Office of the Historian

March 2024

Advisory Committee on Historical Diplomatic Documentation March 11-12, 2024

Minutes

Committee Members

  • James Goldgeier, Chair
  • Kristin Hoganson
  • Adriane Lentz-Smith
  • Sharon Leon
  • Nancy McGovern
  • Timothy Naftali
  • Deborah Pearlstein
  • Kori Schake
  • Sarah Snyder

Office of the Historian

  • Kristin Ahlberg
  • Carl Ashley
  • Margaret Ball
  • Forrest Barnum
  • Sara Berndt
  • Josh Botts
  • Tiffany Cabrera
  • Mandy Chalou
  • Elizabeth Charles
  • Kathryn David
  • Cynthia Doell
  • Lynette Evans-Tiernan
  • Thomas Faith
  • Stephanie Freeman
  • Amy Garrett
  • David Geyer
  • Renée Goings
  • Ben Greene
  • Michelle Guzman
  • Charles Hawley
  • Kerry Hite
  • Adam Howard
  • Richard Hulver
  • Alina Khachtourian
  • Virginia Kinniburgh
  • Laura Kolar
  • Aaron Marrs
  • Michael McCoyer
  • Brad Morith
  • Christopher Morrison
  • David Nickles
  • Nicole Orphanides
  • Paul Pitman
  • Alexander Poster
  • John Powers
  • Kathleen Rasmussen
  • Matthew Regan
  • Amanda Ross
  • Seth Rotramel
  • Daniel Rubin
  • Nathaniel Smith
  • Douglas Sun
  • Brooks Swett
  • Melissa Jane Taylor
  • Chris Tudda
  • Dean Weatherhead
  • Joseph Wicentowski
  • Alex Wieland
  • James Wilson
  • Louise Woodroofe

Bureau of Administration

  • Marci Bayer
  • Jeff Charlston
  • Corynne Gerow
  • Timothy Kootz
  • Mallory Rogoff
  • Marvin Russell

National Archives and Records Administration

  • William Fischer
  • David Langbart
  • Don McIlwain
  • Brewer Thompson

Public

  • Over 50 members of the public

Open Session, March 11

Office of Information Programs and Services Presentation

Adam Howard opened the session by introducing Timothy Kootz and asking him to begin his presentation. Kootz introduced himself, as did his co-presenter Dr. Samuel Stehle, a data scientist in the Department of State’s Center for Analytics. Kootz also thanked the Advisory Committee for its contributions to the work of declassification and transparency.

Kootz briefly outlined how Artificial Intelligence (AI) is bringing about a paradigm shift for declassification programs. He noted that AI augments human review, expedites the declassification process, enhances consistency in declassification decisions, enables proactive releases, could reduce the need for information access requests, and increases public engagement.

Kootz went on to note that A/GIS/IPS handles all aspects of the records lifecycle at the Department of State, then Stehle provided an overview of the work of M/SS/CfA. He explained the CfA is the Department’s “everything data” office. It manages the Enterprise Data Strategy, overseeing issues like software procurement and also conducting analytical work, which is Stehle’s focus. CfA also manages the Enterprise AI Strategy to leverage its possibilities and to ensure responsible use.

Kootz and Stehle then shared and narrated a PowerPoint presentation about Department efforts to use AI in the declassification process.

The Historian opened up the floor for questions.

Sarah Snyder asked for an example of a process disagreement versus a content disagreement. Kootz replied that in the geopolitical atmosphere sensitivity changes over time. The QC process has led to a further question—how do we perform uniform declassification decisions? One facet of this is to update our declassification guides regularly by staying aware of current events and consulting with bureau counterparts for up-to-date knowledge. There are policy and guidance implications in addition to process implications. Do our declassification guides meet expectations and how do we measure and apply it to a human reviewer’s perspective when they may interpret guidance differently?

