From 71a4c258f4506ce60c254f6d08bca9540ca60d27 Mon Sep 17 00:00:00 2001 From: leej3 Date: Mon, 3 Aug 2026 12:02:25 -0400 Subject: [PATCH 01/17] Add USRSE STAMPED poster abstract Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-luna --- 2026-usrse/stamped-abcd-poster-draft.md | 52 +++++++++++++++++++++++++ 1 file changed, 52 insertions(+) create mode 100644 2026-usrse/stamped-abcd-poster-draft.md diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md new file mode 100644 index 0000000..93a615d --- /dev/null +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -0,0 +1,52 @@ +# STAMPED in Practice: Reproducing a Scientific Result + +## Authors + +- John A. Lee , Dartmouth College, ORCID 0000-0001-5884-4247 +- Austin Macdonald , Dartmouth College, ORCID 0000-0002-8124-807X +- Yaroslav O. Halchenko , Dartmouth College, ORCID 0000-0003-3456-2493 + +## Keywords + +STAMPED, reproducibility, research objects, neuroimaging, provenance, AI-assisted science + +## Abstract + +The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and repeat or extend the analysis. Code alone does not provide this: data, parameters, software environments, execution context, and the provenance connecting inputs to outputs must also be identifiable and recoverable. STAMPED defines seven properties for organizing these materials as a durable research object. + +We applied STAMPED to reproduce an analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data. This case study shows how the principles translate into concrete research-design choices and how those choices make a computational result easier to inspect, execute, verify, and reuse. + +- **Self-contained:** Everything needed to understand and run the study was made identifiable from one research-object boundary, reducing reliance on undocumented local context (supported by DataLad and git-annex). +- **Tracked:** Results were linked to their inputs, transformations, execution context, and scientific meaning so their provenance could be inspected (recorded with Git, DataLad run records, `con-duct`, and NIDM/PROV). +- **Actionable:** Procedures and expected outputs were encoded as executable, testable operations instead of instructions requiring interpretation (implemented through BIDS, tested BIDS Apps, Pixi tasks, manifests, and validation commands). +- **Modular:** Data, environments, operations, and results retained distinct identities so components could be reviewed, updated, and reused independently (organized as versioned DataLad components). +- **Portable:** Software dependencies and interfaces were separated from site-specific configuration so execution could move across systems (supported by locked Pixi environments, exact Apptainer images, and standard interfaces). +- **Ephemeral:** Fresh, disposable execution tested whether results followed from the declared research object rather than accumulated machine state (tested with BABS, Slurm, and clean-installation replay). +- **Distributable:** Exact permitted research states remained persistently retrievable under explicit licensing and access rules, supporting reuse while protecting controlled data (implemented through Git/annex siblings and separate access boundaries). + +Although this project demonstrates STAMPED through a scientific reproduction, its broader value lies in applying the principles from the start. Designing the research object alongside the analysis makes later reproduction more thorough and achievable while strengthening review, collaboration, extension, and reuse throughout the study. + +AI-assisted tools can reduce the software burden of computational research by helping translate questions into workflows and automating routine implementation, testing, and documentation. They can also amplify hidden assumptions, undocumented changes, fragile dependencies, and outputs that are difficult to inspect. STAMPED mitigates these risks by bounding context, preserving provenance, defining executable tasks and component interfaces, testing independence from local state, and retaining exact handoff states. This makes AI-assisted work more auditable and reusable, allowing researchers to spend more time applying domain expertise to scientific questions, methods, and interpretation. + +## Acknowledgments + +We thank the authors of the STAMPED principles and the developers and communities behind DataLad, BIDS, BABS, ReproNim, NIDM, Pixi, Apptainer, and `con-duct`. + +AI-assisted content disclosure: This submission was prepared with assistance from OpenAI Codex (GPT-5, accessed August 2026). The system helped synthesize repository evidence and draft the Abstract and Connection to Mission sections. The authors reviewed the source evidence, edited the text, and remain responsible for all claims. + +## References + +1. Macdonald A, Baker CC, To I, Halchenko YO. *STAMPED principles for reproducible research objects*. May 2026. +2. Lee JA. *STAMPED-dl_morphometrics_biases: an ideal-oriented reconstruction of a neuroimaging analysis*. 2026. +3. Nielson DM, Lee JA, Earl E, Moraczewski D, Pereira F. *Age dependent volume estimation biases in recon-all clinical and recon-any*. OHBM 2025 poster. +4. Halchenko YO, et al. DataLad: distributed system for joint management of code, data, and their relationship. *Journal of Open Source Software*. 2021;6(63):3262. +5. Zhao C, et al. A reproducible and generalizable software workflow for analysis of large-scale neuroimaging data collections using BIDS Apps. *Imaging Neuroscience*. 2024;2:imag-2-00074. +6. Center for Open Neuroscience. *con-duct: a lightweight wrapper for monitoring command execution*. RRID:SCR_025436. + +## Connection to Mission, Goals, & Interests of US-RSE Community + +Research software engineers help determine whether a computational result is merely produced once or becomes a durable scientific contribution. STAMPED gives RSEs and researchers a shared vocabulary for the properties that make this difference: coherent research-object boundaries, exact identities, executable procedures, independent components, explicit environments, clean execution, and persistent distribution. This case study translates those properties into a concrete implementation and shows how they can be designed into an analysis from the beginning rather than added after publication. + +The contribution is a reusable decision framework for translating STAMPED principles into choices about research-object boundaries, provenance, execution, validation, and distribution. The worked reproduction exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. Although we demonstrate the approach through a scientific reproduction, such a reproduction is more thorough and achievable when the necessary context, procedures, and evidence are incorporated into the original analysis. + +The project also addresses the conference theme, "Advancing Science in the Age of AI." AI-assisted tools can translate research questions into computational procedures, automate routine implementation, and generate tests and documentation, reducing the software burden on researchers. They can also amplify hidden assumptions, opaque transformations, and environment-sensitive outputs. STAMPED gives RSEs a practical framework for constraining that automation through explicit context, provenance, interfaces, execution tests, and durable handoffs. Researchers can therefore benefit from automation while retaining scientific oversight and focusing their effort on domain questions, methods, and interpretation. From e8c88c3df4a7fcf64c005838b693ab9a3986198a Mon Sep 17 00:00:00 2001 From: leej3 Date: Wed, 5 Aug 2026 11:15:32 -0400 Subject: [PATCH 02/17] Address USRSE abstract review feedback Co-Authored-By: Austin Macdonald <1028657+asmacdo@users.noreply.github.com> Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-sol --- 2026-usrse/stamped-abcd-poster-draft.md | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index 93a615d..9d1a073 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -12,21 +12,21 @@ STAMPED, reproducibility, research objects, neuroimaging, provenance, AI-assiste ## Abstract -The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and repeat or extend the analysis. Code alone does not provide this: data, parameters, software environments, execution context, and the provenance connecting inputs to outputs must also be identifiable and recoverable. STAMPED defines seven properties for organizing these materials as a durable research object. +The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and re-execute or extend the analysis. Code alone is insufficient: data, parameters, software environments, execution context, and provenance must also be identifiable and recoverable. AI-assisted tools can reduce the software development burden, but their speed increases the importance of explicit context, provenance, and reviewable intermediate states. STAMPED defines seven properties for organizing these materials as a durable research object. -We applied STAMPED to reproduce an analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data. This case study shows how the principles translate into concrete research-design choices and how those choices make a computational result easier to inspect, execute, verify, and reuse. +We applied STAMPED while reproducing an analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data. The reproduction used AI-assisted development, allowing us to examine how STAMPED structures the context, evidence, and handoffs surrounding automated work. The case study shows how the principles guide concrete decisions about the organization, execution, documentation, and distribution of a computational analysis. - **Self-contained:** Everything needed to understand and run the study was made identifiable from one research-object boundary, reducing reliance on undocumented local context (supported by DataLad and git-annex). - **Tracked:** Results were linked to their inputs, transformations, execution context, and scientific meaning so their