A Way-Too-Close Reading of the SLMath Director Search

Two SLMaths

There are two Simons Laufer Mathematical Sciences Institutes in the May 2026 Director Leadership Profile, the eleven-page document inviting applications and nominations for the institute’s next director.

The first is the SLMath the document describes, and it sounds, to me, troubled. It faces “existential changes in federal support.” Its rhetoric is mid-pivot toward “applied mathematics, data science, and AI.” It must steward “deep community care” against unspecified operational shortcomings. The next director must balance excellence with innovation, scientific ambition with financial resilience, fundamental mathematics with emerging tools, and continuity with change. Across the role and profile sections, the word “balance” appears once and “balancing” appears three more times. “Preserve while” constructions add to the hedging.

The second SLMath is the one running the programs. Through Fall 2028, every semester program — the institute’s flagship long-form research residencies — is recognizable core mathematics. Spring 2024 was commutative algebra and noncommutative algebraic geometry. Fall 2024 was curvature and special geometric structures. Spring 2025 was extremal combinatorics and probability of discrete structures. Fall 2025 was kinetic theory and stochastic PDEs. Spring 2026 was low-dimensional topology and discrete subgroups of higher-rank Lie groups. Fall 2026 will be motivic homotopy theory and representation theory under quantum field theory. Fall 2027 will be algebraic combinatorics, tropical geometry, and Artin groups and arrangements. Fall 2028 will be inverse problems and geometric measure theory. The calendar is overwhelmingly theoretical mathematics, with a handful of applied-facing exceptions. None of it is artificial intelligence. None of it is data science. This is the program calendar of a serious pure-mathematics convening center doing what it has done for forty years.

The gap between those two SLMaths is what the rest of this essay is about. The institute as it operates is not in obvious crisis. But the institute as it represents itself in this search profile reads, to me, as anxious, defensive, and reaching for a vocabulary that does not match what it actually does.

I. The genre is the message

The leadership profile is eleven pages, contains five photographs, and is produced by WittKieffer, a broad executive search firm whose academic practice includes presidents, provosts, deans, athletic directors, and senior administrators across higher education. Previous MSRI director searches were run internally. This one was not.

The 2013 return of David Eisenbud, who had directed the institute from 1997 to 2007, appears to have proceeded through institute governance rather than through a public executive-search profile. A separate search announced in 2020, which produced Tatiana Toro’s 2021 appointment, was conducted by a trustee committee chaired by Edward Baker. The 2017 m/Oppenheim search at MSRI used an outside firm only for the Director for Advancement and External Relations, a fundraising role where outside firms are conventional. Using WittKieffer for the Director search itself is new.

The form has changed with the firm. The 2020 MSRI Director Search Announcement is roughly four pages. It contains no photographs. It is organized as a position description, with sections called General Responsibilities, Specific Responsibilities, Attributes, and Direct Reports. The Specific Responsibilities are concrete: the 2020 director was to lead the Scientific Advisory Committee on programs and the Human Resources Advisory Committee on participation. The 2026 document is eleven pages, contains five photographs, and is organized around themes: Strategic Direction, Financial Sustainability, Modernization, Convener, Staff, Operations. The concrete committee leadership of 2020 has been absorbed into broader strategic and accountability language. In my reading, the 2020 document told the candidate what the job was. The 2026 document tells the candidate what the institute is going through. Other math institutes post short position descriptions. This is what happens when you hire a search firm: you get a document written for someone who has never heard of SLMath. That may be the point — but it is, I think, a choice about who the institute thinks its next director is.

II. The impossible director

The document asks for one person who is simultaneously an internationally recognized research mathematician, the institute’s principal scientific advocate to NSF and the National Security Agency, a philanthropic fundraiser, a public communicator, and a strategic planner who will modernize operations. The same person must be a culture builder with strong emotional intelligence and an active figure in the global mathematical community. They must also be an AI-aware strategist, a champion of K–12 outreach, a Board partner on governance and trustee recruitment, and a person of “deep community care” who will foster trust, collegiality, and shared purpose across the staff. The compensation: $350,000 to $425,000 per year, on a five-year term with possibility of renewal.

