arxiv: 2605.12514 · v1 · submitted 2026-04-02 · 💻 cs.SI · cs.CV· cs.CY· cs.DL· stat.AP

Recognition: 2 theorem links

· Lean Theorem

Structural Diversity Drives Disruptive Scientific Innovation

Yichun Peng , Saike He , Peijie Zhang , Kang Zhao , Yi Yang , Ning Zhang , Qingpeng Zhang , Daniel Dajun Zeng

show 1 more author

Hao Peng

Authors on Pith no claims yet

Pith reviewed 2026-05-14 22:14 UTC · model grok-4.3

classification 💻 cs.SI cs.CVcs.CYcs.DLstat.AP

keywords structural diversitydisruptive innovationscientific collaborationteam assemblycollaboration networksdisciplinary integrationcausal inferenceteam size

0 comments

The pith

Structural diversity in teams predicts and drives disruptive scientific innovation.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper defines structural diversity as the degree to which a research team bridges multiple distinct knowledge communities through its members' prior collaborations. It shows that this measure reliably forecasts whether the team's output will count as disruptive, beating standard indicators such as the presence of new members or sparse collaboration edges. Using a dataset of 260 million papers and a 2012 NSF policy shift as a natural experiment, the authors demonstrate that structural diversity also reverses the usual penalty of larger team size by turning scale into an advantage for combining ideas. The mechanism is disciplinary integration: higher-diversity teams more readily assemble heterogeneous knowledge into new configurations. This positions team architecture as a design lever for producing breakthroughs rather than an afterthought.

Core claim

Structural diversity, measured as the extent to which a team connects distinct knowledge communities in its prior collaboration network, is a strong causal driver of disruptive innovation. It outperforms team freshness and edge density, interacts positively with team size to offset the curse of scale, and operates through the mechanism of disciplinary integration that allows heterogeneous knowledge to be recombined into novel configurations.

What carries the argument

Structural Diversity (SD), the extent to which a team bridges multiple distinct knowledge communities within its prior collaboration network, which enables disciplinary integration and creative synthesis while mitigating scale penalties.

If this is right

Larger teams produce more disruptive work when they exhibit high structural diversity rather than suffering the usual scale penalty.
Teams with elevated structural diversity achieve greater disciplinary integration, recombining knowledge from separate communities into novel configurations.
Structural diversity outperforms simpler novelty proxies such as team freshness and edge density in predicting disruption.
Organizational choices in team assembly can systematically increase the rate of disruptive discoveries by maximizing bridges across knowledge communities.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Funding agencies could incorporate structural-diversity metrics when evaluating proposed team compositions to favor projects likely to yield breakthroughs.
The same bridging logic may apply to non-academic innovation settings such as corporate R&D consortia or open-source software projects.
Real-time collaboration platforms could surface structural-diversity scores to help scientists assemble teams that deliberately span knowledge communities.

Load-bearing premise

The 2012 NSF policy change creates a clean quasi-natural experiment that isolates structural diversity's causal effect on innovation without confounding from unobserved team traits or field shifts.

What would settle it

If post-2012 NSF-funded teams show no measurable rise in structural diversity or no corresponding increase in disruptive citations relative to pre-policy teams, after accounting for clustering choices, the causal claim would be falsified.

Figures

Figures reproduced from arXiv: 2605.12514 by Daniel Dajun Zeng, Hao Peng, Kang Zhao, Ning Zhang, Peijie Zhang, Qingpeng Zhang, Saike He, Yichun Peng, Yi Yang.