provenance could be inspected (recorded with Git, DataLad run records, `con-duct`, and NIDM/PROV). -- **Actionable:** Procedures and expected outputs were encoded as executable, testable operations instead of instructions requiring interpretation (implemented through BIDS, tested BIDS Apps, Pixi tasks, manifests, and validation commands). -- **Modular:** Data, environments, operations, and results retained distinct identities so components could be reviewed, updated, and reused independently (organized as versioned DataLad components). +- **Actionable:** Procedures and expected outputs were encoded as executable, testable operations instead of instructions requiring interpretation (implemented through BIDS, tested BIDS Apps, Pixi tasks, manifests, validation commands, and `datalad rerun`). +- **Modular:** Data, environments, operations, and results retained distinct identities so components could be reviewed, updated, and reused independently (organized as versioned DataLad subdatasets). - **Portable:** Software dependencies and interfaces were separated from site-specific configuration so execution could move across systems (supported by locked Pixi environments, exact Apptainer images, and standard interfaces). - **Ephemeral:** Fresh, disposable execution tested whether results followed from the declared research object rather than accumulated machine state (tested with BABS, Slurm, and clean-installation replay). - **Distributable:** Exact permitted research states remained persistently retrievable under explicit licensing and access rules, supporting reuse while protecting controlled data (implemented through Git/annex siblings and separate access boundaries). Although this project demonstrates STAMPED through a scientific reproduction, its broader value lies in applying the principles from the start. Designing the research object alongside the analysis makes later reproduction more thorough and achievable while strengthening review, collaboration, extension, and reuse throughout the study. -AI-assisted tools can reduce the software burden of computational research by helping translate questions into workflows and automating routine implementation, testing, and documentation. They can also amplify hidden assumptions, undocumented changes, fragile dependencies, and outputs that are difficult to inspect. STAMPED mitigates these risks by bounding context, preserving provenance, defining executable tasks and component interfaces, testing independence from local state, and retaining exact handoff states. This makes AI-assisted work more auditable and reusable, allowing researchers to spend more time applying domain expertise to scientific questions, methods, and interpretation. +Within this structure, AI-assisted tools can help translate scientific intent into workflows and automate routine implementation, testing, and documentation. Without explicit controls, those efficiencies can amplify hidden assumptions, undocumented changes, fragile dependencies, and difficult-to-inspect outputs. STAMPED mitigates these risks by bounding context, preserving provenance, defining executable tasks and component interfaces, testing independence from local state, and retaining exact handoff states. This allows researchers to benefit from automation while preserving oversight and devoting more attention to scientific questions, methods, and interpretation. ## Acknowledgments @@ -49,4 +49,4 @@ Research software engineers help determine whether a computational result is mer The contribution is a reusable decision framework for translating STAMPED principles into choices about research-object boundaries, provenance, execution, validation, and distribution. The worked reproduction exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. Although we demonstrate the approach through a scientific reproduction, such a reproduction is more thorough and achievable when the necessary context, procedures, and evidence are incorporated into the original analysis. -The project also addresses the conference theme, "Advancing Science in the Age of AI." AI-assisted tools can translate research questions into computational procedures, automate routine implementation, and generate tests and documentation, reducing the software burden on researchers. They can also amplify hidden assumptions, opaque transformations, and environment-sensitive outputs. STAMPED gives RSEs a practical framework for constraining that automation through explicit context, provenance, interfaces, execution tests, and durable handoffs. Researchers can therefore benefit from automation while retaining scientific oversight and focusing their effort on domain questions, methods, and interpretation. +The project also addresses the conference theme, "Advancing Science in the Age of AI," by shifting attention from what AI can do to what evidence AI-assisted work leaves behind. Using STAMPED as a shared point of reference, the poster invites RSEs to compare how their teams record decisions, divide responsibility, and evaluate handoffs, and to consider how those expectations could become community review criteria and institutional practice across research domains. From 125b322da4a6e5c439ab8ca803e90e94253b9070 Mon Sep 17 00:00:00 2001 From: John lee Date: Thu, 6 Aug 2026 10:12:41 -0400 Subject: [PATCH 03/17] Apply suggestions from code review Co-authored-by: Yaroslav Halchenko --- 2026-usrse/stamped-abcd-poster-draft.md | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index 9d1a073..f7ab7ff 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -1,4 +1,4 @@ -# STAMPED in Practice: Reproducing a Scientific Result +# STAMPED in Practice: Reproducing a Scientific Result on ABCD Neuroimaging Study ## Authors @@ -12,19 +12,21 @@ STAMPED, reproducibility, research objects, neuroimaging, provenance, AI-assiste ## Abstract -The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and re-execute or extend the analysis. Code alone is insufficient: data, parameters, software environments, execution context, and provenance must also be identifiable and recoverable. AI-assisted tools can reduce the software development burden, but their speed increases the importance of explicit context, provenance, and reviewable intermediate states. STAMPED defines seven properties for organizing these materials as a durable research object. +The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and re-execute or extend the analysis. Code alone is insufficient: data, parameters, software environments, execution context, and provenance must also be identifiable and recoverable. AI-assisted tools have potential to reduce the software development burden, but their sheer output volume and non-determinstic operation increase the importance of explicit capture of context, provenance, and reviewable intermediate states. STAMPED defines seven properties for organizing these materials as a durable research object. -We applied STAMPED while reproducing an analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data. The reproduction used AI-assisted development, allowing us to examine how STAMPED structures the context, evidence, and handoffs surrounding automated work. The case study shows how the principles guide concrete decisions about the organization, execution, documentation, and distribution of a computational analysis. +We applied STAMPED while reproducing an analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). The reproduction used AI-assisted development, allowing us to examine how STAMPED structures the context, evidence, and handoffs surrounding automated work. The case study shows how the principles guide concrete decisions about the organization, execution, documentation, and distribution of a computational analysis. + +Overall STAMPED define following properties to characterize composed research objects: - **Self-contained:** Everything needed to understand and run the study was made identifiable from one research-object boundary, reducing reliance on undocumented local context (supported by DataLad and git-annex). - **Tracked:** Results were linked to their inputs, transformations, execution context, and scientific meaning so their provenance could be inspected (recorded with Git, DataLad run records, `con-duct`, and NIDM/PROV). -- **Actionable:** Procedures and expected outputs were encoded as executable, testable operations instead of instructions requiring interpretation (implemented through BIDS, tested BIDS Apps, Pixi tasks, manifests, validation commands, and `datalad rerun`). +- **Actionable:** Procedures and expected outputs were encoded as executable, testable operations instead of instructions requiring interpretation (implemented through BIDS, tested BIDS Apps, Pixi tasks, manifests, validation commands, and Datalad run records). - **Modular:** Data, environments, operations, and results retained distinct identities so components could be reviewed, updated, and reused independently (organized as versioned DataLad subdatasets). - **Portable:** Software dependencies and interfaces were separated from site-specific configuration so execution could move across systems (supported by locked Pixi environments, exact Apptainer images, and standard interfaces). -- **Ephemeral:** Fresh, disposable execution tested whether results followed from the declared research object rather than accumulated machine state (tested with BABS, Slurm, and clean-installation replay). +- **Ephemeral:** Fresh, disposable execution tested whether results followed from the declared research object rather than accumulated machine state (tested with BABS, Slurm, and clean-installation replay with `datalad rerun` and alike). - **Distributable:** Exact permitted research states remained persistently retrievable under explicit licensing and access rules, supporting reuse while protecting controlled data (implemented through Git/annex siblings and separate access boundaries). -Although this project demonstrates STAMPED through a scientific reproduction, its