Timothy Naftali wondered how the percentage of 1997 cables exempted differed between human and AI review. Stehle answered that the difference was small—3% compared to a high 2%. Kootz added that historically the Department had a low rate of exemption in cables eligible for declassification review. The difference was small in this case because they were using the results of human review to train the model. Small disagreements were revealed, however, and that still needed to be worked out.

Kori Schake questioned how, since the presenters emphasized the importance of training for specific years, they would amass information to train the model for special projects and other more tailored cases. Stehle responded that this would be the main focus over the next 12 months as they built the model out further. He predicted that this would mean incorporating non-cable materials into the training of the model. Kootz stated that they were considering supervised learning. IPS possesses content that spans many decades. They were also looking at other ways to train the model, including a semi-active learning approach in which the model would learn by absorbing smaller amounts of information and feedback over time.

Another participant wanted to know more about the modeling training process and wondered whether they used any specific algorithms such as LDA or other topic modeling systems. Stehle stated that they did not use topic modeling but rather tried a large suite of machine learning processes. Topic modeling doesn’t help categorize but does help summarize and tells us what kinds of information is present. It does not help in making declass decisions.

James Goldgeier asked how they would use AI to respond to FOIA requests. Kootz replied that one way was through the customer service aspect. Can we help a requester create a stronger request? You may be asking for something but this is what we have in our archives. Requesting that will take a lot of time to process; have you considered this as a much quicker alternative? Have we already received a similar or identical request? Have we released some of these documents already?

A virtual participant wondered what measures were in place to address and mitigate potential biases and errors. Stehle answered that there is a second pair of eyes after the review. Many cables are seen by multiple reviewers, and many are seen by the model and multiple reviewers. Kootz added that neither humans nor the machine is perfect. IPS is willing to take the risk. They weren’t looking for 99.9% efficiency and have purposely lowered the threshold for the sake of public transparency.

Deborah Pearlstein noted that there were very few exemptions in these sets of cables. Considering the endemic problem of overclassification, this could be a very interesting case study. Is there a process where you could incorporate what you are learning here into the front end of classification decision making? Stehle agreed, that was why the model applied its’ exemptions so that reviewers had the opportunity to disagree with the model’s decisions. Kootz stated that whenever a classifier makes either a derivative or original classification decision it creates new data and gives new insights into that area.

Kristin Hoganson asked whether the model could identify documents for earlier declassification review consideration. Also, how does the current resource allocation compare to simply using exclusively human reviewers? Kootz replied that resource allocation was neutral at present but over time there would be an explosion of growth. Future volume increases would mean enhanced capabilities and cost savings. Approximately 50–60 percent of cables are declassified in the first instance. Once State can get the P and N reels digitized, they can also run them against these models. This as a game-changer in helping speed up the process. Ideally, only two areas would remain classified—the exempted and the referrals.

Another participant asked if the use of foreign language can impact results. Jeff Charlston replied that foreign languages are not typically present in cables. It is not a significant issue at present but will be in future datasets. In the 25-year review process, the fact that it is releasable now does not mean it should not have been originally classified. In addition to records complexity, things change in the international arena and can impact declassification decisions.

The Historian thanked the presenters and closed the session.

Remarks from FSI Director Polaschik

Pearlstein opened the session at 11:05 a.m.

Historical Advisory Committee (HAC) Chair James Goldgeier was unable to attend the meeting, so Deborah Pearlstein served as acting Chair. She began the session by seeking a motion to approve the minutes from the December 2023 HAC meeting. The motion was made, seconded, and the minutes were approved. Pearlstein then turned the meeting over to Howard, and he, in turn, introduced Foreign Service Institute (FSI) Director Joan Polaschik, who spoke next. She began by introducing the latest addition to the OH staff, Richard Hulver, who is the new Team Lead of the Oral History section of the Historical Studies Division. She also noted staff changes in FSI leadership. She noted the departure of FSI Deputy Director, Shelby Smith-Wilson, who is now serving as the Principal Deputy Assistant Secretary for the Bureau of International Narcotics and Law Enforcement Affairs. Smith-Wilson had been with FSI for eighteen months and was essential in developing FSI’s core curriculum as well as the Department's new learning policy. FSI is now in the process of recruiting a new Deputy Director. In the meantime, Howard will be serving as Acting FSI Deputy Director, while OH Deputy Director Goings will serve as Acting OH Director. Polaschik expects the recruiting process for a new FSI Deputy Director will take no more than two months.

