconciling a single transaction costs between $3,000 - $3,500 when searching, imaging, coding, computer support, accounting services, quality review, and statistical analysis are taken into account. AR-566 at 33–03-09. While Interior estimated that the 2003 Plan would cost $335 million to implement, its received only $127 million in congressional appropriations for IIM historical accounting between 2003 - 2007. In any event, the agency now estimates that the 2003 Plan would have cost approximately $1.675 billion to complete. See AR-566 at 33-03-05 (noting that “2003 Plan is now estimated to cost about five times the original cost estimate”). Interior expected that it could complete the historical accounting activities described in the 2003 plan -- including the reconciliation of hundreds of thousands of transactions in land- based accounts -- by 2007, but by year’s end it had completed a small fraction of those reconciliations. Indeed, at the time of trial, it had not even finalized its accounting plan for pre-1985 transactions in land-based accounts and the remaining Judgment/Per Capita accounts. Judicial decisions and congressional action following the issuance of the 2003 HSA Plan affected the agency’s implementation of its historical accounting project. Shortly after the plan’s issuance, Judge Lamberth issued a structural injunction rejecting the Department’s plan and mandating a significant expansion of the historical accounting project. Cobell X, 283 F. Supp. 2d 1 (D.D.C. 2003). On November 10, 2003, less than two months after the issuance of that opinion (and after receiving estimates from the Interior Department that compliance with the structural injunction would cost the agency approximately $3 billion, Tr. 200:17-201:1 (Cason)), Congress included a provision in an appropriations act that granted Interior relief from performing an historical accounting of the IIM trust until “the earlier of the following shall have occurred: (a) Congress [amends the 1994 Act] to delineate the specific historical accounting obligations of the Department of the Interior with respect to the Individual Indian Money Trust; or (b) December 31, 2004.” Department of the Interior and Related Agencies Appropriations Act, 2004, Pub. L. No. 108-108, 117 Stat. 1241, 1263 (2003). On December 10, 2004, exactly one year after the passage of the appropriations bill, the Court of Appeals for the D.C. Circuit reversed Cobell X, holding that the structural injunction had no legal basis after the passage of the 2004 appropriations bill. Cobell XIII, 392 F.3d 461 (D.C. Cir. 2004). Congress failed to step up to the problem, however -- it did not “delineate the specific historical accounting obligations of the Department of the Interior with respect to the Individual Indian Money Trust” before December 31, 2004 -- so, on February 23, 2005, Judge Lamberth reinstated his structural injunction. Cobell XIV, 357 F. Supp. 2d 298 (D.D.C. 2005). The reinstated injunction was stayed soon thereafter, Cobell v. Norton, 2005 U.S. App. LEXIS 5788 (D.C. Cir. Apr. 7, 2005), and eventually reversed in late 2005, Cobell XVII, 428 F.3d 1070. During this turbulent period, defendants did perform some historical accounting work, even if that work did not always conform to the scope or methodology described in the 2003 Plan. The results of work done during that period informed the agency’s adjustments to its historical accounting project. Defendants began to implement the 2003 plan for statistical sampling of post-1985 land-based transactions within the Alaska Region, but that beginning was aborted because of the judicial decisions and congressional actions described above. Tr. 971:16-972:9 (Scheuren). The land-based transaction reconciliation work performed after the issuance of the 2003 Plan was largely part of a project entitled “Litigation Support Accounting,” or LSA, which was conducted between October or November 2003 and November 2004. Tr. 199:1 (Cason); Tr. 974:7-12 (Scheuren). Interior borrowed the term “litigation support” from the appropriations bill discussed above, which temporarily halted both the agency’s obligation to perform an historical accounting of the IIM trust and, arguably, temporarily prohibited any work other than “litigation support” on land-based IIM accounts: [O]f the amounts available under this heading not to exceed $45,000,000 shall be available for records collection and indexing, imaging and coding, accounting for per capita and judgment accounts, accounting for tribal accounts, reviewing and distributing funds from special deposit accounts, and program management of the Office of Historical Trust Accounting, including litigation support. Pub. L. 108-108 at 23, (emphasis added). Noticeably absent from this list of permissible expenditures in the period between November 10, 2003 and December 31, 2004 was any allowance for reconciliation of land-based accounts. Interior interpreted the reference to “litigation support,” however, as authorizing an accounting project that involved analyzing -- as well as reducing liability arising out of -- a small sample of land-based accounts. Tr. 199:11-23, Tr. 205:16-19 (Cason). Part of the rationale for the Litigation Support Accounting project was that it would develop information that would facilitate Congressionally directed settlement discussions that were occurring during that time. Tr. 62:2-64:5 (Cason). The government retained NORC to draw samples for the LSA project, which ultimately involved the reconciliation of 6,599 transactions from land-based accounts in the electronic era. NORC’s goal in performing the LSA project was to deepen its understanding of the system in order to better target future historical accounting efforts. Tr. 974:18-975:1 (Scheuren). NORC -- along with various accounting firms assisting with the LSA project -- learned many lessons while performing this relatively small number of reconciliations. First, it learned that the cost of reconciling transactions in land-based accounts is significant. Tr. 964:23-965:16 (Scheuren). Second, however, NORC encountered far fewer errors and missing records than it had expected to discover. Tr. 977:15-22 (Scheuren). The large sample sizes contained in the 2003 Plan had been based on Interior’s assumption -- an assumption informed by historical reports, anecdotal evidence, and decades of criticism -- that records would be missing, erroneous, and in disarray. Those assumptions, NORC concluded, were overblown and incorrect. The positive results of the LSA project caused NORC to suggest that the Interior Department reassess the sample design in the 2003 Plan. Tr. 1003:3-11 (Scheuren). In fact, although NORC had reconciled less than 3% of the transactions to be reconciled under the 2003 HSA Plan, AR-566 at 33-3-22, it recommended that Interior cease conducting reconciliation work on post-1985 land-based transactions. NORC based this recommendation on the small number of errors it had encountered, and on its estimation that additional work would be expensive and would not meaningfully lower the margin of error around estimated error rates. Tr. 1004:17-1005:7 (Scheuren); 3/15/07 email from Scheuren to Edwards, AR-427 at 4-5. NORC did make certain