broader value lies in applying the principles from the start. Designing the research object alongside the analysis makes later reproduction more thorough and achievable while strengthening review, collaboration, extension, and reuse throughout the study. +Although this project demonstrates STAMPED through a scientific reproduction, its broader value lies in applying the principles from the start along with the FAIR principles mandating use of agreed upon standards. Designing the research object alongside the analysis makes later reproduction more thorough and achievable while strengthening review, collaboration, extension, and reuse throughout the study. Within this structure, AI-assisted tools can help translate scientific intent into workflows and automate routine implementation, testing, and documentation. Without explicit controls, those efficiencies can amplify hidden assumptions, undocumented changes, fragile dependencies, and difficult-to-inspect outputs. STAMPED mitigates these risks by bounding context, preserving provenance, defining executable tasks and component interfaces, testing independence from local state, and retaining exact handoff states. This allows researchers to benefit from automation while preserving oversight and devoting more attention to scientific questions, methods, and interpretation. From 6b36e139403bc69ef83ca01af1f3bed1858c1c7e Mon Sep 17 00:00:00 2001 From: Cody Baker <51133164+CodyCBakerPhD@users.noreply.github.com> Date: Thu, 6 Aug 2026 11:42:22 -0400 Subject: [PATCH 04/17] Create stamped-princples.md --- 2026-usrse/stamped-princples.md | 37 +++++++++++++++++++++++++++++++++ 1 file changed, 37 insertions(+) create mode 100644 2026-usrse/stamped-princples.md diff --git a/2026-usrse/stamped-princples.md b/2026-usrse/stamped-princples.md new file mode 100644 index 0000000..4e00c63 --- /dev/null +++ b/2026-usrse/stamped-princples.md @@ -0,0 +1,37 @@ +# Pragmatic practices for reproducible and open science through case stories and principles + + +## Presenters + +Cody C.Baker , Center for Open Neuroscience, Dartmouth College, 0000-0002-0829-4790 + + +## Keywords + +Reproducible research - Open science - Provenance + +## Abstract + +Neuroscience increasingly depends on the interplay of code, data, and computational environments, yet the record of how they were used together is often incomplete, scattered across repositories, wikis, and notebooks, or lost entirely. +This fragmentation undermines rigor, reproducibility, reusability, and efficiency in BRAIN Initiative pipelines that routinely span multiple institutions, archives, and compute platforms. +Existing frameworks such as FAIR and FAIR4RS govern discovery and interoperability of digital objects, but do not specify how research objects should be structured and managed so they can be re-executed, extended, and audited. +The community lacks a shared vocabulary for this operational layer. + +Building on the YODA and VAMP traditions from neuroimaging, and on patterns that have independently converged across geophysics, genomics, statistics, and neuroimaging over three decades, we formalize seven principles a research object should satisfy: Self-containment, Tracking, Actionability, Modularity, Portability, Ephemerality, and Distributability, collectively STAMPED. +Each spans a spectrum from practical minimum to aspirational ideal, so adoption is non-prescriptive and incremental. +Formal LinkML schemas, an interactive compliance checklist, and the curated collection of examples are provided as enabling tools to this end. + +We demonstrate STAMPED through two major neuroscience pipelines. +OpenNeuroDerivatives reorganized derivative neuroimaging datasets so they exist as independent Ephemeral units that reference raw inputs as subdatasets rather than nesting under them, removing an upward dependency that previously violated Self-containment, Modularity, and Portability. +DANDI Compute, utilizing the Allen Institute for Neural Dynamics electrophysiology pipeline, packages spike-sorting outputs into nested BIDS-derivative units in which each leaf contains the exact code, runtime logs, outputs, and provenance metadata needed to re-execute the analysis, satisfying STAMPED end-to-end. + +These adoptions show that STAMPED provides a tool-agnostic, incrementally adoptable vocabulary that lets researchers, reviewers, collaborators, and emerging AI agents evaluate and improve the operational maturity of computational neuroscience. +By making research objects re-executable and inspectable by construction, STAMPED converts reproducibility from an aspiration into a measurable property of everyday neuroscience practice. + +## References + +Austin Macdonald, Cody Baker, Isaac To, Yaroslav O. Halchenko, “STAMPED principles for reproducible research objects”, 26-May-2026. [Online]. Available: osf.io/preprints/metaarxiv/f3h82_v1. +Michelle Barker, et al. Introducing the FAIR Principles for research software. Scientific Data, 9(1):622, October 2022. ISSN 2052-4463. +Michael Hanke, et al. YODA: YODA’s organigram on data analysis, 2018. Slides. +Alessio P Buccino, Arjun Sridhar, David Feng, Karel Svoboda, Joshua H Siegle (2026). Efficient and reproducible pipelines for spike sorting large-scale electrophysiology data. eLife, 15:RP110170. +Michael Hanke. What is DataLad and what can it do for you? 2023. https://files.inm7.de/mih/pres/talks/whatisdatalad_2023.html#/ From ba333f94c05cae46442ae3096d3e85dc0ff52247 Mon Sep 17 00:00:00 2001 From: leej3 Date: Thu, 6 Aug 2026 10:59:25 -0400 Subject: [PATCH 05/17] Add Snapper pre-commit hook for USRSE abstract Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-luna --- .pre-commit-config.yaml | 6 ++++++ 1 file changed, 6 insertions(+) create mode 100644 .pre-commit-config.yaml diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml new file mode 100644 index 0000000..f394726 --- /dev/null +++ b/.pre-commit-config.yaml @@ -0,0 +1,6 @@ +repos: +- repo: https://github.com/TurtleTech-ehf/snapper + rev: v0.9.1 + hooks: + - id: snapper + files: ^2026-usrse/stamped-abcd-poster-draft\.md$ From ff1a468c21797385efb0c6de82dbe11bc304ea75 Mon Sep 17 00:00:00 2001 From: leej3 Date: Thu, 6 Aug 2026 14:08:06 -0400 Subject: [PATCH 06/17] Normalize USRSE abstract with Snapper Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-luna --- 2026-usrse/stamped-abcd-poster-draft.md | 140 +++++++++++++++--------- 1 file changed, 86 insertions(+), 54 deletions(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index f7ab7ff..0ff787e 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -1,54 +1,86 @@ -# STAMPED in Practice: Reproducing a Scientific Result on ABCD Neuroimaging Study - -## Authors - -- John A. Lee , Dartmouth College, ORCID 0000-0001-5884-4247 -- Austin Macdonald , Dartmouth College, ORCID 0000-0002-8124-807X -- Yaroslav O. Halchenko , Dartmouth College, ORCID 0000-0003-3456-2493 - -## Keywords - -STAMPED, reproducibility, research objects, neuroimaging, provenance, AI-assisted science - -## Abstract - -The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and re-execute or extend the analysis. Code alone is insufficient: data, parameters, software environments, execution context, and provenance must also be identifiable and recoverable. AI-assisted tools have potential to reduce the software development burden, but their sheer output volume and non-determinstic operation increase the importance of explicit capture of context, provenance, and reviewable intermediate states. STAMPED defines seven properties for organizing these materials as a durable research object. - -We applied STAMPED while reproducing an analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). The reproduction used AI-assisted development, allowing us to examine how STAMPED structures the context, evidence, and handoffs surrounding automated work. The case study shows how the principles guide concrete decisions about the organization, execution, documentation, and distribution of a computational analysis. - -Overall STAMPED define following properties to characterize composed research objects: - -- **Self-contained:** Everything needed to understand and run the study was made identifiable from one research-object boundary, reducing reliance on undocumented local context (supported by DataLad and git-annex). -- **Tracked:** Results were linked to their inputs, transformations, execution context, and scientific meaning so their provenance could be inspected (recorded with Git, DataLad run records, `con-duct`, and NIDM/PROV). -- **Actionable:** Procedures and expected outputs were encoded as executable, testable operations instead of instructions requiring interpretation (implemented through BIDS, tested BIDS Apps, Pixi tasks, manifests, validation commands, and Datalad run records). -- **Modular:** Data, environments, operations, and results retained distinct identities so components could be reviewed, updated, and reused independently (organized as versioned DataLad subdatasets). -- **Portable:** Software dependencies and interfaces were separated from site-specific configuration so execution could move across systems (supported by locked Pixi environments, exact Apptainer images, and standard interfaces). -- **Ephemeral:** Fresh, disposable execution tested whether results followed from the declared research object rather than accumulated machine state (tested with BABS, Slurm, and clean-installation replay with `datalad rerun` and alike). -- **Distributable:** Exact permitted research states remained persistently retrievable under explicit licensing and access rules, supporting reuse while protecting controlled data (implemented through Git/annex siblings and separate access boundaries). - -Although this project demonstrates STAMPED through a scientific reproduction, its broader value lies in applying the principles from the start along with the FAIR principles mandating use of agreed upon standards. Designing the research object alongside the analysis makes later reproduction