I think the list is overloaded. The intersection of those qualifications in living mathematicians is, I’d guess, small. The subset willing to take a five-year administrative term at that salary, while presumably maintaining a tenurable appointment at Berkeley, is smaller still. It seems to me that the document is not really describing a person. It is enumerating a strategic plan and giving the plan a human-shaped header. Every unresolved tension in the institute’s recent thinking feels loaded onto one job description.

The five-year term is also worth noting. David Eisenbud served as MSRI director across roughly two decades in two stints. Brian Conrey has long led AIM. Brendan Hassett has run ICERM since 2016, and NSF just renewed his institute through 2030 with $16.5 million. Dimitri Shlyakhtenko has been at IPAM since 2017. Kevin Corlette has led IMSI since its founding in 2020. I found no comparable public Director search currently underway at the other established NSF math institutes. The five-year-with-renewal frame, presented as neutral, seems to accommodate a job description it would be hard to inhabit for much longer than five years.

III. Artificial intelligence is the bet, packaged as balance

The document names artificial intelligence three times in the role and profile sections, four in the full PDF. It refers obliquely to “emerging tools and directions,” “emerging methodologies,” and “new tools and perspectives” several more times. Every reference is hedged. The director must show “deep scientific judgment about how new tools and directions should inform, complement, or reshape fundamental mathematics.” Strategic decisions about AI must “strengthen rather than dilute” the institute’s foundational commitments. Even curiosity is qualified: the document seeks “intellectual curiosity and openness to innovation, including sustained engagement with emerging technologies.”

AI is everywhere in mathematics right now — in the funding landscape, in the discourse, and increasingly in the research itself. I suspect SLMath genuinely wants to move in this direction, not merely gesture at it. That makes the document’s hedging more concerning, not less. The semester programs through Fall 2028 are core mathematics. None is artificial intelligence. None is data science. The 2020 MSRI Director Search Announcement does not contain the phrases artificial intelligence, machine learning, or data science. Something changed between 2020 and 2026, but the program calendar did not.

The actual AI work lives in a separate format. The new month-long AxIOM (Accelerating Innovation in Mathematics) program track, debuting in Spring 2027, has four inaugural programs. Two are explicitly AI or AI-adjacent: “Machine Learning for Mathematics” (March 1–26, 2027) and “Building the Mathematical Library of the Future” (March 15–April 9, 2027), which appears devoted to formal methods, proof assistants, and Lean-style mathematical libraries. The institute is partitioning AI work into a parallel format that runs alongside, rather than within, the semester programs that carry its identity. The leadership profile does not name that partition or explain how AxIOM relates to the institute’s core identity. If SLMath is genuinely beginning an AI pivot, the community deserves to know what that means for the place that has been, for forty years, a pure-mathematics convening center.

NSF already has an institute for this. In August 2025, NSF launched ICARM at Carnegie Mellon, the Institute for Computer-Aided Reasoning in Mathematics, whose stated mission is to help mathematicians “effectively integrate artificial intelligence, machine learning, formal methods and automated reasoning into their research.” SLMath has historically been the fundamental-mathematics institute. If it now positions itself as AI-relevant, it is either competing with ICARM for the same federal mandate, carving out a different niche, or speaking to non-federal funders for whom the NSF institute landscape is less salient. Its one AI-facing event so far is a joint workshop with the Simons Institute for the Theory of Computing — partnering with external AI capacity rather than building its own.

The most honest version of what the document could have said: We believe AI tools are becoming important to mathematics, and we are beginning to host that work alongside our core programs. Here is what that means for the institute’s identity going forward. The document never says anything that direct. It hedges every reference, asks the next director to exercise “deep scientific judgment” about the balance, and leaves the community to guess whether the institute is genuinely pivoting or just positioning itself for funders. I think the math community should be asking that question out loud. The country’s flagship pure-mathematics convening institute is signaling a shift it will not plainly describe.

IV. The quiet demotion of equity infrastructure

The most documentable critique of the leadership profile is what has been removed from the role itself.

The 2020 MSRI Director Search Announcement listed, as one of nine required Attributes of the next director:

Persistent and deliberate commitment to action with respect to inclusivity, equity, inclusion, and justice.