recommendations for additional historical accounting work, which will be described in further detail below. I.2.C. Producing the historical statements of account Interior’s 2007 historical accounting project utilizes several different computer systems, contracting firms, and methodologies. It involves six discrete steps, which are set out in the 2007 HSA Plan in deceptively simple terms. AR-565 at 33­02-08-09. They will be described broadly below, and in further detail in the discussion that follows. The first step is to gather credit and debit ledgers and to assemble data from each IIM account. Id. Next, the agency reconciles selected transactions (see chart above) to determine accuracy -- comparing ledgers to contemporaneous records -- and tests the integrity of the data systems. Id. After completing that process, the DOI petitions the court for permission to assemble and mail HSA packages to beneficiaries. Id. Upon receipt of the court’s permission, the agency assembles HSA packages and mails them to beneficiaries. Id. Each HSA package will contain a list of all transactions in the beneficiary’s account along with available references for each transaction, as well as a statement regarding the accuracy and completeness of the information provided. Id. Finally, the Department will implement an administrative appeal process, by which account holders may challenge information contained within the HSAs within a specific period of time. Id. I.2.D. Information Contained in the Historical Statements of Account Historical Statements of Account are to include two main elements: a report of the transaction history within each account, listing all transactions posted to IRMS or TFAS, and “Interior’s conclusions” regarding the accuracy and completeness of the accounting provided. AR-565 at 33-02-03-04; 33-02-06-11. Information that will not be included in the HSAs will be discussed in the further detail below. I.2.E. Beneficiaries Receiving Historical Statements of Account Under the 2007 HSA Plan, Interior intends to prepare HSAs for all IIM accounts that were open at any time between October 25, 1994 and December 31, 2000. For all IIM accounts open at any point on or between those dates, the agency intends to prepare an HSA listing transactions dating back to the opening of the account or June 24, 1938, whichever is later. Defendants will not prepare HSAs for accounts closed before October 25, 1994 or opened after December 31, 2000. AR-565 at 33-02-03-06. Defendants estimate that HSAs may be prepared for approximately 364,772 accounts. AR-565 at 33-02-11. It is worth noting that there is not a 1:1:1 relationship among accounts, beneficiaries, and allotments. A beneficiary may have multiple accounts relating to a single allotment, or a beneficiary may own multiple allotments that have never been leased and never led to the creation of IIM accounts. For this reason, the number of accounts reconciled will not be equal to the number of beneficiaries receiving accounting statements. I.2.F. Transactions Reconciled under the 2007 HSA Plan i. Judgment and Per Capita accounts In its 2007 Historical Accounting Plan, Interior explains that When a tribe receives money as part of a legal judgment or negotiated settlement, the tribe may elect or be required to pass along some, or all, of the money to its enrolled members or their descendents [sic]. Similarly, other income to the tribe may be distributed to tribal members in accordance with a tribal resolution specifying the amount of the payment and the eligibility of members to receive payment. In most cases, money is distributed directly to tribal members; however, for minors or other tribal members who are not eligible to receive direct payment, the money is deposited on their behalf into an IIM account. Interior’s 2007 HSA Plan, AR-565 at 33-02-11. The majority of the Judgment and Per Capita accounts -­indisputably the easiest funds to reconcile -- have been reconciled under an 100% transaction-by-transaction approach. Interior has no immediate plans to reconcile the remaining 13,122 funds; it has put that work on hold indefinitely in order to focus on higher priority accounting work. AR-565 at 33-02-13. The reconciliations that have already been completed were conducted pursuant to a contract with Chevarria, Dunne & Lamey and reviewed by the government’s quality control contractor, Grant Thorton. The steps followed by CD&L in reconciling Judgment and Per Capita accounts were as follows: (1) query IRMS and TFAS to produce account transaction histories; (2) use document indexes to identify record boxes; (3) search for records necessary to reconcile transactions; (4) compare account details with supporting documentation; (5) calculate interest based on historical interest factor and compare with interest posted; (6) produce and seek authorization to mail HSAs. Reconciliations were performed according to the Accounting Standards Manual. Some of the funds were listed in CD&L’s reports as “fully reconciled,” while others were listed as “partially reconciled.” Scant attention was paid to the reconciliation of these accounts in the October 2007 trial. ii. Land-Based Accounts: 1985 - 2000 The LSA project reconciled non-interest transactions in land-based accounts posted on IRMS and TFAS between 1985-2000. The sample design for the LSA project involved a 100% reconciliation of the 2,099 transactions of $100,000 or more. The sample of lower value transactions reconciled consisted of 289 transactions taken from the Eastern Region, 442 transactions from the Alaska region, and, initially, a sample of approximately 20,000 transactions from the remaining 10 regions (the “national sample”). Tr. 990:12-994:10 (Scheuren); see chart § I.2.A.ii. The 20,000+ transactions in the national sample were drawn from 1,020 accounts across 10 regions, and were divided into four replicates of approximately 5,000 transactions each. In the end, the LSA contractors only analyzed transactions from one of the four replicates. Tr. 993:7-994:7 (Scheuren). The transactions analyzed were within a replicate of 5,128 transactions, but contractors attempted to reconcile only 3,769 of those transactions because the others were apparently outside the scope of the analysis (interest transactions, transactions occurring after 2000, etc.). Between the Eastern Region sample (289 transactions), the Alaska Region sample (442 transactions), and National Sample replicate (3,769 transactions), contractors attempted to reconcile 4,500 lower-value transactions pursuant to the LSA project. Accounting firms working with the sampled transactions provided by NORC were able to reconcile 2,363 of the 2,372 debit transactions sampled, and 2,117 of 2,128 credit transactions sampled. Tr. 997:1-12, Tr. 1000:2-12 (Scheuren); AR-438 (tables 4 and 7). A posted transaction from the IRMS or TFAS database was considered “reconciled” if contractors found supporting documents sufficient to determine whether the transaction was correct or erroneous. Tr. 997:5-8 (Scheuren); see Section 1.2.G. below. No errors were found among the reconciled debit transactions, and 25 errors were found among the reconciled credit transactions. Tr. 997:1-12, Tr. 1000:7-20 (Scheuren). For statistical purposes, NORC also considered all unreconciled transactions erroneous (nine debit transactions and eleven credit transactions). 