more thorough and achievable while strengthening review, collaboration, extension, and reuse throughout the study. - -Within this structure, AI-assisted tools can help translate scientific intent into workflows and automate routine implementation, testing, and documentation. Without explicit controls, those efficiencies can amplify hidden assumptions, undocumented changes, fragile dependencies, and difficult-to-inspect outputs. STAMPED mitigates these risks by bounding context, preserving provenance, defining executable tasks and component interfaces, testing independence from local state, and retaining exact handoff states. This allows researchers to benefit from automation while preserving oversight and devoting more attention to scientific questions, methods, and interpretation. - -## Acknowledgments - -We thank the authors of the STAMPED principles and the developers and communities behind DataLad, BIDS, BABS, ReproNim, NIDM, Pixi, Apptainer, and `con-duct`. - -AI-assisted content disclosure: This submission was prepared with assistance from OpenAI Codex (GPT-5, accessed August 2026). The system helped synthesize repository evidence and draft the Abstract and Connection to Mission sections. The authors reviewed the source evidence, edited the text, and remain responsible for all claims. - -## References - -1. Macdonald A, Baker CC, To I, Halchenko YO. *STAMPED principles for reproducible research objects*. May 2026. -2. Lee JA. *STAMPED-dl_morphometrics_biases: an ideal-oriented reconstruction of a neuroimaging analysis*. 2026. -3. Nielson DM, Lee JA, Earl E, Moraczewski D, Pereira F. *Age dependent volume estimation biases in recon-all clinical and recon-any*. OHBM 2025 poster. -4. Halchenko YO, et al. DataLad: distributed system for joint management of code, data, and their relationship. *Journal of Open Source Software*. 2021;6(63):3262. -5. Zhao C, et al. A reproducible and generalizable software workflow for analysis of large-scale neuroimaging data collections using BIDS Apps. *Imaging Neuroscience*. 2024;2:imag-2-00074. -6. Center for Open Neuroscience. *con-duct: a lightweight wrapper for monitoring command execution*. RRID:SCR_025436. - -## Connection to Mission, Goals, & Interests of US-RSE Community - -Research software engineers help determine whether a computational result is merely produced once or becomes a durable scientific contribution. STAMPED gives RSEs and researchers a shared vocabulary for the properties that make this difference: coherent research-object boundaries, exact identities, executable procedures, independent components, explicit environments, clean execution, and persistent distribution. This case study translates those properties into a concrete implementation and shows how they can be designed into an analysis from the beginning rather than added after publication. - -The contribution is a reusable decision framework for translating STAMPED principles into choices about research-object boundaries, provenance, execution, validation, and distribution. The worked reproduction exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. Although we demonstrate the approach through a scientific reproduction, such a reproduction is more thorough and achievable when the necessary context, procedures, and evidence are incorporated into the original analysis. - -The project also addresses the conference theme, "Advancing Science in the Age of AI," by shifting attention from what AI can do to what evidence AI-assisted work leaves behind. Using STAMPED as a shared point of reference, the poster invites RSEs to compare how their teams record decisions, divide responsibility, and evaluate handoffs, and to consider how those expectations could become community review criteria and institutional practice across research domains. +# STAMPED in Practice: Reproducing a Scientific Result on ABCD Neuroimaging Study + +## Authors + +- John A. Lee , Dartmouth College, ORCID 0000-0001-5884-4247 +- Austin Macdonald , Dartmouth College, ORCID 0000-0002-8124-807X +- Yaroslav O. Halchenko , Dartmouth College, ORCID 0000-0003-3456-2493 + +## Keywords + +STAMPED, reproducibility, research objects, neuroimaging, provenance, AI-assisted science + +## Abstract + +The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and re-execute or extend the analysis. +Code alone is insufficient: data, parameters, software environments, execution context, and provenance must also be identifiable and recoverable. +AI-assisted tools have potential to reduce the software development burden, but their sheer output volume and non-determinstic operation increase the importance of explicit capture of context, provenance, and reviewable intermediate states. +STAMPED defines seven properties for organizing these materials as a durable research object. + +We applied STAMPED while reproducing an analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). +The reproduction used AI-assisted development, allowing us to examine how STAMPED structures the context, evidence, and handoffs surrounding automated work. +The case study shows how the principles guide concrete decisions about the organization, execution, documentation, and distribution of a computational analysis. + +Overall STAMPED define following properties to characterize composed research objects: + +- **Self-contained:** Everything needed to understand and run the study was made identifiable from one research-object boundary, reducing reliance on undocumented local context (supported by DataLad and git-annex). +- **Tracked:** Results were linked to their inputs, transformations, execution context, and scientific meaning so their provenance could be inspected (recorded with Git, DataLad run records, `con-duct`, and NIDM/PROV). +- **Actionable:** Procedures and expected outputs were encoded as executable, testable operations instead of instructions requiring interpretation (implemented through BIDS, tested BIDS Apps, Pixi tasks, manifests, validation commands, and Datalad run records). +- **Modular:** Data, environments, operations, and results retained distinct identities so components could be reviewed, updated, and reused independently (organized as versioned DataLad subdatasets). +- **Portable:** Software dependencies and interfaces were separated from site-specific configuration so execution could move across systems (supported by locked Pixi environments, exact Apptainer images, and standard interfaces). +- **Ephemeral:** Fresh, disposable execution tested whether results followed from the declared research object rather than accumulated machine state (tested with BABS, Slurm, and clean-installation replay with `datalad rerun` and alike). +- **Distributable:** Exact permitted research states remained persistently retrievable under explicit licensing and access rules, supporting reuse while protecting controlled data (implemented through Git/annex siblings and separate access boundaries). + +Although this project demonstrates STAMPED through a scientific reproduction, its broader value lies in applying the principles from the start along with the FAIR principles mandating use of agreed upon standards. +Designing the research object alongside the analysis makes later reproduction more thorough and achievable while strengthening review, collaboration, extension, and reuse throughout the study. + +Within this structure, AI-assisted tools can help translate scientific intent into workflows and automate routine implementation, testing, and documentation. +Without explicit controls, those efficiencies can amplify hidden assumptions, undocumented changes, fragile dependencies, and difficult-to-inspect outputs. +STAMPED mitigates these risks by bounding context, preserving provenance, defining executable tasks and component interfaces, testing independence from local state, and retaining exact handoff states. +This allows researchers to benefit from automation while preserving oversight and devoting more attention to scientific questions, methods, and interpretation. + +## Acknowledgments + +We thank the authors of the STAMPED principles and the developers and communities behind DataLad, BIDS, BABS, ReproNim, NIDM, Pixi, Apptainer, and `con-duct`. + +AI-assisted content disclosure: This submission was prepared with assistance from OpenAI Codex (GPT-5, accessed August 2026). +The system helped synthesize repository evidence and draft the Abstract and Connection to Mission sections. +The authors reviewed the source evidence, edited the text, and remain responsible for all claims. + +## References + +1. Macdonald A, Baker CC, To I, Halchenko YO. +*STAMPED principles for reproducible research objects*. +May 2026. + +2. Lee JA. +*STAMPED-dl_morphometrics_biases: an ideal-oriented reconstruction of a neuroimaging analysis*. +2026. + +3. Nielson DM, Lee JA, Earl E, Moraczewski D, Pereira F. *Age dependent volume estimation biases in recon-all clinical and recon-any*. +OHBM 2025 poster. + +4. Halchenko YO, et al. DataLad: distributed system for joint management of code, data, and their relationship. +*Journal of Open Source Software*. +2021;6(63):3262. + +5. Zhao C, et al. A reproducible and generalizable software workflow for analysis of large-scale neuroimaging data collections using BIDS Apps. +*Imaging Neuroscience*. 