**Figure 1.** Figure 1: SD as a robust predictor of disruptive scientific innovation. (A-B) Illustration of how SD captures the potential of distinct knowledge communities for innovation. (C) Each point represents one discipline (19 in total). The dashed line shows a statistically positive correlation between the mean SD (standardized) and the mean CD Index (field-normalized) across all areas of science. Bubble size is proportion… view at source ↗

**Figure 2.** Figure 2: Disciplinary heterogeneity of SD and the mitigation of “Curse of Scale”. (A) Heterogeneous effects of SD on disruptive innovation across 19 disciplines. Each horizontal bar shows the regression coefficient (p < 0.001 for all 19 disciplines) between SD (standardized) and CD Index (field-normalized) by fitting the same regression model in [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

**Figure 3.** Figure 3: Robustness test using the Propensity Score Matching framework further supports the positive association between SD and disruptive innovation. The analysis is based on 690, 972 Computer Science papers. (A) Kernal density estimation of the CD Index for treated teams (Top quartile SD, orange) and their matched control teams (Bottom quartile SD, blue). The distribution of CD Index for high-SD teams is statisti… view at source ↗

**Figure 4.** Figure 4: A quasi-natural experiment based on the NSF policy change in 2012 supports the correlation between SD and disruptive innovation. (A) The percentage of teams by the number of Connected Components (CC) before (2010-2011, N = 18, 267) and after (2012-2013, N = 23, 191) the NSF policy implementation. All papers (N = 41, 458) were NSF-funded in the AMiner dataset. The inset highlights a sharp increase in teams… view at source ↗

**Figure 5.** Figure 5: Disciplinary Integration (DI) mediates a sizable fraction of the effect of SD on disruptive innovation. (A) Path diagram linking team SD to their CD Index via disciplinary integration. SD significantly predicts DI (β = 0.153, p < 0.001), which in turn predicts CD Index (β = 0.028, p < 0.001). About 18% of the effect of SD on CD (total effect size: 0.024, p < 0.001) can be explained by DI (indirect effect s… view at source ↗

read the original abstract

Scientific innovation increasingly depends on collaboration, yet the organizational structure that fosters breakthrough ideas remains poorly understood. Existing metrics - such as team size or compositional diversity - capture readily observable characteristics but not the deeper architecture of collaboration. We introduce Structural Diversity (SD): the extent to which a team bridges multiple distinct knowledge communities within its prior collaboration network. Using a century-scale dataset of 260 million scientific publications (1900-2025) and combining causal inference with a quasi-natural experiment based on a U.S. National Science Foundation policy change in 2012, we show that SD is a powerful and robust predictor of disruptive innovation, outperforming traditional team novelty indicators such as team freshness and edge density. Moreover, SD positively interacts with team size and is able to mitigate the well-known "curse of scale" by transforming scale from a liability into a resource for creative synthesis. We find that one mechanism underlying this effect is Disciplinary Integration (DI): teams with higher SD can more effectively combine heterogeneous knowledge into novel configurations. Our findings position SD as both a new theoretical construct and an actionable design principle for organizing scientific collaboration. By linking the architecture of team assembly to the dynamics of creative discovery, our work offers a structural explanation for how collective intelligence can be systematically engineered to foster disruptive innovation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Structural Diversity is a new network bridging measure that the paper claims predicts disruptive innovation and offsets large-team downsides, but the 2012 NSF policy shock leaves room for confounding and the community detection step looks sensitive.

read the letter

The main takeaway is that this paper defines Structural Diversity as the degree to which a team's prior collaboration network bridges distinct knowledge communities, then shows it outperforms standard novelty metrics and interacts with team size to reduce the curse of scale. They support this with a century-long dataset of 260 million papers and a quasi-experiment around the 2012 NSF policy shift, plus a proposed mechanism called Disciplinary Integration. The scale of the data and the shift from team composition to prior network structure are the clearest strengths; the construct feels distinct from existing freshness or edge-density measures. The attempt to tie network architecture to actual knowledge recombination is also a step forward. The identification strategy is the softest part. The policy change may have altered funding levels, team composition rules, or field incentives at the same time, so it is not obvious that the estimates isolate structural diversity rather than those other factors. Community detection for the knowledge communities is another point of fragility; different algorithms or resolution parameters can reassign teams and change which ones count as high-diversity, yet the abstract gives no sign of stability checks across alternatives. Without those, the reported superiority could partly reflect measurement choices. The work is aimed at people who study scientific collaboration networks and science policy. A reader already working on team assembly or innovation metrics would find the new angle worth thinking through, even if the causal piece needs more work. It should go to peer review. The question is policy-relevant, the data effort is substantial, and the core idea is fresh enough that referees can usefully pressure the identification and robustness sections.