9/30/2005 NORC LSA Report, AR-438 at 14, 17. Both variable and attribute sampling were employed in the LSA project, as contractors noted both the presence of errors as well the total amount of dollars in error. The population from which samples were drawn for the LSA -- the “target population” -- was not the entire population analyzed under the 2007 plan, nor was it the entire “electronic era” population. Tr. 2149:1-8, Tr. 2169:119-22 (Hinkins). Certain transactions from the electronic ledger era were not in the population sampled in the LSA project, as not all transactional data from the “electronic era” had been restored to the data set when the samples were drawn. Tr. 689:11-24 (Herman). NORC acknowledges that the positive inferences that can be drawn from the LSA work are only applicable to the population sampled -- that is, to transactions between 1985-2000 in land- based accounts available in electronic form at the time the sampling was conducted. Tr. 2152:9-16, Tr. 2152:17-2153:3 (Hinkins). Interior believes that other projects will address transactions missing from the target population in the LSA work, such as the Data Completeness Validation and Land-to-Dollars tests discussed below. Tr. 2171:5-8, Tr. 2172:1-5 (Hinkins). Interior contractor Michelle Herman also suggested that Interior plans to draw another sample after the entire data set is restored, but no defense witnesses testified as to the details of such plans at trial. Tr. 574:8-574:1, Tr. 702:16-24, Tr. 704:10­14 (Herman). Indeed, defendants’ position on post-LSA sampling and reconciliation of land-based transactions within the “electronic era” after data restoration projects is confusing. In some instances, Interior indicates that it has completed its sampling and transaction reconciliation of “electronic era” land-based transactions through the Litigation Support Accounting project described above. See, e.g., AR-566 at 33-03-14 (“no further reconciliation is necessary for the electronic era”). Elsewhere, the agency expresses an intention to conduct further sampling after data gaps are restored to the IRMS and TFAS databases. See, e.g., AR-566 at 33-03-19 (“Interior plans to reconcile a sample of these restored transactions to determine if they have an error rate significantly different from that found in the LSA sample.”); Susan Hinkins Responding Expert Report [Dkt. 3393-2] at 23 (“[T]he 2007 Plan clearly shows that DOI has not claimed to have completed tests for the Electronic Ledger Era.”). A close reading of the 2007 HSA Plan reveals that further testing of restored, “electronic era” transactions from land-based accounts is not a task that Interior will necessarily accomplish before land-based account HSAs from the “electronic era” are mailed. According to the plan, the completion of the Data Completeness Validation, Interest Recalculation, and Land­to-Dollars Posting Tests constitute “all the [remaining] steps necessary to mail land-based account HSAs[.]” AR-566 at 33-03­ 18. Because the government represents (1) that the mailing of HSAs discharges its accounting duty and (2) that HSAs can be mailed to beneficiaries prior to further reconciliation of land- based transactions within the electronic era after the data set is restored, my reading of the 2007 Plan is that Interior considers the land-based electronic era reconciliation work to be complete. iii. Pre-1985 transactions in land-based accounts Interior’s plan for sampling and reconciling “paper era” transactions -- a project Interior has yet to begin -- remains vague. Defendants estimate that approximately 65,000 of the nearly 268,000 land-based IIM accounts within the historical accounting population were open before 1985. AR-565 at 33-02-10. This reconciliation work will be more expensive, since the records are older. They may have been destroyed. If they can be located at all, they must be translated into digital form and uploaded into an electronic database before reconciliations can be attempted. On May 25, 2007, NORC recommended to the Interior Department that it begin its “paper ledger era” land-based reconciliation work by testing the hypothesis that the error rate in the “paper ledger era” is similar to the error rate in the “electronic era.” 5/25/07 memo from Scheuren to Edwards, AR-426. Six days later, DOI officially adopted this approach by including it in its 2007 HSA Plan. AR-565 at 33-02-15. The initial hypothesis recommended was influenced by, among other things, a meta-analysis conducted by NORC. A “meta-analysis” is a process by which statisticians consider studies from several different sources on similar or related subjects “to draw together what is known about the subject”. Tr. 1019:14-19 (Scheuren). The purpose of NORC’s meta-analysis was to examine whether historical studies of the IIM trust -- an incomplete but voluminous collection of around 900 documents, including some 263 audits -- suggested that the proposed hypothesis was way off base. Tr. 1020:9-1021:6, Tr. 1043:14-21, Tr. 1066:21-25 (Scheuren). NORC’s meta-analysis played a minor role in the adoption of the initial hypothesis for “paper ledger era” testing: it did not result in the hypothesis, but was instead one technique NORC used in considering whether the hypothesis was within the realm of possibility. Tr. 1021:16-1022:2 (Scheuren). Dr. Scheuren described it as a “due diligence step.” Tr. 1020:3 (Scheuren). Nothing NORC uncovered in its meta-analysis caused the statisticians to reconsider their initial hypothesis that the error rate in the “paper ledger era” resembles that of the “electronic ledger era”. Tr. 1022:4-7 (Scheuren). Complete copies of all documents relied upon in the meta-analysis have not been disclosed, nor has NORC’s internal work product summarizing those documents. See, e.g., Tr. 1443:25-1444:6 (Duncan). Following NORC’s recommendation, Interior will draw a small sample of accounts from a virtual electronic ledger compiled from the “paper ledger era,” and a small sample of transactions from within those accounts. AR-565 at 33-02-15; Tr. 1018:9-17 (Scheuren). The exact size of those samples has not been revealed. If Interior discovers that the error rate in the “paper ledger era” is similar to that of the “electronic ledger era,” the conclusion will be that “the findings of the electronic ledger era apply to the paper ledger era as well” and no further sampling will be conducted. AR-565 at 33-02-15. If, however, they are not, the Interior Department plans to broaden the sample of paper era transactions, and will probably have to extend its projected schedule to complete its work. Tr. 269:7-19 (Cason). 