That attribute has been entirely removed from the 2026 document. None of the words inclusivity, equity, inclusion, or justice appears anywhere in the 2026 role responsibilities or candidate criteria. The few places those words survive in the 2026 PDF are atmospheric. Inclusion appears once in the About section in the phrase “intentional inclusion of graduate students, postdoctoral fellows, and early-career faculty alongside senior leaders.” Inclusivity and equity appear only in the Berkeley city-description page, in a paragraph about the city’s civic character. In 2020 the institute committed the director to action on equity. In 2026 the institute relocated those words to a description of where the institute is geographically situated.

A second documentary loss: the 2020 document tasked the director with leading a Human Resources Advisory Committee “in developing policies and activities to recruit, support, and increase participation and contributions of historically marginalized groups.” Neither the HRAC nor the phrase historically marginalized groups appears in the 2026 document. The 2026 Direct Reports list contains no committee chair role. The 2026 Core Responsibilities list contains six bullets, none of which mention equity, access, broadening participation, diversity, inclusion, or any cognate concept.

The closest 2026 language has been displaced into the Personal Qualities section as a candidate quality: “Demonstrates a persistent commitment to developing the U.S. STEM workforce and to expanding opportunity within the mathematical sciences.” To me, this is the language of federal compliance reporting. The word underrepresented appears once in the entire document, in the About section, in a sentence about Summer Graduate Schools as programs “supporting underrepresented groups in mathematics.”

The structure of the displacement matters. A Core Responsibility is something the institution holds the role accountable for. An Attribute (in 2020) was something the director was required to demonstrate. A Personal Quality (in 2026) is less directly accountable than either. Between 2020 and 2026 the institute has moved its explicit equity commitment from an Attribute the director had to demonstrate, plus a committee the director had to lead, to a Personal Quality the candidate is asked to bring. The work becomes less clearly attached to the director’s formal accountability.

This matters because the substantive work is real. MSRI-UP, the institute’s undergraduate research program, has run for more than a decade. Each summer it brings a cohort averaging about seventeen undergraduates to the Berkeley campus for six weeks of intensive research with faculty mentors, and connects participants with peers and mentors who support graduate-school pathways. For many students, MSRI-UP is not outreach branding. It is the bridge from undergraduate mathematics into the research profession. The most recent published evaluation covers 2007–2018 and tracks 205 alumni. Among the 184 alumni for whom degree-status data was available, 157 — 85% — had enrolled in or completed graduate programs in the mathematical sciences. Survey respondents were overwhelmingly from underrepresented racial and ethnic groups in mathematics, and nearly half identified as female. By these measures, MSRI-UP is one of the most consequential undergraduate research programs for underrepresented students in any American mathematical sciences institution.

ADJOINT — the African Diaspora Joint Mathematics Workshops — is more recent. The 2023 program brought together four research groups comprising twelve researchers, who reported a mean recommendation score of 4.82 of 5 and a mean continue-the-program score of 4.91 of 5. The 2024 evaluation of the MSRI postdoctoral fellowship for 2016–2023 documents 81% of postdocs rating their experience excellent or very good, with comparatively strong representation of women and underrepresented minority fellows relative to national doctoral baselines.

The 2026 leadership profile names none of these programs. The phrase MSRI-UP does not appear. The phrase ADJOINT does not appear. The phrase African Diaspora does not appear.

SLMath has not stopped doing equity work. But between 2020 and 2026, explicit equity accountability disappeared from the role criteria for the institute’s leader. The timing is hard to separate from the federal climate around DEI language. Adapting to political risk is a choice an institution can defend. Removing the commitment and hoping no one compares the two documents is, in my opinion, just retreat.

The 2021 MSRI press release announcing Toro’s appointment said that addressing issues of equity and inclusion of underrepresented groups in the mathematical sciences had been “a guiding principle” in every setting where she had worked. The 2020 search document, written to recruit her, named that principle as a required attribute. The 2026 search document, written to replace her, has removed it from the role.

V. The financial alarm is real, but its target is not where the document points

The most extraordinary phrase in the leadership profile sits in the Strengthen Financial Sustainability section: “In addition to stewarding SLMath’s major NSF grant renewals, think proactively about long-term scenarios for sustaining SLMath’s mission and independence, including planning for the possibility of existential changes in federal support.”