Polaschik followed these personnel changes by briefly describing structural changes now taking place at FSI. The process of establishing an Office of the Provost is underway. The Provost will essentially be the chief academic officer for FSI. The application period for the Provost position closed at the end of February and FSI is currently waiting to receive the list of candidate referrals from the U.S. Office of Personnel Management so the interview process can commence. While this lengthy process is ongoing, FSI is concurrently hiring the Division Directors for the Office of the Provost as well as a Vice Provost. FSI is also working on creating its new Board of Visitors, which will be modeled on the HAC. Polaschik hopes to have the Board in place by August or September. She described the board as having three components: former Department officials with vast Department knowledge; experts in national security (including those with expertise in the areas of climate, global health, China, emerging technologies, multilateralism, and economic statecraft); and experts in adult education. Polaschik also stressed the importance of OH and its staff of historical experts to FSI and FSI training, as well as OH’s work to inform the Department’s policymaking process. Polaschik concluded by highlighting the efforts of John Powers, the Declassification Coordination, Publishing, and Digital Initiatives Director, and his team in working with inter-agency partners to design and pilot a new process to manage and streamline the backlog in the FRUS declassification. She noted that Powers will provide more details during the afternoon closed session. She stressed that OH expects more FRUS publications this year than in the past few years. At this point, Howard noted that OH had just released a volume from the Carter subseries, National Security Policy, 1977 to 1980, which is the first FRUS publication for 2024. As a final thought, Polaschik noted OH’s role in assisting Secretary of State Blinken’s efforts to create a more diverse, equitable, and accessible Department by co-hosting with the Operations Center the fourth annual Ebenezer Bassett Black History Month lecture on how black activism shaped U.S. foreign policy.

HAC member Kristin Hoganson followed Polaschik by asking Howard how he thought his role as Acting FSI Deputy Director might open up avenues to integrate FRUS into FSI training and enhance its role in education and training throughout the Department. Howard responded by noting that in addition to his efforts and those of other OH staffers to engage with audiences outside the Department about FRUS, he has been meeting with various audiences within the Department including Chiefs of Mission prior to their departure for posts abroad and also with Principal Deputy Assistant Secretaries. He will continue to reach out to various parts of the Department and beyond to build the OH brand.

Remarks from the General Editor

Howard then turned the session over to FRUS General Editor Kathleen Rasmussen. She began with an update on FRUS research this past quarter, which has focused on the George W. Bush (GWB) presidential records currently housed at the National Declassification Center. At present, FRUS historians are researching and copying relevant high-level records that will be of interest to all FRUS historians working on the GWB subseries. She further noted that the six-member GWB Working Group has been making significant progress drafting a preliminary outline of what the subseries will look like. The working group will be providing the HAC with an update in the closed session at the June meeting. She anticipates the working group will have a draft plan completed later in the year. In addition to this GWB subseries work, FRUS historians conducted research in two other presidential archives. One historian visited the William Clinton Presidential Library in Little Rock, Arkansas and another historian conducted classified research at the Ronald Reagan Presidential Library in Simi Valley. She noted that the classified records at these two libraries have not yet been transferred to Washington. She thanked archivist Kara Ellis at the Clinton Library and archivist Cate Sewell at the Reagan Library.