2024;2:imag-2-00074. + +6. Center for Open Neuroscience. +*con-duct: a lightweight wrapper for monitoring command execution*. +RRID:SCR_025436. + + +## Connection to Mission, Goals, & Interests of US-RSE Community + +Research software engineers help determine whether a computational result is merely produced once or becomes a durable scientific contribution. +STAMPED gives RSEs and researchers a shared vocabulary for the properties that make this difference: coherent research-object boundaries, exact identities, executable procedures, independent components, explicit environments, clean execution, and persistent distribution. +This case study translates those properties into a concrete implementation and shows how they can be designed into an analysis from the beginning rather than added after publication. + +The contribution is a reusable decision framework for translating STAMPED principles into choices about research-object boundaries, provenance, execution, validation, and distribution. +The worked reproduction exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. +Although we demonstrate the approach through a scientific reproduction, such a reproduction is more thorough and achievable when the necessary context, procedures, and evidence are incorporated into the original analysis. + +The project also addresses the conference theme, "Advancing Science in the Age of AI," by shifting attention from what AI can do to what evidence AI-assisted work leaves behind. +Using STAMPED as a shared point of reference, the poster invites RSEs to compare how their teams record decisions, divide responsibility, and evaluate handoffs, and to consider how those expectations could become community review criteria and institutional practice across research domains. From 58441f4c9b5be2d47adfe8d6726b3bb8ae6f5fe4 Mon Sep 17 00:00:00 2001 From: leej3 Date: Thu, 6 Aug 2026 14:47:43 -0400 Subject: [PATCH 07/17] fix smaller feedback items --- 2026-usrse/stamped-abcd-poster-draft.md | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index 0ff787e..932a2dd 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -14,16 +14,16 @@ STAMPED, reproducibility, research objects, neuroimaging, provenance, AI-assiste The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and re-execute or extend the analysis. Code alone is insufficient: data, parameters, software environments, execution context, and provenance must also be identifiable and recoverable. -AI-assisted tools have potential to reduce the software development burden, but their sheer output volume and non-determinstic operation increase the importance of explicit capture of context, provenance, and reviewable intermediate states. -STAMPED defines seven properties for organizing these materials as a durable research object. +AI-assisted tools can reduce software-development effort, but their high-volume, nondeterministic output makes explicit context, provenance, and reviewable intermediate states increasingly important. +STAMPED () defines seven properties for organizing these materials as a durable research object. We applied STAMPED while reproducing an analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). -The reproduction used AI-assisted development, allowing us to examine how STAMPED structures the context, evidence, and handoffs surrounding automated work. +The reproduction used AI-assisted development, allowing us to examine how STAMPED structures the context, evidence, and reviewable checkpoints surrounding automated work. The case study shows how the principles guide concrete decisions about the organization, execution, documentation, and distribution of a computational analysis. -Overall STAMPED define following properties to characterize composed research objects: +Together, the seven STAMPED properties characterize a composed research object: -- **Self-contained:** Everything needed to understand and run the study was made identifiable from one research-object boundary, reducing reliance on undocumented local context (supported by DataLad and git-annex). +- **Self-contained:** Everything needed to understand and run the study was available within one research-object boundary, eliminating the need to look elsewhere for context (supported by DataLad and git-annex). - **Tracked:** Results were linked to their inputs, transformations, execution context, and scientific meaning so their provenance could be inspected (recorded with Git, DataLad run records, `con-duct`, and NIDM/PROV). - **Actionable:** Procedures and expected outputs were encoded as executable, testable operations instead of instructions requiring interpretation (implemented through BIDS, tested BIDS Apps, Pixi tasks, manifests, validation commands, and Datalad run records). - **Modular:** Data, environments, operations, and results retained distinct identities so components could be reviewed, updated, and reused independently (organized as versioned DataLad subdatasets). @@ -36,7 +36,7 @@ Designing the research object alongside the analysis makes later reproduction mo Within this structure, AI-assisted tools can help translate scientific intent into workflows and automate routine implementation, testing, and documentation. Without explicit controls, those efficiencies can amplify hidden assumptions, undocumented changes, fragile dependencies, and difficult-to-inspect outputs. -STAMPED mitigates these risks by bounding context, preserving provenance, defining executable tasks and component interfaces, testing independence from local state, and retaining exact handoff states. +STAMPED mitigates these risks by bounding context, preserving provenance, defining executable tasks and component interfaces, testing independence from local state, and preserving explicit states for human review. This allows researchers to benefit from automation while preserving oversight and devoting more attention to scientific questions, methods, and interpretation. ## Acknowledgments @@ -78,9 +78,9 @@ Research software engineers help determine whether a computational result is mer STAMPED gives RSEs and researchers a shared vocabulary for the properties that make this difference: coherent research-object boundaries, exact identities, executable procedures, independent components, explicit environments, clean execution, and persistent distribution. This case study translates those properties into a concrete implementation and shows how they can be designed into an analysis from the beginning rather than added after publication. -The contribution is a reusable decision framework for translating STAMPED principles into choices about research-object boundaries, provenance, execution, validation, and distribution. -The worked reproduction exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. +The contribution is a worked example of how STAMPED principles guide choices about research-object boundaries, provenance, execution, validation, and distribution. +It exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. Although we demonstrate the approach through a scientific reproduction, such a reproduction is more thorough and achievable when the necessary context, procedures, and evidence are incorporated into the original analysis. The project also addresses the conference theme, "Advancing Science in the Age of AI," by shifting attention from what AI can do to what evidence AI-assisted work leaves behind. -Using STAMPED as a shared point of reference, the poster invites RSEs to compare how their teams record decisions, divide responsibility, and evaluate handoffs, and to consider how those expectations could become community review criteria and institutional practice across research domains. +Using STAMPED as a shared point of reference, the poster invites RSEs to compare how their teams record decisions, divide responsibility, and evaluate AI-assisted work, and to consider how those expectations could become community review criteria and institutional practice across research domains. From 373c43bb3a3c919366d2eef1ba159a045eb8612f Mon Sep 17 00:00:00 2001 From: leej3 Date: Thu, 6 Aug 2026 17:20:16 -0400 Subject: [PATCH 08/17] Refine STAMPED case-study framing Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-sol --- 2026-usrse/stamped-abcd-poster-draft.md | 21 +++++++-------------- 1 file changed, 7 insertions(+), 14 deletions(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index 932a2dd..deddd1b 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -17,11 +17,11 @@ Code alone is insufficient: data, parameters, software environments, execution c AI-assisted tools can reduce software-development effort, but their high-volume, nondeterministic output makes explicit context, provenance, and reviewable intermediate states increasingly important. STAMPED () defines seven properties for organizing these materials as a durable research object. -We applied STAMPED while reproducing an analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). -The reproduction used AI-assisted development, allowing us to examine how STAMPED structures the context, evidence, and reviewable checkpoints surrounding automated work. -The case study shows how the principles guide concrete decisions about the organization, execution, documentation, and distribution of a computational analysis. +We used STAMPED to assess an existing analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). +We reworked the analysis to more closely align with the ideals laid out in the STAMPED principles. +We made extensive use of AI in the development, testing, and documentation of the work, helping to assess how such practices interact with the application of the principles. -Together, the seven STAMPED properties characterize a composed research object: +Given an analysis pipeline and a chosen set of tools, we created a research object more closely aligned with the principles: - **Self-contained:** Everything needed to understand and run the study was available within one research-object boundary, eliminating the need to look elsewhere for context (supported by DataLad and git-annex). - **Tracked:** Results were linked to their inputs, transformations, execution context, and scientific meaning so their provenance could be inspected (recorded with Git, DataLad run records, `con-duct`, and NIDM/PROV). @@ -31,13 +31,10 @@ Together, the seven STAMPED properties characterize a composed research object: - **Ephemeral:** Fresh, disposable execution tested whether results followed from the declared research object rather than accumulated machine state (tested with BABS, Slurm, and clean-installation replay with `datalad rerun` and alike). - **Distributable:** Exact permitted research states remained persistently