Referee Report

3 major / 3 minor

Summary. The paper introduces Structural Diversity (SD) as the extent to which a scientific team bridges multiple distinct knowledge communities in its prior collaboration network. Using a dataset of 260 million publications spanning 1900-2025 and a quasi-natural experiment based on the 2012 NSF policy change, it claims that SD is a robust predictor of disruptive innovation that outperforms traditional metrics such as team freshness and edge density, positively interacts with team size to mitigate the 'curse of scale,' and operates in part through Disciplinary Integration (DI) by enabling more effective combination of heterogeneous knowledge.

Significance. If the central claims hold after addressing identification and measurement issues, the work would be significant for advancing the science of science by providing a new network-structural construct that explains breakthrough innovation beyond compositional diversity. The large-scale data, combination of predictive and causal approaches, and actionable implications for team assembly represent strengths that could influence both theory and practice in organizing collaborative research.

major comments (3)

[Identification strategy / Methods] The quasi-natural experiment in the identification strategy section relies on the 2012 NSF policy change as an exogenous shock to SD, but provides no explicit DiD specification, controls for concurrent changes in funding amounts or team eligibility, or placebo tests; this leaves open the possibility that residual confounding from field-specific incentives or unobserved team characteristics drives the reported effects rather than SD itself.
[Definition and measurement of SD] SD is operationalized via community detection on the prior collaboration network to identify 'distinct knowledge communities,' yet no robustness checks across algorithms (e.g., Louvain vs. modularity maximization) or resolution parameters are reported; this choice directly affects team classification and could artifactually inflate the superiority of SD over edge density or freshness in the predictive regressions.
[Results / Predictive models] The results section claims SD outperforms traditional novelty indicators and mitigates the curse of scale via positive interaction with team size, but without specific tables showing coefficient magnitudes, incremental R², out-of-sample prediction metrics, or stability after alternative community partitions, the comparative advantage remains unsubstantiated.

minor comments (3)

[Abstract] The abstract states a dataset through 2025; clarify the actual endpoint of the data collection and any forward-looking elements.
[Introduction / Conceptual framework] Introduce formal equations for SD and DI early in the text rather than relying on verbal description, and ensure all variables are defined before first use.
[Figures] Network figures should include legends for community labels and sensitivity to parameter choices to aid interpretability.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which have helped us identify areas where the manuscript can be strengthened. We address each major comment below and commit to revisions that directly respond to the concerns while preserving the core contributions of the work.

read point-by-point responses

Referee: [Identification strategy / Methods] The quasi-natural experiment in the identification strategy section relies on the 2012 NSF policy change as an exogenous shock to SD, but provides no explicit DiD specification, controls for concurrent changes in funding amounts or team eligibility, or placebo tests; this leaves open the possibility that residual confounding from field-specific incentives or unobserved team characteristics drives the reported effects rather than SD itself.

Authors: We agree that greater transparency in the identification strategy is warranted. In the revised manuscript we will present the complete DiD specification, add controls for changes in funding amounts and team eligibility, and report placebo tests on pre-policy periods and unrelated fields. These additions will directly mitigate concerns about residual confounding and reinforce that the effects operate through the policy-induced change in structural diversity. revision: yes
Referee: [Definition and measurement of SD] SD is operationalized via community detection on the prior collaboration network to identify 'distinct knowledge communities,' yet no robustness checks across algorithms (e.g., Louvain vs. modularity maximization) or resolution parameters are reported; this choice directly affects team classification and could artifactually inflate the superiority of SD over edge density or freshness in the predictive regressions.