According to Rasmussen, once the compilers have finished research and document selection and annotation, FRUS volumes are edited and sent out for the declassification process. Since the last HAC meeting, three volumes have been submitted for editing and declassification. Rasmussen concluded by highlighting the recent publication of National Security Policy, 1977–1980, which documents U.S. national security interests, capabilities, and strategies during the Carter administration. She then described some of the volumes’ highlights, including the Carter administration’s nuclear strategy and key Carter administration national security officials’ deliberations on effective nuclear deterrence as well as continuity and survivability of the U.S. government in the event of nuclear attack. She stressed the hard work involved in publishing this volume and the efforts of OH and across the inter-agency: the Department of State’s Office of Information and Program Services and FRUS coordinator Keri Lewis, the CIA, the Departments of Defense, Energy, Homeland Security, and Justice, the NSC, and the Inter-Agency Security Classification Appeals Panel (ISCAP). She concluded by noting the OH FRUS staff who worked on the volume: the late Peter Kraemer, James Wilson, Chris Tudda, Carl Ashley, Heather McDaniel, Stephanie Eckroth, Mandy Chalou, Amanda Ross, and Nicole Orphanides. Once Rasmussen finished, OH Deputy Director Renée Goings added OH’s appreciation of John Powers, who, before joining OH, worked hard at the NSC to help bring this volume to publication.

Remarks from the Chief of Declassification, Publishing, and Digital Initiatives

John Powers, OH Chief of the FRUS Declassification Coordination, Publishing, and Digital Initiatives Division (DPD) then provided the HAC with an update. He began by thanking the Office of the Secretary of Defense’s Records and Declassification Division for their efforts during the pandemic and over the course of three years to eliminate a backlog of almost 1,300 records that required Department of Defense first review by adopting new policies and processes during the pandemic. They also processed an additional 1,000 new FRUS referrals during the same time period. Powers thanked J.D. Smith and his staff for this effort. Powers also thanked Smith and his staff for their help in getting the National Security Policy volume published. Powers then stated that the National Security Policy volume was the second volume published in four months. The other volume, Eastern Europe, 1981–1988, was published last December. As did Rasmussen, Powers thanked the NSC, ISCAP, and Keri Lewis for their help with the National Security Policy volume. Powers next detailed the current publishing record of the FRUS series: twenty-seven Carter volumes have been published with seven volumes remaining; the Reagan subseries has forty-two volumes either currently being researched, in declassification, or in the post-declassification process; and, finally, one George H.W. Bush volume is already in the post-declassification process, three are in declassification review, and twenty-nine are currently being researched or will be researched soon.

Powers then described his experience the past eight months since joining OH, which has included meeting with OH partners and understanding OH processes and procedures. He now better understands the complexities of publishing FRUS and recognizes the need to update certain processes, which have not kept pace with the changes in the way that government works since the FRUS statute was enacted in 1991 and revised with Executive Order 13526 in 2009. He recognizes that FRUS processes need to identify new strategies to enable greater success, and this includes incorporating new technologies into these strategies. He referenced the presentation of Kootz and Stehle about the paradigm shift taking place in IPS declassification processes. He stressed that that IPS is undertaking opportunities for OH to improve and modernize its processes. He also stressed that cooperation with OH partners is important given a climate of limited resources government wide. He finished by noting that DPD will be embarking on various pilot programs over the course of the year and will update the HAC as these programs develop.