In my assessment, the actual financial picture does not support that level of alarm. In 2017, MSRI’s endowment was approximately $23 million against an operating budget of $9.9 million. The m/Oppenheim search at the time set a goal of $100 million by 2030. In 2022, James and Marilyn Simons and Henry and Marsha Laufer made a $70 million unrestricted gift, described in the announcement as the largest unrestricted endowed gift to a U.S.-based mathematics institute. The gift effectively achieved the 2030 endowment target eight years early. The institute described as facing “existential” funding risk is among the best-endowed mathematics institutes in the country.

The operating budget tells a different story. The 2023–24 SLMath annual report shows that 52% of program-member funding came from NSF, 3% from NSA, 11% from endowments, and 34% from private sources. For short-term scientific activities, NSF’s share rose to 73%. At a standard 4% draw, a $100 million endowment yields roughly $4 million per year — meaning that endowment income alone cannot replace NSF-supported activity without changing the operating model. A healthy endowment does not eliminate operating risk. The current NSF cooperative agreement, DMS-2424139, covers 2025 through 2030. The next director will preside over the proposal cycle that determines federal support from 2030 onward.

The institute is endowment-rich and operating-budget-vulnerable. But the document will not say which federal changes it means, or who in the institute would be most affected. It just says “existential changes” and leaves the next director to figure out, in public, what the search committee would not put in writing.

The document also asks the director to remain “attentive to the values and sensitivities of the mathematical community regarding funding sources and institutional independence.” But it does not say whose values, or which sensitivities. The NSA, which funds ADJOINT and appears in several SLMath budget lines, has made mathematicians uncomfortable for decades. Corporate tech money comes with strings. Defense-adjacent foundations raise their own questions. The document names none of these. It tells the candidate that people have feelings about funders, without saying which funders or which feelings.

The $70 million Simons-Laufer gift was publicly framed as strengthening long-term independence. The leadership profile reads to me as if the institute is being pushed toward more donor dependence, not less — toward new philanthropy, corporate partnerships, and a pitch to funders who are not NSF. The document does not say why.

VI. What else the document does not say

The leadership profile also leaves several questions conspicuously unanswered.

The first unanswered question concerns Tatiana Toro. Her five-year appointment runs through summer 2027. She is, by every public measure, completing her tenure in good standing. She received the 2025 Solomon Lefschetz Medal during her directorship. The institute has functioned without visible governance failure. The current NSF cooperative agreement was awarded on her watch. The leadership profile makes no reference to her. It does not thank her, characterize her tenure, or specify whether the institute is in a phase of continuity or change relative to her vision. At other institutions, search documents often acknowledge the outgoing leader, in part because doing so tells the new leader where they stand.

A second unanswered question concerns the simultaneous Deputy Director search. That search was launched in February 2026 by Terence Tao, in his capacity as Vice-Chair of the SLMath Board of Trustees. Two senior leadership searches running in the same cycle is notable. The current Deputy Director’s status and continuity are not addressed in the Director profile, even though the Director profile promises the new Director will work “in close partnership with the Deputy Director.”

A third unanswered question concerns the “modernization” language. The profile asks the next director to “lead the next phase of organizational maturation by streamlining administrative processes, modernizing systems, and improving data and documentation practices.” It also asks for “transparency around roles, decision-making, and accountability” and for reinforcement of SLMath as “a people-centered, values-driven workplace.” In my experience, in a search document, language like this usually means something went wrong, not that the institute is dreaming big. The document does not say what. It frames the issues as personality traits the next director should have, which leaves the reader to guess whether there are specific institutional problems behind the ask.

A fourth unanswered question concerns the Board itself. The trustees appear in the document as a constituency the director must “work closely with” and “engage,” but no trustee voice says what the trustees actually want the institute to become. The board’s composition is worth noting. Three current trustees — Vincent Della Pietra, Stephen Della Pietra, and Henry Laufer — are listed under Renaissance Technologies, the quantitative hedge fund founded by Jim Simons. Marilyn Simons, also on the board, represents the Simons Foundation. The institute is named after two Renaissance Technologies families. I believe a document that asks its next director to remain “attentive to the values and sensitivities of the mathematical community regarding funding sources” might start by acknowledging the concentration of a single hedge fund’s alumni and donors on its own governing board.

Maybe these are just conventions of executive search documents. But the result is a document that asks the math community to evaluate a leadership transition without giving the community the information needed to do so.