retrievable under explicit licensing and access rules, supporting reuse while protecting controlled data (implemented through Git/annex siblings and separate access boundaries). -Although this project demonstrates STAMPED through a scientific reproduction, its broader value lies in applying the principles from the start along with the FAIR principles mandating use of agreed upon standards. -Designing the research object alongside the analysis makes later reproduction more thorough and achievable while strengthening review, collaboration, extension, and reuse throughout the study. +Collectively, the decisions on how to implement the principles provide a worked example of how to use STAMPED to guide choices about research-object boundaries, provenance, execution, validation, and distribution. +The worked example exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. +Although we demonstrate the approach through a scientific reproduction, such an effort is more convenient if integrated from the start of the analysis. -Within this structure, AI-assisted tools can help translate scientific intent into workflows and automate routine implementation, testing, and documentation. -Without explicit controls, those efficiencies can amplify hidden assumptions, undocumented changes, fragile dependencies, and difficult-to-inspect outputs. -STAMPED mitigates these risks by bounding context, preserving provenance, defining executable tasks and component interfaces, testing independence from local state, and preserving explicit states for human review. -This allows researchers to benefit from automation while preserving oversight and devoting more attention to scientific questions, methods, and interpretation. ## Acknowledgments @@ -78,9 +75,5 @@ Research software engineers help determine whether a computational result is mer STAMPED gives RSEs and researchers a shared vocabulary for the properties that make this difference: coherent research-object boundaries, exact identities, executable procedures, independent components, explicit environments, clean execution, and persistent distribution. This case study translates those properties into a concrete implementation and shows how they can be designed into an analysis from the beginning rather than added after publication. -The contribution is a worked example of how STAMPED principles guide choices about research-object boundaries, provenance, execution, validation, and distribution. -It exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. -Although we demonstrate the approach through a scientific reproduction, such a reproduction is more thorough and achievable when the necessary context, procedures, and evidence are incorporated into the original analysis. - The project also addresses the conference theme, "Advancing Science in the Age of AI," by shifting attention from what AI can do to what evidence AI-assisted work leaves behind. Using STAMPED as a shared point of reference, the poster invites RSEs to compare how their teams record decisions, divide responsibility, and evaluate AI-assisted work, and to consider how those expectations could become community review criteria and institutional practice across research domains. From 81fe828d311d0dc6dcb1e0e06cedc9ddcd9a6170 Mon Sep 17 00:00:00 2001 From: leej3 Date: Thu, 6 Aug 2026 17:20:39 -0400 Subject: [PATCH 09/17] State the practical payoff for RSEs Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-sol --- 2026-usrse/stamped-abcd-poster-draft.md | 1 + 1 file changed, 1 insertion(+) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index deddd1b..525ffb7 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -14,6 +14,7 @@ STAMPED, reproducibility, research objects, neuroimaging, provenance, AI-assiste The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and re-execute or extend the analysis. Code alone is insufficient: data, parameters, software environments, execution context, and provenance must also be identifiable and recoverable. +For RSEs, the practical payoff is less time reconstructing context and more efficient review, maintenance, collaboration, and extension of an analysis. AI-assisted tools can reduce software-development effort, but their high-volume, nondeterministic output makes explicit context, provenance, and reviewable intermediate states increasingly important. STAMPED () defines seven properties for organizing these materials as a durable research object. From db1597e2d5c8db7f48d1a4b43d555a37aa35d56b Mon Sep 17 00:00:00 2001 From: leej3 Date: Thu, 6 Aug 2026 17:21:09 -0400 Subject: [PATCH 10/17] Compress the STAMPED implementation summary Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-sol --- 2026-usrse/stamped-abcd-poster-draft.md | 11 ++--------- 1 file changed, 2 insertions(+), 9 deletions(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index 525ffb7..a96ec4e 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -22,15 +22,8 @@ We used STAMPED to assess an existing analysis presented at OHBM 2025 on age-dep We reworked the analysis to more closely align with the ideals laid out in the STAMPED principles. We made extensive use of AI in the development, testing, and documentation of the work, helping to assess how such practices interact with the application of the principles. -Given an analysis pipeline and a chosen set of tools, we created a research object more closely aligned with the principles: - -- **Self-contained:** Everything needed to understand and run the study was available within one research-object boundary, eliminating the need to look elsewhere for context (supported by DataLad and git-annex). -- **Tracked:** Results were linked to their inputs, transformations, execution context, and scientific meaning so their provenance could be inspected (recorded with Git, DataLad run records, `con-duct`, and NIDM/PROV). -- **Actionable:** Procedures and expected outputs were encoded as executable, testable operations instead of instructions requiring interpretation (implemented through BIDS, tested BIDS Apps, Pixi tasks, manifests, validation commands, and Datalad run records). -- **Modular:** Data, environments, operations, and results retained distinct identities so components could be reviewed, updated, and reused independently (organized as versioned DataLad subdatasets). -- **Portable:** Software dependencies and interfaces were separated from site-specific configuration so execution could move across systems (supported by locked Pixi environments, exact Apptainer images, and standard interfaces). -- **Ephemeral:** Fresh, disposable execution tested whether results followed from the declared research object rather than accumulated machine state (tested with BABS, Slurm, and clean-installation replay with `datalad rerun` and alike). -- **Distributable:** Exact permitted research states remained persistently retrievable under explicit licensing and access rules, supporting reuse while protecting controlled data (implemented through Git/annex siblings and separate access boundaries). +To improve the research object across the seven properties, we combined DataLad and git-annex for composition and state; Git, DataLad run records, `con-duct`, and NIDM/PROV for provenance; tested BIDS Apps and Pixi tasks for executable interfaces; Apptainer, BABS, and Slurm for portable, fresh execution; and persistent Git/annex siblings with separate access boundaries for distribution. +Used together, these tools made data, environments, operations, and results more identifiable, executable, modular, portable, reviewable, and retrievable while exposing interactions among the principles. Collectively, the decisions on how to implement the principles provide a worked example of how to use STAMPED to guide choices about research-object boundaries, provenance, execution, validation, and distribution. The worked example exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. From d8fb257730c7897d9587be356ddf14206eb054d6 Mon Sep 17 00:00:00 2001 From: leej3 Date: Thu, 6 Aug 2026 17:21:27 -0400 Subject: [PATCH 11/17] Promise practical implementation details Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-sol --- 2026-usrse/stamped-abcd-poster-draft.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index a96ec4e..5a6f425 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -26,7 +26,7 @@ To improve the research object across the seven properties, we combined DataLad Used together, these tools made data, environments, operations, and results more identifiable, executable, modular, portable, reviewable, and retrievable while exposing interactions among the principles. Collectively, the decisions on how to implement the principles provide a worked example of how to use STAMPED to guide choices about research-object boundaries, provenance, execution, validation, and distribution. -The worked example exposes the effort, evidence, and tradeoffs involved, giving RSEs an implementation pattern they can adapt across domains. +The poster reports the practical details of this process—the effort and judgment required, problems encountered, tradeoffs made, evidence produced, and interactions among principles and tools—and invites RSEs to consider which parts apply to their own shared or domain-specific challenges. Although we demonstrate the approach through a scientific reproduction, such an effort is more convenient if integrated from the start of the analysis. From ba759cef8324d0c7df606a4481aa87adc2460bd3 Mon Sep 17 00:00:00 2001 From: leej3 Date: Thu, 6 Aug 2026 17:21:49 -0400 Subject: [PATCH 12/17] Frame the reconstruction as an AI test run Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-sol --- 2026-usrse/stamped-abcd-poster-draft.md | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index 5a6f425..f6af02d 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -20,7 +20,8 @@ STAMPED () defines seven properties for organizi We used STAMPED to assess an existing analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). We reworked