1.2.G. Reconciliation Process i. Contractors involved in the reconciliation work Interior’s historical accounting project now underway is a classic case of government outsourcing. It employs five accounting firms, two historian firms, and firms to assist in statistical matters and project design. AR-565 at 33-02-07. Of the five accounting firms, three (Clifton Gunderson LLC, Deloitte & Touche LLP, and Reznick Group, P.C.) are performing reconciliations and account analyses; one (FTI Consulting, Inc.) is conducting forensic accounting; and one (Grant Thornton LLP) is performing quality control reviews of all the historical accounting activities. Id. FTI worked on the first three steps in the historical accounting as detailed in the 2007 Plan, Tr. 438:2-441:20 (Herman), and has been primarily responsible for gathering account data on IIM transactions, Tr. 438:2-10 (Herman). FTI has also done reconciliation work as part of the LSA project, along with the other accounting firms. Tr. 438:18-439:13 (Herman). Michelle Herman is a Managing Director of FTI Consulting. Tr. 434:7-11 (Herman). Over the past ten years -- practically her entire post-graduate career -- she has worked for three contractor firms involved with the historical accounting project. Tr. 437:9-17, Tr. 435:4-6 (Herman). The National Opinion Research Center (NORC) works on statistical matters. AR-565 at 33-02-7. Doctors Fritz Scheuren and Susan Hinkins have led most of the historical accounting work at NORC. Morgan, Angel & Associates, LLC, and Historical Research Associates Inc., which specializes on Indian issues, provide information on leasing, the allotment process, reservation histories, and other matters. AR-565 at 33-02-07. The five accounting firms and NORC have contributed to the development and updating of the Accounting Standards Manual. AR-8 at 2. NORC was engaged to address measurement and quality issues for Interior’s July 2, 2002 report to Congress; statistical sampling was not a significant consideration at that time. Tr. 932:12-20 (Scheuren); AR-561. NORC was thereafter engaged to design and propose a sampling approach for Interior’s January 2003 Plan. Tr. 934:11-15 (Scheuren). NORC designed the sample for testing by Grant Thornton with respect to performing quality assurance reviews of reconciliation work performed by Interior’s accounting contractors. Tr. 1032:9-17 (Scheuren). ii. Accounting Standards Manual Contractors doing the reconciliation work are guided by the Accounting Standards Manual (ASM), a 581-page compilation of historical accounting standards. AR-566 at 33-03-11. Its purpose is to ensure quality and uniformity among the various firms performing historical accounting work; it is also used by the firm performing quality control services. Tr. 2109:20-24 (Dunne). The ASM governs both “paper ledger era” and “electronic ledger era” reconciliations. Tr. 580:2-12 (Herman); Tr. 2111:9­19 (Dunne). It details a reconciliation process in three steps: (1) identifying document types that might support a given transaction and searching for those documents; (2) tracing the transaction back to its initial entry in the IIM system using documents collected in step (1); and (3) quality control procedures, both within the firm conducting the reconciliation and by the QC firm, Grant Thorton. Accounting Standards Manual, AR-8 at 44-01-13. Contractors are to assess reconciliation results on a scale of 1 to 4, depending on how conclusively the supporting documents verify the transaction posted. The ASM makes reference to auditing standards, but as the historical accounting work is a “consulting engagement” and not an audit, accountants and other consultants using the ASM are not bound by those standards. Tr. 2119:16-19 (Dunne). iii. Materials Relied Upon in the Reconciliation Process Pursuant to the Accounting Standards Manual, accountants reconciling transactions are to specify whether documentation supporting a particular transaction is “level 1,” “level 2,” “level 3,” or “level 4.” All documents within levels one and two are utilized or created by the federal government in the process of managing the IIM trust; a document meets the level one standard only if it is “utilized or created by the government in the normal operations of processing [IIM] transactions.” Tr. 2112:15-19 (Dunne). Level three is the subject of considerable controversy, as it may be assigned case-by-case, using “alternative procedures” undefined by the manual, thus allowing accountants to use their judgment in validating transactions with atypical supporting documentation. Tr. 2113:8­23 (Dunne). Well over half of the 581-page ASM is comprised of examples of supporting documentation that may be used in the reconciliation process. Different types of supporting documents are used for reconciliation, depending on whether the transaction is a receipt or a disbursement. Tr. 2121:16-2122:5 (Dunne). A receipt (or credit transaction) must be linked to a document reflecting expectation of a collection, such as a lease. Tr. 2121:21-23 (Dunne). A disbursement (or debit transaction) requires evidence that the government issued the check, such as a check register. Tr. 2122:2-18 (Dunne). iv. Systems relied upon The reconciliation work uses transactions systems (IRMS and, later, TFAS), and ownership systems (LRIS and, later, TAAMS). In order to reconcile “electronic era” transactions, DOI contractor FTI loaded data from TFAS and the IIM module of IRMS -- along with restored transactions discovered in the course of the Data Completeness Validation project –- into a database called the “SQL server”. Tr. 449:13-24, Tr. 450:5-17, Tr. 453:1­ 454:17 (Herman). The financial information has also been loaded onto the Account Reconciliation Tool, or ART. Tr. 596:15-25 (Herman). The ART, demonstrated at trial, is a powerful database that allows for extensive searching, linking, and reconciling of transactions with supporting documentation. Tr. 599:1-644:14 (Herman). Because the IRMS database was inconsistently used across agencies and some transactions are missing from all systems, the database is incomplete. Tr. 1036:8-15 (Scheuren); Tr. 2042:1-4 (Christie). The Data Completeness Validation project (described below) aims to find “electronic ledger era” data that is not available on IRMS or TFAS and then input the “restored” data into the ART as it is discovered. Ownership information utilized in performing transaction reconciliations is typically derived from the LRIS excerpt BIA provided OHTA at the beginning of its historical accounting work. Tr. 614:10-17, Tr. 639:24-640:5, Tr. 760:6-8 (Herman). Contractors engaged in reconciliation work rely on that excerpt in determining ownership interests in allotments, although the LRIS ownership data is incomplete. v. Definition of error The validity of defendants’ definition of “errors” in its historical accounting analysis was the subject of considerable testimony, confusion, and disagreement at trial. It is not for me to determine the best “error” definition for purposes of statistical analysis, but it is worth sorting out what defendants’ error rates do and do not reflect. Defendants’ plan employs both attribute and variable sampling. Attribute sampling answers yes or no questions (such as whether there is a discrepancy between a transaction recorded on IRMS and the supporting documentation), while variable sampling records the value of the error/difference (such as, “recorded transaction is for $10 less than supporting documentation would indicate”). As I understand it, there are differences in the definition of “error” between the different historical accounting tests. For example, in discussing