Authors: We will add a dedicated robustness section that replicates all main results using the Louvain algorithm, alternative modularity-maximization implementations, and a range of resolution parameters. The revised manuscript will show that SD retains its predictive advantage and its interaction with team size across these specifications, confirming that the findings are not artifacts of the chosen community-detection procedure. revision: yes
Referee: [Results / Predictive models] The results section claims SD outperforms traditional novelty indicators and mitigates the curse of scale via positive interaction with team size, but without specific tables showing coefficient magnitudes, incremental R², out-of-sample prediction metrics, or stability after alternative community partitions, the comparative advantage remains unsubstantiated.

Authors: We will expand the results section with new tables that report coefficient magnitudes, incremental R² values, out-of-sample prediction metrics (including cross-validated performance), and replication under alternative community partitions. These additions will provide quantitative documentation of SD's comparative performance and its role in offsetting the curse of scale. revision: yes

Circularity Check

0 steps flagged

No circularity: SD defined from network structure and tested via external policy shock

full rationale

The paper defines Structural Diversity (SD) as the extent to which teams bridge distinct knowledge communities in the prior collaboration network, then measures its association with disruptive innovation (a separate citation-based outcome) using a quasi-natural experiment based on the 2012 NSF policy change. This identification strategy relies on an exogenous policy shift rather than fitting SD parameters directly to the innovation outcome or renaming fitted quantities as predictions. No self-definitional loops, fitted-input-as-prediction steps, or load-bearing self-citations appear in the derivation chain. The central claim remains independently testable against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 2 invented entities

The central claim rests on two untested elements: (1) the validity of the 2012 NSF policy change as an exogenous shock that affects structural diversity without directly affecting innovation through other channels, and (2) the robustness of the community-detection procedure used to partition the collaboration network into 'distinct knowledge communities.' No free parameters are explicitly named, but any clustering threshold or resolution parameter would function as one.

axioms (2)

domain assumption The 2012 NSF policy change serves as a valid quasi-natural experiment that shifts structural diversity independently of other innovation drivers.
Invoked to support causal claims; no further justification supplied in the abstract.
domain assumption Knowledge communities can be reliably recovered from the historical collaboration network via standard community-detection methods.
Required for the definition of Structural Diversity; sensitivity to algorithm or resolution parameter is unaddressed.

invented entities (2)

Structural Diversity (SD) no independent evidence
purpose: Quantify the extent to which a team bridges multiple distinct knowledge communities in its prior collaboration network.
Newly defined construct that is the primary independent variable.
Disciplinary Integration (DI) no independent evidence
purpose: Mechanism explaining how high-SD teams combine heterogeneous knowledge into novel configurations.
Postulated mediator between SD and disruptive innovation.

pith-pipeline@v0.9.0 · 5555 in / 1744 out tokens · 62834 ms · 2026-05-14T22:14:42.500442+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We define SD as the ratio of connected components (CCs) to the total number of team members based on the team’s historical co-authorship network prior to a focal publication
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

SD is a powerful and robust predictor of disruptive innovation, outperforming traditional team novelty indicators such as team freshness and edge density

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

30 extracted references · 30 canonical work pages

[1]

Are ideas getting harder to find?American Economic Review, 110(4):1104–1144, 2020

Nicholas Bloom, Charles I Jones, John Van Reenen, and Michael Webb. Are ideas getting harder to find?American Economic Review, 110(4):1104–1144, 2020

work page 2020
[2]

Team assembly mechanisms determine collaboration network structure and team performance.Science, 308(5722):697–702, 2005

Roger Guimera, Brian Uzzi, Jarrett Spiro, and Luis A Nunes Amaral. Team assembly mechanisms determine collaboration network structure and team performance.Science, 308(5722):697–702, 2005

work page 2005
[3]

The increasing dominance of teams in production of knowledge.Science, 316(5827):1036–1039, 2007

Stefan Wuchty, Benjamin F Jones, and Brian Uzzi. The increasing dominance of teams in production of knowledge.Science, 316(5827):1036–1039, 2007

work page 2007
[4]

Large teams develop and small teams disrupt science and technology.Nature, 566(7744):378–382, 2019