the analysis to more closely align with the ideals laid out in the STAMPED principles. -We made extensive use of AI in the development, testing, and documentation of the work, helping to assess how such practices interact with the application of the principles. +We used AI extensively in development, testing, and documentation, making the reconstruction a practical test of the STAMPED vision for AI-assisted research. +In this setting, the principles provided a structure for bounding context, preserving inspectable states, connecting automated outputs to evidence, and establishing checkpoints for human review. To improve the research object across the seven properties, we combined DataLad and git-annex for composition and state; Git, DataLad run records, `con-duct`, and NIDM/PROV for provenance; tested BIDS Apps and Pixi tasks for executable interfaces; Apptainer, BABS, and Slurm for portable, fresh execution; and persistent Git/annex siblings with separate access boundaries for distribution. Used together, these tools made data, environments, operations, and results more identifiable, executable, modular, portable, reviewable, and retrievable while exposing interactions among the principles. From 1475a4186b43067f4d1ad13b421af9fa880f6997 Mon Sep 17 00:00:00 2001 From: leej3 Date: Thu, 6 Aug 2026 17:22:18 -0400 Subject: [PATCH 13/17] Use confident poster-centered language Co-Authored-By: codex-cli 0.143.0 / gpt 5.6-sol --- 2026-usrse/stamped-abcd-poster-draft.md | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index f6af02d..1b48787 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -18,13 +18,13 @@ For RSEs, the practical payoff is less time reconstructing context and more effi AI-assisted tools can reduce software-development effort, but their high-volume, nondeterministic output makes explicit context, provenance, and reviewable intermediate states increasingly important. STAMPED () defines seven properties for organizing these materials as a durable research object. -We used STAMPED to assess an existing analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). -We reworked the analysis to more closely align with the ideals laid out in the STAMPED principles. -We used AI extensively in development, testing, and documentation, making the reconstruction a practical test of the STAMPED vision for AI-assisted research. -In this setting, the principles provided a structure for bounding context, preserving inspectable states, connecting automated outputs to evidence, and establishing checkpoints for human review. +This poster presents our application of STAMPED to an existing analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). +We use the principles to guide improvements to the research object surrounding the analysis. +We use AI extensively in development, testing, and documentation, making the reconstruction a practical test of the STAMPED vision for AI-assisted research. +In this setting, the principles provide a structure for bounding context, preserving inspectable states, connecting automated outputs to evidence, and establishing checkpoints for human review. -To improve the research object across the seven properties, we combined DataLad and git-annex for composition and state; Git, DataLad run records, `con-duct`, and NIDM/PROV for provenance; tested BIDS Apps and Pixi tasks for executable interfaces; Apptainer, BABS, and Slurm for portable, fresh execution; and persistent Git/annex siblings with separate access boundaries for distribution. -Used together, these tools made data, environments, operations, and results more identifiable, executable, modular, portable, reviewable, and retrievable while exposing interactions among the principles. +To improve the research object across the seven properties, we combine DataLad and git-annex for composition and state; Git, DataLad run records, `con-duct`, and NIDM/PROV for provenance; tested BIDS Apps and Pixi tasks for executable interfaces; Apptainer, BABS, and Slurm for portable, fresh execution; and persistent Git/annex siblings with separate access boundaries for distribution. +Used together, these tools make data, environments, operations, and results more identifiable, executable, modular, portable, reviewable, and retrievable while exposing interactions among the principles. Collectively, the decisions on how to implement the principles provide a worked example of how to use STAMPED to guide choices about research-object boundaries, provenance, execution, validation, and distribution. The poster reports the practical details of this process—the effort and judgment required, problems encountered, tradeoffs made, evidence produced, and interactions among principles and tools—and invites RSEs to consider which parts apply to their own shared or domain-specific challenges. From ef051862061d21b071a621ad4e478f074365b5cf Mon Sep 17 00:00:00 2001 From: leej3 Date: Fri, 7 Aug 2026 11:07:57 -0400 Subject: [PATCH 14/17] Take another pass --- 2026-usrse/stamped-abcd-poster-draft.md | 24 +++++++++++++----------- 1 file changed, 13 insertions(+), 11 deletions(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index 1b48787..190252b 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -12,24 +12,25 @@ STAMPED, reproducibility, research objects, neuroimaging, provenance, AI-assiste ## Abstract -The scientific value of a computational finding depends on whether others can understand how it was produced, inspect its evidence, and re-execute or extend the analysis. -Code alone is insufficient: data, parameters, software environments, execution context, and provenance must also be identifiable and recoverable. -For RSEs, the practical payoff is less time reconstructing context and more efficient review, maintenance, collaboration, and extension of an analysis. -AI-assisted tools can reduce software-development effort, but their high-volume, nondeterministic output makes explicit context, provenance, and reviewable intermediate states increasingly important. -STAMPED () defines seven properties for organizing these materials as a durable research object. +RSEs routinely encounter computational analyses whose code is available but whose data, parameters, software environments, execution context, and provenance are difficult to reconstruct. +This missing context makes analyses harder to review, maintain, transfer, and extend. +AI-assisted tools can accelerate development, testing, and documentation, but their high-volume, nondeterministic output further complicates these tasks and heightens the need for explicit context and reviewable intermediate states. +STAMPED (https://stamped-principles.org) defines seven properties for organizing these materials as a durable and more useful research object. This poster presents our application of STAMPED to an existing analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). We use the principles to guide improvements to the research object surrounding the analysis. We use AI extensively in development, testing, and documentation, making the reconstruction a practical test of the STAMPED vision for AI-assisted research. -In this setting, the principles provide a structure for bounding context, preserving inspectable states, connecting automated outputs to evidence, and establishing checkpoints for human review. -To improve the research object across the seven properties, we combine DataLad and git-annex for composition and state; Git, DataLad run records, `con-duct`, and NIDM/PROV for provenance; tested BIDS Apps and Pixi tasks for executable interfaces; Apptainer, BABS, and Slurm for portable, fresh execution; and persistent Git/annex siblings with separate access boundaries for distribution. -Used together, these tools make data, environments, operations, and results more identifiable, executable, modular, portable, reviewable, and retrievable while exposing interactions among the principles. +Following a review of the original analysis, we use a coordinated set of tools to improve the research object across the seven properties: DataLad and git-annex for composition and versioned state; Git, DataLad run records, con-duct, and NIDM/PROV for provenance; tested BIDS Apps and Pixi tasks for executable interfaces; Apptainer, BABS, and Slurm for portable, fresh execution; and persistent Git/annex siblings with separate access boundaries for distribution. +We show how these tools work in concert to make the data, environments, operations, and results more identifiable, executable, modular, portable, reviewable, and retrievable. Collectively, the decisions on how to implement the principles provide a worked example of how to use STAMPED to guide choices about research-object boundaries, provenance, execution, validation, and distribution. The poster reports the practical details of this process—the effort and judgment required, problems encountered, tradeoffs made, evidence produced, and interactions among principles and tools—and invites RSEs to consider which parts apply to their own shared or domain-specific challenges. Although we demonstrate the approach through a scientific reproduction, such an effort is more convenient if integrated from the start of the analysis. +The poster reports the practical details of this process: the effort and judgment required, problems encountered, tradeoffs made, evidence produced, and interactions among principles and tools. +In doing so, it shows why these practices are easier to incorporate during study design than to add retrospectively. +It invites RSEs to compare these experiences with their own shared or domain-specific challenges. ## Acknowledgments @@ -68,7 +69,8 @@ RRID:SCR_025436. Research software engineers help determine whether a computational result is merely produced once or becomes a durable scientific contribution. STAMPED gives RSEs and researchers a shared vocabulary for the properties that make this difference: coherent research-object boundaries, exact identities, executable procedures, independent