the LSA project, Dr. Scheuren testified that any transactions LSA contractors attempted but were unable to reconcile were treated as errors, even though they had no reason to believe that the inability to reconcile those transactions reflected errors. Tr. 997:13-1001:15 (Scheuren). Dr. Hinkins also testified that, when contractors informed NORC that they were unable to reconcile transactions, NORC’s statisticians treated those unreconciled transactions as errors. Tr. 2155:9-22 (Hinkins). Michelle Herman testified that, in conducting transaction reconciliations, any disparity of two cents or more was recorded as an error. Tr. 615:21-616:2 (Herman). After some internal debate within Interior, the agency decided that, in the transaction-by­transaction reconciliations, the failure to locate supporting documentation should be recorded as an error. Tr. 1949:15-20, Tr. 1958:15-17 (Zippin). In other instances, however, it appears that the failure to locate supporting data is not recorded as an error. The 2007 HSA Plan notes that an error indicates only that “the actual transaction posting to an account is different from the amount expected to be posted based on the contemporaneous records.” AR-565 at 33-02-30. James Cason testified that an inability to find supporting data is not automatically recorded as an error. Tr. 187:14-188:8, Tr. 190:20-191:1 (Cason). Michelle Herman testified that she “consider[s] an error a transaction that wasn’t posted correctly, not transactions that are missing from the record set.” Tr. 789:15-20 (Herman). It appears that the definition of error employed by NORC in its statistical analysis of the LSA reconciliation work, and relied upon for the claimed 1% error rate, is broader than the definition used elsewhere in the historical accounting project. Error statistics, however, do not reflect the failure to reconcile transactions intentionally left out of the LSA sample because of expected reconciliation difficulties, NORC Interim LSA Report dated 12/28/04, AR-416 at 51-09-17, or transactions left out of the sample because they had not yet been “restored” through the DCV project. I.2.H. Data Completeness Validation Project Michelle Herman has designed and directed the Data Completeness Validation (DCV) Project for Interior: a battery of tests on IIM data that together comprise a major component of the historical accounting project referenced in the 2007 Plan. According to Interior’s plan, the project goals are “(1) identifying and resolving gaps in electronic data, (2) verifying the transfer of accounts and balances through system conversions (from paper ledgers to IRMS and from IRMS to TFAS), and (3) assessing the integrity of the underlying electronic data.” AR-566 at 33-03-15. Ms. Herman testified that the primary goal is “to assess the completeness of the electronic ledger era for historical statements, and when I mention completeness, it’s of the data today versus the data as posted historically.” Tr. 458:2-6 (Herman). This project has been a massive undertaking. At FTI, between four and eight employees have worked between three and four years conducting DCV testing on some 113 million transactions, or 50,000 annual man-hours, Tr. 502:19-503:9, Tr. 465:9-21 (Herman), not including the work of other contractors who support DCV testing through projects like scanning and coding, etc. The DCV project carries out tests for electronic data gaps and system conversion that were contemplated in the 2003 HSA Plan. Tr. 451:1-10 (Herman). The data gaps test is to determine whether the electronic data stored in the IRMS system was complete, and the system conversion test was to make sure that data were successfully transferred from the IRMS system to the TFAS system when Interior rolled out TFAS between 1998-2000. Id.; Tr. 450:20-21 (Herman). The DCV project primarily relies on information taken from the IIM module of IRMS from between 1985 ­2000 and information from TFAS from between 1998 - present. To some extent, the DCV project has also relied upon TAAMS in the account number analysis. Tr. 453:9-11 (Herman). FTI has produced a national report describing its DCV work in all 12 BIA regions, and an additional six more detailed reports on six specific regions. Tr. 455:1-12 (Herman). The six regions for which DCV work is largely complete are the largest ones, including the first three to convert from IRMS to TFAS. Tr. 456:15-18; 21-24 (Herman). Within each of the regional reports, FTI describes the months and years of available data for each agency within the region, the transactions restored to the database, accounts out of balance, results of the “reused accounts” test, and the results of transaction mapping efforts. Tr. 482:2-483:12 (Herman); see DX-153A, DX-154A, DX-155A, DX­156A, DX-157A, DX-158A (updated regional reports). Transaction mapping is another component of the DCV project. Tr. 465:1-4 (Herman). Transfer transactions –- such as a debit from an SDA account and a credit to one or more IIM accounts –- should sum to zero, so when FTI identifies instances where that has not happened, it attempts to identify documentation explaining the apparent (or actual) discrepancy. Tr. 484:15-484:17 (Herman). Efforts to map transactions can result in one of several outcomes: (1) the transfer can be mapped successfully (sum of transfers is zero); (2) the transfer can be an “explained difference” (sum of transfers is not zero, but not because an error was made); (3) the transfer can be unresolved (sum of transfers is not zero, FTI awaiting further documentation to complete forensic analysis); (4) the transfer can reveal a posting error; or (5) the transfer can indicate that data is missing from the system. Tr. 496:14-20 (Herman). Bookkeeping mistakes uncovered during transaction mapping will be reflected on HSAs in the “statement of known difference” column. Tr. 520:1-7 (Herman). Transaction mapping takes place electronically in almost all cases. Tr. 498:1-499:-1, Tr. 500:4­7 (Herman). When discrepancies cannot be mapped with electronic information on the SQL server, FTI will put in a request (through OHTA) for searchers to look for supporting documents at the AIRR in Lenexa. Tr. 504:16-23 (Herman). If the searchers find supporting documentation, the documents are loaded onto the ART system which can be accessed by FTI in Los Angeles. Transaction mapping was helpfully demonstrated by Michelle Herman during the October 2007 trial. Examples of mapping activities include comparing balance files and transactions within a single account (the “roll-forward analysis”), Tr. 525:13-528:15, Tr. 552:2-21 (Herman), analyzing the transfer of account balances from paper ledgers to the IRMS system, Tr. 523:23-525:2 (Herman), and check disbursement mapping, to follow disbursement data available in the database through to Treasury records. Tr. 535:4-536:13 (Herman). When FTI workers (or other DCV contractors) discover a hard copy document (such as a bill for collection, journal voucher, or a transaction register print out) indicating that a certain transfer took place that is not reflected on the IRMS and TFAS systems available on the SQL database, they input the transaction into the database in an effort to complete the data set: these are referred to as “restored transactions.” Tr. 529:19-530:9 (Herman). Transactions occurring both before and after 1985 have been restored to the electronic data set, so that the database will ultimately be utilized for “paper era” accounting work, too. At the time of trial, approximately 113 million transfers had been mapped as part of this effort. Tr. 551:16-19 (Herman). FTI had placed approximately 80,000 DCV- related documents requests of the AIRR searchers; 35,000 remained outstanding. Tr. 505:3-10 (Herman). It can take searchers from a week to longer than a year to find a document at AIRR. Tr. 505:11-14 (Herman). Because account numbers have been re-used over time, HSAs cannot be prepared according only to account numbers. The DCV project thus assigns each account a new, unique “Native American Account Number” or NAAN, and each beneficiary a unique “Native American Beneficiary Number” or NABN. Tr. 460:1-461:11 (Herman). FTI discovered approximately 645,000 IIM accounts that are open, or have been open, on the IRMS and TFAS systems between 1985 and present. Tr. 459:11-22 (Herman). All but 786 accounts had been analyzed in the re-used account test at the time of trial. Tr. 487:15-24 (Herman). This set of accounts is broader than the set of accounts for which Interior is performing an historical accounting. Tr. 487:25-288:3 (Herman). ii. Results of DCV Testing FTI and other contractors had restored 451,875 transactions to the electronic data set at the time of trial: 276,118 of those transactions occurred after February 1985, and 175,757 occurred before February 1985. 9/30/07 DCV Interim Report, DX-152A at 20; Tr. 721:21-722:15 (Herman). The record does not establish what percentage of originally missing transactions this figure represents. Plaintiffs’ expert suggested that the number may be in the realm of only 7%, but because significant pieces of information were not included in his calculations, I do not consider his estimate reliable. Tr. at 1387:13-1390:14, Tr. 1469:19-1476:18 (Duncan). FTI has also mapped over 6 million of the 6.4 million checks issued to IIM beneficiaries during the “electronic era.” Tr. 535:2-536:13 (Herman). More than 97% of the checks analyzed between 1987 and 2002 were successfully mapped. Tr. 565:2-566:6 (Herman). At the time of trial, FTI had confirmed that 266,285 accounts successfully transitioned from IRMS to TFAS. DX-152A at 20. Of the 355,320 accounts FTI has identified as falling within the scope of the historical accounting Interior now plans to accomplish, 54,618 accounts still have unreconciled differences or data gaps, which FTI believes is largely attributable to the unavailability of many balance files before 1989. DX-152 at 271. FTI has completed mapping the majority (93.9%) of IRMS and TFAS transactions from the electronic era. DX-152A. I.2.I. Posting Test/Land-to-Dollars Test i. Test purpose The LSA project looks only at transactions posted to IRMS and TFAS, it does not detect the “failure to collect, deposit, and record collection transactions.” 3/31/07 memo from Jeffrey Zippin to Susan Hinkins re Land to Dollar Completeness Test at Horton Agency, AR-435 at 38-01-01. To identify any such omissions, Interior’s historical accounting plan includes a test designed to start with the land and examine whether revenue generated by IIM allotments has actually made it into the proper IIM accounts. Tr. 2121:13-15 (Dunne). The goal of this Land-to- Dollars (or “posting”) test -- which is to be conducted on a sample basis -- is to confirm that funds generated by leased allotments were actually received and properly entered into the IIM system. AR-565 at 33-02-21. Since Interior has only recently instituted an accounts receivable system, this test responds to a very credible concern that not all payments owed by lessees were received into the IIM trust. ii. Initial effort: the Alaska prototype In 2003, NORC documented its initial effort to design a sample for such a test. AR-389 at 47-03-04. Thirty-three land- related documents were selected from Lee’s Summit in the Alaska Region. The process of linking these documents to transactions proved to be challenging, as the documents necessary for the test were in multiple locations, in formats and maintained according to systems that were subject to wide variations between regions. Id. at 47-03-10. NORC’s 2003 conclusion was that the sampling process utilized in the Land-to-Dollars testing needed to be “rethought.” Id. iii. Pilot Test at the Horton Agency In 2006, Interior conducted a pilot Land-to-Dollars test at the Horton Agency in Kansas. 3/31/07 memo from Susan Hinkins to Jeffery Zippin re: Land to Dollar Completeness Test at Horton Agency, AR-435 at 38-01-01. The Horton Agency was chosen because it processed farming and grazing leases, making more straightforward document collection possible, and because agency staff were not already burdened by requests from the LSA project. Id. at 38-01-02. Interior does not claim that the results of this test are representative of collection and posting patterns throughout the IIM trust, and expects that further testing in other locations will be more challenging. Id. Interior does not have a complete list of land records, and compiling a complete list of such documents would be, at the very least, extremely difficult. Id. at 38-01-01. Trial testimony revealed, indeed, that lease information in some cases has been routinely deleted from Interior’s electronic systems. Tr. 1750:18-1751:13 (Infield). NORC, however, believes that a complete inventory is not necessary to successfully conduct these sorts of tests, AR­435 at 38-01-02, and Interior has adopted this view. Tr. 247:5­17 (Cason). NORC discovered that land records at the Horton Agency were in good order. AR-435 at 38-01-02. NORC analyzed 21 of the 812 allotments within the Horton Agency, AR-435 at 38-01-03, reviewing agency data to determine what sort of revenue generating activity was processed by Horton between 1985 - 2000. Of the 21 allotments analyzed, only ten showed evidence of farming and grazing revenue generation during that time period; two showed evidence of potential revenue from subsurface interests which were not analyzed. Id. For the ten allotments that did appear to have farming/grazing revenue during those 16 years, NORC drew a sample (a sample of a sample) of expected annual revenues. Id. at 38-01-04. Revenue generating documents were located for 33 of 35 samples analyzed, and for 32 of those 33 samples (all of the ones within the temporal scope of the study), accountants verified that “revenue was received for the listed owners.” Id. There is no indication that the test considered the accuracy of revenue postings, or whether the “listed owners” were the correct owners. The test was performed on a very small sample of allotments, and no statistical conclusions can be drawn from it. AR-435 at 38-01-03. iv. Plans for future posting tests OHTA plans to conduct more land-to-dollars tests: at least one test for each of the twelve BIA regions. Michelle Herman testified that some were underway at the time of the evidentiary hearing, but no evidence reflecting these tests was introduced or contained in the Administrative Record. Tr. 