Howard followed Powers and opened the session up for questions from those present and those attending virtually. Lentz-Smith had the first question, which she addressed to Powers and asked about the nature of the modernization he envisioned. He said he saw the necessity of being more effective and efficient in how records are exchanged between OH and the other stakeholder agencies. Powers also said there is the need to use new technologies to help OH and stakeholders make better decisions about what needs to be declassified and what shouldn’t be declassified. He added that in addition to thinking about ways to better use technology, OH was also planning to publish FRUS in smaller, more manageable batches—i.e., publishing by volume chapter as declassified, rather than as a singular whole, which necessitates waiting for all volume chapters to be declassified before publication could proceed. This, he asserted, would better assist OH and its partners who may be under budgetary and resource constraints and also help all involved in getting closer to meeting the 120-day statutory requirement for review. He further noted that in order to streamline the appeal process and to have finite decisions then process changes will need to be made in how OH refers records to partner agencies. Goings followed up Powers by detailing OH’s three-year modernization plan. The first part is digitization in order for OH to get better intellectual control over the tens of thousands of records OH has by indexing and tagging them so OH knows what records it has and the status of those records. The records could then be captured and managed in a database. OH is working with IPS to develop this technology. HAC member Hoganson then asked if OH’s modernization was focused solely within the Department happening concurrently at partner agencies. Power said this was happening across all the stakeholder agencies but acknowledged that there are some technological challenges in such cooperation as some of the agencies use systems at a different classification level than others.

No further questions followed, and Pearlstein concluded the session at 11:41 a.m.

Closed Session, March 11

Report from the Office of Information Programs and Services (IPS)

Goings introduced Mallory Rogoff at 1:00 p.m.

Rogoff introduced herself to the Committee, then introduced Marci Bayer, the Records and Information Manager of the Overseas Branch in IPS and said they would discuss their efforts to supplement the declassification of Department of State records with AI and how it can augment traditional review of documents by State reviewers. These efforts, the Department believes, will also increase transparency and make more documents available to the public. She would also discuss other records besides the cables and electronic records that Kootz discussed in Monday morning’s session. Bayer will discuss the Central Foreign Policy File for 1980 and the steps they are taking to transfer records to NARA. Rogoff said IPS is exploring digitization "buy-in" options for bureaus to fund their own digitization efforts, which would help facilitate digitization since funding would come from those bureaus rather than from the Bureau of Administration.

Bayer provided an update on the 1980 Central Foreign Policy File. She noted that IPS had started joint activities with NARA to transfer the 1980 Central Foreign Policy File. IPS worked with NARA’s electronic accessioning unit on a plan to transfer the electronic files and engaged with the National Declassification Center to gather the declassification decisions on the P and N Reels.

Rogoff provided an overview on some transparency initiatives that the Records Management Program (A/GIS/IPS/RA) is working on in the context of M–23–07. She noted that the Declassification Program is having great success utilizing AI tools to make review determinations on cables that is already increasing transparency, and then followed up with the status of paper records. She provided an overview of the Record Program's annual Senior Agency Official for Records Management (SAORM) Report—a public report. On the SAORM, she said that the Department will not meet the goal to "transfer all permanent records electronically" to NARA prior to June 2024. This was because digitization on a mass scale is a huge challenge for IPS and resource intensive. She also noted that some of these paper records are valuable in and of themselves and don’t really lend themselves to be digitized and then destroyed. The Department will therefore be submitting an exemption request as allowed by M–23–07 on the basis that digitizing all of the Department's paper record holdings will be overly burdensome and the intrinsic value in a variety of Department record types would be lost via digitization. Some of these record types include: maps, treaties, historical documents, senior officials' notebooks, and handwritten notes.

Schake asked when Rogoff will report that the Department will be unable to comply with the law and whether she will able to predict when the Department will come into compliance. Rogoff replied that they hoped that the exemption they are requesting will be approved, which would then substitute for compliance.

McGovern asked whether NARA has been asked about the carbon footprint that will be created by all the digitization initiatives. Rogoff replied that IPS hoped that the importance of preserving so many unique paper records would eliminate the carbon footprint concerns. She thinks NARA understands this and will agree to the exemption.

Hoganson asked about the path forward on post-1980 P Reels. She recalled what Rogoff said in a previous meeting about someone determining that P Reels were not of high value and said that some 4,800 P Reels have yet to be transferred. There seems to be a disagreement about what to do with these important historical records.