Lingfei Wu, Dashun Wang, and James A Evans. Large teams develop and small teams disrupt science and technology.Nature, 566(7744):378–382, 2019

work page 2019
[5]

death of the renaissance man

Benjamin F Jones. The burden of knowledge and the “death of the renaissance man”: Is innovation getting harder?The Review of Economic Studies, 76(1):283–317, 2009

work page 2009
[6]

Atypical combinations and scientific impact.Science, 342(6157):468–472, 2013

Brian Uzzi, Satyam Mukherjee, Michael Stringer, and Ben Jones. Atypical combinations and scientific impact.Science, 342(6157):468–472, 2013

work page 2013
[7]

The possible role of resource requirements and academic career-choice risk on gender differences in publication rate and impact.PloS one, 7(12):e51332, 2012

Jordi Duch, Xiao Han T Zeng, Marta Sales-Pardo, Filippo Radicchi, Shayna Otis, Teresa K Woodruff, and Lu´ ıs A Nunes Amaral. The possible role of resource requirements and academic career-choice risk on gender differences in publication rate and impact.PloS one, 7(12):e51332, 2012

work page 2012
[8]

The preeminence of ethnic diversity in scientific collaboration.Nature communications, 9(1):5163, 2018

Bedoor K AlShebli, Talal Rahwan, and Wei Lee Woon. The preeminence of ethnic diversity in scientific collaboration.Nature communications, 9(1):5163, 2018. 17

work page 2018
[9]

Division of labor in collaborative knowledge production: The role of team size and interdisciplinarity.Research Policy, 49(6):103987, 2020

Carolin Haeussler and Henry Sauermann. Division of labor in collaborative knowledge production: The role of team size and interdisciplinarity.Research Policy, 49(6):103987, 2020

work page 2020
[10]

Multi-university research teams: Shift- ing impact, geography, and stratification in science.science, 322(5905):1259–1262, 2008

Benjamin F Jones, Stefan Wuchty, and Brian Uzzi. Multi-university research teams: Shift- ing impact, geography, and stratification in science.science, 322(5905):1259–1262, 2008

work page 2008
[11]

Prominent but less productive: The impact of interdisciplinarity on scientists’ research.Administrative Science Quarterly, 62(1):105–139, 2017

Erin Leahey, Christine M Beckman, and Taryn L Stanko. Prominent but less productive: The impact of interdisciplinarity on scientists’ research.Administrative Science Quarterly, 62(1):105–139, 2017

work page 2017
[12]

Fresh teams are asso- ciated with original and multidisciplinary research.Nature human behaviour, 5(10):1314– 1322, 2021

An Zeng, Ying Fan, Zengru Di, Yougui Wang, and Shlomo Havlin. Fresh teams are asso- ciated with original and multidisciplinary research.Nature human behaviour, 5(10):1314– 1322, 2021

work page 2021
[13]

Structural diversity in social contagion.Proceedings of the national academy of sciences, 109(16):5962–5966, 2012

Johan Ugander, Lars Backstrom, Cameron Marlow, and Jon Kleinberg. Structural diversity in social contagion.Proceedings of the national academy of sciences, 109(16):5962–5966, 2012

work page 2012
[14]

Structural holes and good ideas.American Journal of Sociology, 110(2):349– 399, 2004

Ronald S Burt. Structural holes and good ideas.American Journal of Sociology, 110(2):349– 399, 2004

work page 2004
[15]

Collaboration networks, structural holes, and innovation: A longitudinal study.Administrative Science Quarterly, 45(3):425–455, 2000

Gautam Ahuja. Collaboration networks, structural holes, and innovation: A longitudinal study.Administrative Science Quarterly, 45(3):425–455, 2000

work page 2000
[16]

Shaurya Rohatgi

Jason Priem, Heather Piwowar, and Richard Orr. Openalex: A fully-open index of scholarly works, authors, venues, institutions, and concepts.arXiv preprint arXiv:2205.01833, 2022

work page arXiv 2022
[17]

The central role of the propensity score in observational studies for causal effects.Biometrika, 70(1):41–55, 1983