587:3-17 (Herman); AR-566 at 18. Nor was any evidence presented to demonstrate how (or whether) Interior plans to address apparently missing lease data: a problem encountered in two of the 35 samples analyzed in the Horton Agency pilot. AR­435 at 38-01-04. I.2.J. Interest Recalculation Project Each historical statement of account will include interest posted to the beneficiary’s account in the electronic record. Tr. 115:22-25 (Cason). Interior will perform independent recalculations of the interest posted in the ledgers to confirm that the interest listed in the HSAs reflects the interest actually earned. Tr. 114:15-19 (Cason). The interest postings listed in each account will be recalculated using historical interest factors. Tr. 1963:23-1964:4 (Zippin). IIM funds earn interest in the following manner: monies from IIM accounts that are entitled to earn interest are pooled for investment in certain types of securities or bonds that pay a dividend or earn interest. Tr. 1966:18-24, Tr. 1967:7-13 (Zippin). At certain intervals, the interest earned by the pooled funds is transferred into a single Special Deposit Account. In accordance with applicable interest payment rules, the dollars in that SDA are then divided by the total IIM dollars in accounts to which the interest is applicable. That fraction is then used to calculate interest owned to eligible IIM accounts. Tr. 1965:12-1967:13 (Zippin). At trial, Jeffrey Zippin hesitantly confirmed that the “calculation assumes . . . that the pooled funds that were invested reflects [sic] accurately the total of all of the individual IIM accounts.” Tr. 1967:14-18 (Zippin) (“I’ll have to say that would be my understanding. I don’t have direct knowledge of that.”). The pooled investments did not always reflect the complete sum of the IIM account balances, however, and the Interest Recalculation Project does not address any problems relating to understated interest resulting from this discrepancy. Tr. 1967:22-1969:7 (Zippin); see also 914:17-917:25 (Winter). The ASM instructs contractors to recalculate interest by applying the interest factor described above to historical balances in IIM accounts. Tr. 2122:8-10 (Dunne). In their HSAs, beneficiaries will be informed of the new, independent calculation of interest, and any limitations on that calculation. Tr. 114:21-115:2 (Cason); Tr. 1968:23-1969:4 (Zippin). This project responds to OHTA’s concerns over whether interest was calculated and credited properly to IIM accounts. Tr. 1964:5-9 (Zippin); 9/13/02 Memo from Chavarria to Edwards, “Issues Regarding Interest Yields,” AR-367 at 60-26-02; 11/1/02 Memo from Chavarria to Zippin, “IIM Trust Funds System Level Issues,” AR-361 at 60-20-02-05. After applying this test to Judgment and Per Capita accounts, Interior concluded that most discrepancies in individual accounts were minor, typically only a dollar over or two dollars under the posted interest amount. Tr. 116:4-9 (Cason). The interest recalculation test proposed in the 2007 HSA Plan has significant limitations. Problems with pooled IIM investments have been extensively documented. See, e.g., AR-361 at 60-20-02. Unfortunately, Interior’s approach “is of minimal value if the actual balance on which the interest is calculated [is] believed to be inaccurate.” Office of Historical Trust Accounting, Bank of America Contract #1435-01-02-CT-85121 (undated), AR-579 at 31-06-01. Actual investments will not be reconstructed. The interest factor “is not about interest earned; it’s about the total amount that was available to be paid from the interest, and the factor is what [Interior] use[d] to apply to the money in the accounts.” Tr. 1966:10-17 (Zippin). At trial, OHTA acknowledged that recalculating the historical interest factor will only reflect how much interest should have been posted to IIM accounts if the pooled funds accurately reflect the IIM funds available for investment. Tr. 1968:15-25 (Zippin). I.2.K. Land title records i. Problems with land title records Concerns over the accuracy of ownership records have plagued the IIM trust for years. Early in the historical accounting project, Interior’s accountants advised that “project teams will need the ownership records for subject land and will need to know that the ownership records are correct.” 3/12/02 Ernst & Young “OHTA Accounting Methods” memo, AR-25 at 1-18-03. NORC has also advised defendants that “land ownership records must be verified” in order to confirm the correct distribution of income derived from land allotments as part of the historical accounting. 12/2/02 NORC memo re: sampling plan, AR-235 at 08­02-07. Unfortunately, testimony at trial echoed findings in the Misplaced Trust Report (as well as other studies) suggesting that Interior’s electronic land ownership records are inconsistently updated, inaccurate, and unreliable. Tr. 2064:6-11 (Christie). There was anecdotal testimony that, at least in some regional offices, LRIS was not timely implemented; that the IRMS ownership module was outdated, inconsistently used, or and not relied upon at all, and that, when agencies sent their ownership data to Land Title Records Offices (LTROs), it was not always uploaded to LRIS in a timely fashion. Tr. 1742:21-1743:15 (Infield); Tr. 1318:21­1322:14 (Redthunder). Interior’s contractors have discovered, however, that although LRIS is the only electronic source of historical ownership information, some data that cannot be found on LRIS can be found elsewhere in hard copy. NORC Analysis of LRIS Tract History Reports, AR-405 at 50-02-10. Note that LRIS is not used for disbursing IIM funds, but instead as an historical record of allotment ownership. AR-405 at 50-02-05. For that reason, LRIS data is critical to an historical accounting. ii. Land Title Records Office test Beginning in 2001, defendants started testing land title records kept at BIA Land Title Records Offices (LTROs). See NORC’s 11/25/01 “Design Report on Sampling and Economic Applications,” AR-399 at 49-03-17. After completing this project, Interior’s “contractor determined that the LRIS data examined adequately reflected land ownership reported by the Bureau of Indian Affairs’ (BIA) LTROs[;] [and] [a]s a result of the project, Interior is confident in the information held within the LTROs for use in the historical accounting.” Interior’s Thirteenth Status Report dated May 1, 2003, AR-549 at 24-18-02. The land title records study was performed by NORC and OHTA. Four reports in the administrative record reveal the study’s design and major findings. See AR-404; AR-405; AR-406; AR-407. The purpose of the test was to determine whether the land title systems could be relied upon in the historical accounting. AR-407 at 50-04-05. Using statistical sampling, NORC examined randomly selected allotments and traced a small number of tract ownership histories from LRIS, through probate proceedings for some of those tracts. AR-407 at 50-04-03; AR-565 at 33-02-19. Between November 2001 and May 2002, NORC and OHTA visited LTROs in Aberdeen, Alaska, Albuquerque, Anadarko, Billings, Muskogee, Portland, and Sacramento. AR-407 at 50-04­03-05. The intention was to examine: (1) the completeness and accuracy of the tract records underlying