Rogoff acknowledged the disagreement and how IPS can digitize these incredibly resource-intensive records. IPS’s liaison to the NDC indicate that the P Reels are redundant to cables they have already reviewed. That’s where the determination that they were not high value records was made. The low appetite to process those records comes from that decision.

Hoganson asked whether they will ever be transferred. Rogoff said she recognizes the dilemma and wants to transfer some of the material, but she wants NARA to compromise and recognize that transferring some records is better than none.

Botts asked if the cost of digitization falls, and releasing other records increases, will IPS reach the point where digitizing is more cost effective? Rogoff said that yes, it is likely that the more valuable and high-level, born-digital electronic records will be easier to do if IPS can maximize and refine AI reviewing. The fewer paper records going forward will make it easy.

McGovern said electronic records are a good test bed for AI and other reviews, but noted there are going to be quality assurance issues. AI is not a magical tool that will solve all declassification problems. Rogoff agreed with this assessment.

Goings thanked Rogoff and Bayer for appearing before the Committee.

Report from the National Declassification Center

Howard then introduced Bill Fischer. Fischer reported on the consolidation of holdings of the Presidential Libraries to the National Declassification Center. As of November 2023, Johnson and H.W. Bush records had been moved. This leaves records in three libraries: Nixon, Reagan, and Clinton. The budgetary requirements and the need to schedule and coordinate moving with the Air Force further complicates the process.

Naftali, who was absent, sent a written question asking if archivists would be able to continue to work on processing and releasing materials. Fischer said yes and also explained that the NDC had been holding some returns during the pandemic that were now going back to the libraries. He also commented that retaining the records would give the library archivists more time to separate out unclassified materials that could remain in the reading rooms.

Hoganson asked about the implications for OH if the records remain at the libraries. Fischer responded that the libraries should be able to take FRUS researchers. Hoganson followed up asking that if the goal is access for researchers, how is moving records to the NDC versus the libraries improving access. Fischer explained that the libraries could only process so much and it is more efficient at the NDC.

Howard asked if the NDC was working with the libraries to bring any archivists to DC to help NDC understand the records and processes. Fischer said NDC would welcome TDY staff but this was not currently planned. McIlwain also commented that NDC staff were in touch with the library archivists and had one POC for each library to help them understand the records as needed. He stated NARA was hoping to segregate out unclassified materials from the Bush 43 records and return them to the libraries for public access in an expeditious manner. Hoganson asked how the NDC prioritized which records to process. Fischer explained that they knew some things would be “quick wins.” He discussed donated collections that needed review, one example being the Rostow papers at the Johnson Library, and that unclassified Obama records had never been sorted. He also mentioned that an NSC waiver could be applied to a number of Bush 43 holdings. He introduced Brewer Thompson who had NATO expertise and would be working on some Eisenhower collections with NATO equities.

Goings welcomed Thompson to her new position and thanked her for her work assisting the office.

Pearlstein asked about the possible use of the DOS pilot AI system, as described during the IPS presentation. He said they are open to the idea but would need a system to host the product. Fisher noted that items like OCRed cables might be a good subject for an AI application.

Howard asked if the presidents have a say in whether their records stay with their libraries or remain with NARA in Washington. Fisher stated that ex-presidents would likely have a say about where unclassified materials ultimately resided but that all classified material would remain in Washington going forward. Langbart explained there were a number of rules for presidential libraries, for example the ex-president must raise 60% endowment funds (up from an original 20%) to support library maintenance.

Rasmussen asked for an NDC staffing update. Fischer reported on the vacancies filled. Pearlstein requested that Fischer send the job announcements to the HAC if they were having issues filling positions as they could publicize the openings more widely. Fisher noted that the entry level positions had lots of promotion potential and would be excellent opportunities for newly minted archival specialists.

Closed Session, March 12

Planning for the Clinton Subseries The General Editor and members of the Clinton plan working group briefed the committee and the office on the scope of the Clinton FRUS plan, and the edits the plan has undergone since its first inception.