Paul R Rosenbaum and Donald B Rubin. The central role of the propensity score in observational studies for causal effects.Biometrika, 70(1):41–55, 1983

work page 1983
[18]

The gender gap in scholarly self-promotion on social media.Nature Communications, 16(1):5552, 2025

Hao Peng, Misha Teplitskiy, Daniel M Romero, and Em˝ oke-´Agnes Horv´ at. The gender gap in scholarly self-promotion on social media.Nature Communications, 16(1):5552, 2025

work page 2025
[19]

Haochuan Cui, Yiling Lin, Lingfei Wu, and James A. Evans. Aging and the narrowing of scientific innovation, 2025

work page 2025
[20]

A dynamic network measure of technological change.Management science, 63(3):791–817, 2017

Russell J Funk and Jason Owen-Smith. A dynamic network measure of technological change.Management science, 63(3):791–817, 2017

work page 2017
[21]

The hierarchy of the sciences?American Journal of Sociology, 89(1):111– 139, 1983

Stephen Cole. The hierarchy of the sciences?American Journal of Sociology, 89(1):111– 139, 1983

work page 1983
[22]

Neural embed- dings of scholarly periodicals reveal complex disciplinary organizations.Science Advances, 7(17):eabb9004, 2021

Hao Peng, Qing Ke, Ceren Budak, Daniel M Romero, and Yong-Yeol Ahn. Neural embed- dings of scholarly periodicals reveal complex disciplinary organizations.Science Advances, 7(17):eabb9004, 2021

work page 2021
[23]

Arnetminer: extrac- tion and mining of academic social networks

Jie Tang, Jing Zhang, Limin Yao, Juanzi Li, Li Zhang, and Zhong Su. Arnetminer: extrac- tion and mining of academic social networks. InProceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 990–998, 2008

work page 2008
[24]

Randomization analysis of experimental data: The fisher randomization test comment.Journal of the American statistical association, 75(371):591–593, 1980

Donald B Rubin. Randomization analysis of experimental data: The fisher randomization test comment.Journal of the American statistical association, 75(371):591–593, 1980

work page 1980
[25]

Is science becoming more interdisciplinary? measuring and mapping six research fields over time.scientometrics, 81(3):719–745, 2009

Alan Porter and Ismael Rafols. Is science becoming more interdisciplinary? measuring and mapping six research fields over time.scientometrics, 81(3):719–745, 2009. 18

work page 2009
[26]

OECD Frascati Manual. Frascati manual 2015: Guidelines for collecting and reporting data on research and experimental development, the measurement of scientific, technological and innovation activities, 2015

work page 2015
[27]

The chaperone effect in scientific publishing.Proceedings of the National Academy of Sciences, 115(50):12603–12607, 2018

Vedran Sekara, Pierre Deville, Sebastian E Ahnert, Albert-L´ aszl´ o Barab´ asi, Roberta Sina- tra, and Sune Lehmann. The chaperone effect in scientific publishing.Proceedings of the National Academy of Sciences, 115(50):12603–12607, 2018

work page 2018
[28]

Acceptance in top journals shows large disparities across name-inferred ethnicities.Preprint at SocArXiv, 2021

Hao Peng, Karim Lakhani, and Misha Teplitskiy. Acceptance in top journals shows large disparities across name-inferred ethnicities.Preprint at SocArXiv, 2021

work page 2021
[29]

A new readability yardstick.Journal of Applied Psychology, 32(3):221–233, 1948

Rudolf Flesch. A new readability yardstick.Journal of Applied Psychology, 32(3):221–233, 1948

work page 1948
[30]

Promotional language and the adoption of innovative ideas in science.Proceedings of the National Academy of Sciences, 121(25):e2320066121, 2024

Hao Peng, Huilian Sophie Qiu, Henrik Barslund Fosse, and Brian Uzzi. Promotional language and the adoption of innovative ideas in science.Proceedings of the National Academy of Sciences, 121(25):e2320066121, 2024. 19

work page 2024