IIM accounts; (2) whether ownership transfers resulting from probate were accurately captured; and (3) how the land and lease records might be used in the historical accounting. Id. NORC encountered difficulties in executing these tests. This court’s December 5, 2001, temporary restraining order related to IT security shut down the LRIS computer database and significantly affected NORC’s performance of these tests. Also, because each LTRO was different, NORC had to customize its approach to each office’s filing systems. AR-407 at 50-04-06. In November 2001, Interior conducted its first review of an LTRO in Albuquerque, AR-407 at 50-04-05, before the court’s shutdown of Interior’s computers. Albuquerque was the only LTRO where Interior was fully able to examine LRIS reports. LRIS was unavailable in Alaska by March 2002, but NORC was able to review the LTRO’s offline data. AR-407 at 50-04-07. Aberdeen, the largest LTRO, had a title service area covering the Great Plains and Midwest BIA regions, comprising approximately 7.2 million acres of allotted lands (individual and tribal). AR-404 at 50­01-06. NORC sampled 50 tracts at the Aberdeen LTRO and then drew a sub-sample of 12 tracts within that population. AR-404 at 50­01-07. NORC reported that “the contents of the Aberdeen LTRO systems that we examined appear complete and accurate, within the scope of the checking we did. We therefore are ready to recommend their fuller use in the historical accounting to come.” AR-404 at 50-01-08. NORC reported similar findings in other LTROs. AR-407 at 50-04-07-13; AR-565 at 33-02-19; AR-404 at 50-01-13 (Anchorage), 50-01-25-26 (Anadarko), 50-01-32-33 (Billings), 50­01-44-45 (Portland). Tests examining ownership changes by operation of probate orders were narrowed to focus on recent beneficiaries, however, using the electronic database that covered from the mid-1980s forward. The results of the ownership testing were based on a random sample taken from LRIS Tract History Reports (“THRs”) in the Aberdeen, Albuquerque, Anadarko, Billings, and Portland LTROs. AR-405 at 50-01-04. NORC found no material errors in the LRIS probate entries it sampled, AR-405 at 50-01-04, but the sample used for this testing addressed only 99 probated estates and was, therefore, not large enough for estimating an “error rate” among the probate entries at conventional assurance levels. AR-405 at 50-01-04-05. NORC qualified its report on the LTROs in the following manner: “Given the importance of LRIS for the historical accounting, it is clearly desirable to use the knowledge gained so far to design a larger statistical sample which will allow us to estimate the accuracy of the data on LRIS with a higher level of confidence.” AR-405 at 50-02-22-23. For reasons unexplained on this record, no further testing has been conducted, or, apparently, is contemplated. I.2.L. Mailing historical statements of account Interior asserts in its 2007 Plan that its duty to account is discharged by the mailing of HSAs. AR-565 at 33-02­ 09. Interior acknowledges that, while HSA packages will be mailed to the most current known address for each beneficiary, it will encounter difficulty identifying current addresses for former account holders. Even many current account holders are “whereabout unknowns,” for whom Interior lacks accurate mailing addresses. AR-565 at 33-02-08. Consequently, Interior expects that many HSAs will be returned. Id. at 33-02-09. Interior’s quality control procedures for the HSA mailing process involve sampling HSAs to confirm that the envelope includes the relevant inserts and that the inserts are consistent with the listed addresses. Id. at 57-14-04-05. I.2.M. Anticipated administrative appeals process Interior is currently drafting regulations for an administrative appeals process under which a beneficiary who objects to his or her HSA can petition OHTA for review. AR-566 at 33-03-20-21. Generally, the agency would like to establish a process by which beneficiaries must challenge information in the HSA within a certain number of days from the date of the HSA, in order to allow the agency to fix the mistake in the first instance; the failure to raise a challenge within the set time period would likely result in the HSA becoming “conclusively deemed accurate and complete for all purposes.” See Draft Administrative Appeals Proposed Rule, [Dkt. 3356-2] at 26. The agency plans to issue a notice of proposed rule-making in the Federal Register to seek public comment on the rule. Tr. 88:2­15, 221:24-222:11 (Cason). At the present time, the Department is unable to issue a notice of proposed rule-making in the Federal Register because of the class communication orders in this case, which defendants have moved to vacate [Dkt. 3348]. Plaintiffs have strenuously opposed the administrative appeals process they expect defendants to propose based on previous drafts released [Dkt. 3356]. A ruling on that motion will follow the release of this opinion. I.3. Areas Outside the Historical Accounting Plan A central purpose of the October 2007 bench trial was to allow a more thorough consideration of whether the historical accounting underway is legally sufficient. That process requires consideration, not only of the activities that are underway in the historical accounting project, but also of the accounting activities that will not be performed under the 2007 Plan. A review of areas that Interior has declared outside the scope of its accounting effort follows. I.3.A. Account sampling Interior has no plans to reconcile each and every transaction within any single land-based account: the sampling design in the 2007 Plan involves reconciling a small number of transactions from within a small number of land-based accounts. Tr. 2164:15-17, Tr. 2167:10-16 (Hinkins); Tr. 1084:13-1085:5, Tr. 1105:7-10 (Scheuren); Tr. 111:8-10 (Cason). It is Interior’s judgment that this approach is the best balance between the twin aims of controlling costs and providing reasonable assurances of the accuracy of transactions within all land-based accounts in the sampled pool. Tr. 1085:3-7 (Scheuren). I.3.B. Reconciliation of account balances The focus of the historical accounting work is on the reconciliation of sampled transactions. Defendant’s contractor confirms that “[t]he 2007 plan clearly states that an accuracy and completeness statement will be provided with respect to transactions, not with respect to account balances. . . . NORC was never asked to design a sample which would provide the individual with a statistical statement about his or her account balance.” Hinkins’ Response to Duncan’s Expert Report, DX-4 at 36; see also Tr. at 2161:11-23, Tr. 2163:3-16, Tr. 2167:10-16 (Hinkins). Interior’s position appears to be that, as a result of sampled transaction reconciliations and tests designed to probe the reliability and completeness of the systems themselves, beneficiaries’ HSAs should provide meaningful assurance that account balances are accurate. At trial, much was made of the discrepancy between the total balance of all subsidiary IIM accounts and Interior’s general ledger balance, or control account. Tr. 907:19-917:25 (Winter). A Comptroller General Report from 1982 indicated a $25 million