VII. The community’s responsibility

To be fair, the drafters of the 2026 profile faced real constraints. The federal environment around explicit DEI language has become legally and politically hazardous in ways that did not apply in 2020. A broader search firm can extend the candidate pool, professionalize the process, benchmark compensation, and support candidate confidentiality. A longer document allows fuller description of the institute’s current strategic situation. The AxIOM format may be experimental rather than evasive — a way to host AI work without permanently committing the semester calendar. The financial-sustainability language reflects a real concern that NSF-dependent institutes face uncertain futures regardless of their endowment health. The “modernization” language may name the kind of institutional improvements any organization undertakes after a leadership transition. Read this way, the 2026 profile is a prudent adaptation to changed circumstances, and the document’s silences are professionalism rather than evasion.

Two things complicate that reading. First, the equity-attribute removal is not a matter of interpretation: the 2020 document said the director must show “persistent and deliberate commitment to action with respect to inclusivity, equity, inclusion, and justice,” and the 2026 document does not. Adapting to political risk is defensible. Doing it silently is a different kind of choice. Second, there is a gap between the institute’s anxious tone and its actual condition. An institute with a $100 million endowment, the current NSF renewal in hand, a celebrated outgoing director, and a four-decade convening record is not in obvious crisis. The document strikes me as more anxious than the institute’s actual situation warrants.

The math community tends to treat leadership decisions like these as inside baseball. Boards of trustees handle these matters. Search committees take soundings. Press releases announce outcomes. The community responds with congratulations or, occasionally, with private grumbling. Public criticism is rare.

This norm has costs. The current leadership profile is the product of a process mathematicians as a body did not shape and will not formally evaluate. The board chose to hire through a national executive-search firm rather than through a peer-led trustee committee. It chose to organize the document around AI and operational modernization rather than around mathematical vision. It chose to leave MSRI-UP and ADJOINT unnamed. It chose to remove the explicit equity-commitment attribute that its own 2020 search document required. It chose to keep silent on the outgoing director and on the simultaneous Deputy search. It chose to frame financial sustainability in language detached from the actual endowment picture. The community will live with these choices for at least five years. That, I think, is worth noticing.

After targeted searches of their public news and announcement pages, I found no public statement on this search from the American Mathematical Society, the Society for Industrial and Applied Mathematics, or the Mathematical Association of America. I found no joint statement from the community of NSF math-institute directors on the changing role of the institutes in the AI moment or on the federal political environment they collectively face. The closest thing to a public conversation is Terence Tao’s blog post announcing the Deputy Director search, written in his Board capacity, which contains no commentary on strategic direction.

The next director, whoever they prove to be, will inherit the contradictions this document declines to resolve. They will have to decide what proportion of the institute’s identity is AI-shaped and what proportion remains pure-mathematics-shaped. They will have to decide whether MSRI-UP and ADJOINT remain core programs or quietly become discretionary. They will have to decide what to say publicly about NSF, NSA, the Simons Foundation, and the political environment the document has named only as “existential.” They will have to decide what it means to lead a mathematical institute at a moment when the community has not done the work of saying, in public, what its institutes should be doing.

The leadership profile is not the place that conversation will happen. It is the document that shows how thin the public conversation has been.

Coda

A close reading reveals an institution that has done many things right. It has built a roughly $100 million endowment. It has sustained a serious pure-mathematics convening function across four decades. It has produced what I consider one of the country’s most consequential undergraduate-research pipelines for students from underrepresented groups in mathematics. It made the pathbreaking appointment of Tatiana Toro. It has earned the renewed confidence of the National Science Foundation. The actual institute, I believe, is in better shape than its own search document admits.

The search document describes a different institute. Between 2020 and 2026 it removed equity from its director’s required attributes. It added AI to its rhetoric without adding it to its semester programs. It replaced a four-page position description with an eleven-page leadership profile. It produced a search document that names funding risk as existential while sitting on the largest unrestricted endowed gift to a U.S.-based mathematics institute. My sense is that the mathematical community has stopped asking its institutes to describe themselves in mathematics’ own voice. The institutions have, accordingly, stopped doing so. The next director will speak whatever language the community asks of them. The leadership profile is the early evidence of which language is being asked.