components, explicit environments, clean execution, and persistent distribution. -This case study translates those properties into a concrete implementation and shows how they can be designed into an analysis from the beginning rather than added after publication. +This poster grounds that vocabulary in a practical implementation, showing how RSE decisions about provenance, execution, validation, and distribution shape the reviewability and reuse of computational research across domains. -The project also addresses the conference theme, "Advancing Science in the Age of AI," by shifting attention from what AI can do to what evidence AI-assisted work leaves behind. -Using STAMPED as a shared point of reference, the poster invites RSEs to compare how their teams record decisions, divide responsibility, and evaluate AI-assisted work, and to consider how those expectations could become community review criteria and institutional practice across research domains. +The project also addresses the conference theme, “Advancing Science in the Age of AI,” by shifting attention from what AI can do to what evidence AI-assisted work leaves behind. +The poster invites RSEs to compare how their teams bound context, record decisions, divide work between automation and human reviewers, and evaluate AI-assisted outputs. +These comparisons can help identify shared expectations for reviewing AI-assisted research and inform longer-term community and institutional practices. \ No newline at end of file From 0f41fa0d15c0ba2464f82ab69084ec3fc55bd7ed Mon Sep 17 00:00:00 2001 From: leej3 Date: Fri, 7 Aug 2026 11:43:12 -0400 Subject: [PATCH 15/17] fix formatting --- 2026-usrse/Makefile | 4 +++ 2026-usrse/stamped-abcd-poster-draft.md | 39 ++++++++++--------------- 2 files changed, 20 insertions(+), 23 deletions(-) diff --git a/2026-usrse/Makefile b/2026-usrse/Makefile index c5c9c3d..d1cf654 100644 --- a/2026-usrse/Makefile +++ b/2026-usrse/Makefile @@ -65,6 +65,10 @@ PDF_LATEX_OPTS := --to pdf \ -V toccolor=darkgreen \ -V 'header-includes=\definecolor{darkblue}{HTML}{00008B}\definecolor{darkgreen}{HTML}{006400}' +# This draft is longer than the one-page abstract template; keep this +# adjustment scoped to this document rather than changing shared defaults. +stamped-abcd-poster-draft.pdf: PDF_LATEX_OPTS := $(subst fontsize=11pt,fontsize=10pt,$(PDF_LATEX_OPTS)) + # PDF via HTML (wkhtmltopdf / weasyprint) — same look as the HTML render PDF_HTML_OPTS := --to html5 --css=https://cdn.jsdelivr.net/npm/water.css@2/out/water.css diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index 190252b..87922c2 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -32,6 +32,10 @@ The poster reports the practical details of this process: the effort and judgmen In doing so, it shows why these practices are easier to incorporate during study design than to add retrospectively. It invites RSEs to compare these experiences with their own shared or domain-specific challenges. +```{=latex} +\newpage +``` + ## Acknowledgments We thank the authors of the STAMPED principles and the developers and communities behind DataLad, BIDS, BABS, ReproNim, NIDM, Pixi, Apptainer, and `con-duct`. @@ -42,28 +46,17 @@ The authors reviewed the source evidence, edited the text, and remain responsibl ## References -1. Macdonald A, Baker CC, To I, Halchenko YO. -*STAMPED principles for reproducible research objects*. -May 2026. - -2. Lee JA. -*STAMPED-dl_morphometrics_biases: an ideal-oriented reconstruction of a neuroimaging analysis*. -2026. - -3. Nielson DM, Lee JA, Earl E, Moraczewski D, Pereira F. *Age dependent volume estimation biases in recon-all clinical and recon-any*. -OHBM 2025 poster. - -4. Halchenko YO, et al. DataLad: distributed system for joint management of code, data, and their relationship. -*Journal of Open Source Software*. -2021;6(63):3262. - -5. Zhao C, et al. A reproducible and generalizable software workflow for analysis of large-scale neuroimaging data collections using BIDS Apps. -*Imaging Neuroscience*. 2024;2:imag-2-00074. - -6. Center for Open Neuroscience. -*con-duct: a lightweight wrapper for monitoring command execution*. -RRID:SCR_025436. - +1. Macdonald A, Baker CC, To I, Halchenko YO; *STAMPED principles for reproducible research objects*; May 2026; [preprint](https://github.com/stamped-principles/stamped-paper) + +2. Lee JA; *STAMPED-dl_morphometrics_biases: an ideal-oriented reconstruction of a neuroimaging analysis*; 2026; [repository](https://github.com/STAMPED-dl-morphometrics-biases/STAMPED-dl_morphometrics_biases) + +3. Nielson DM, Lee JA, Earl E, Moraczewski D, Pereira F; *Age dependent volume estimation biases in recon-all clinical and recon-any*; OHBM 2025 poster; [doi:10.17605/OSF.IO/P3KNS](https://doi.org/10.17605/OSF.IO/P3KNS) + +4. Halchenko YO, et al.; DataLad: distributed system for joint management of code, data, and their relationship; *Journal of Open Source Software*; 2021;6(63):3262; [doi:10.21105/joss.03262](https://doi.org/10.21105/joss.03262) + +5. Zhao C, et al.; A reproducible and generalizable software workflow for analysis of large-scale neuroimaging data collections using BIDS Apps; *Imaging Neuroscience*; 2024;2:imag-2-00074; [doi:10.1162/imag_a_00074](https://doi.org/10.1162/imag_a_00074) + +6. Center for Open Neuroscience; *con-duct: a lightweight wrapper for monitoring command execution*; RRID:SCR_025436; [repository](https://github.com/con/duct) ## Connection to Mission, Goals, & Interests of US-RSE Community @@ -73,4 +66,4 @@ This poster grounds that vocabulary in a practical implementation, showing how R The project also addresses the conference theme, “Advancing Science in the Age of AI,” by shifting attention from what AI can do to what evidence AI-assisted work leaves behind. The poster invites RSEs to compare how their teams bound context, record decisions, divide work between automation and human reviewers, and evaluate AI-assisted outputs. -These comparisons can help identify shared expectations for reviewing AI-assisted research and inform longer-term community and institutional practices. \ No newline at end of file +These comparisons can help identify shared expectations for reviewing AI-assisted research and inform longer-term community and institutional practices. From f723ca9b8897d3d9a37166bae6ff9bc7bde3a054 Mon Sep 17 00:00:00 2001 From: leej3 Date: Fri, 7 Aug 2026 11:44:47 -0400 Subject: [PATCH 16/17] Restore inline citations in USRSE abstract Co-Authored-By: codex-cli 0.143.0 / gpt-5.6-sol --- 2026-usrse/stamped-abcd-poster-draft.md | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index 87922c2..e353c14 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -15,13 +15,13 @@ STAMPED, reproducibility, research objects, neuroimaging, provenance, AI-assiste RSEs routinely encounter computational analyses whose code is available but whose data, parameters, software environments, execution context, and provenance are difficult to reconstruct. This missing context makes analyses harder to review, maintain, transfer, and extend. AI-assisted tools can accelerate development, testing, and documentation, but their high-volume, nondeterministic output further complicates these tasks and heightens the need for explicit context and reviewable intermediate states. -STAMPED (https://stamped-principles.org) defines seven properties for organizing these materials as a durable and more useful research object. +STAMPED (https://stamped-principles.org) defines seven properties for organizing these materials as a durable and more useful research object [1]. -This poster presents our application of STAMPED to an existing analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org). -We use the principles to guide improvements to the research object surrounding the analysis. +This poster presents our application of STAMPED to an existing analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org) [3]. +We use the principles to guide improvements to the research object surrounding the analysis [2]. We use AI extensively in development, testing, and documentation, making the reconstruction a practical test of the STAMPED vision for AI-assisted research. -Following a review of the original analysis, we use a coordinated set of tools to improve the research object across the seven properties: DataLad and git-annex for composition and versioned state; Git, DataLad run records, con-duct, and NIDM/PROV for provenance; tested BIDS Apps and Pixi tasks for executable interfaces; Apptainer, BABS, and Slurm for portable, fresh execution; and persistent Git/annex siblings with separate access boundaries for distribution. +Following a review of the original analysis, we use a coordinated set of tools to improve the research object across the seven properties: DataLad and git-annex for composition and versioned state [4]; Git, DataLad run records, con-duct, and NIDM/PROV for provenance [4,6]; tested BIDS Apps and Pixi tasks for executable interfaces [5]; Apptainer, BABS, and Slurm for portable, fresh execution [5]; and persistent Git/annex siblings with separate access boundaries for distribution [4]. We show how these tools work in concert to make the data, environments, operations, and results more identifiable, executable, modular, portable, reviewable, and retrievable. Collectively, the decisions on how to implement the principles provide a worked example of how to use STAMPED to guide choices about research-object boundaries, provenance, execution, validation, and distribution. @@ -61,7 +61,7 @@ The authors reviewed the source evidence, edited the text, and remain responsibl ## Connection to Mission, Goals, & Interests of US-RSE Community Research software engineers help determine whether a computational result is merely produced once or becomes a durable scientific contribution. -STAMPED gives RSEs and researchers a shared vocabulary for the properties that make this difference: coherent research-object boundaries, exact identities, executable procedures, independent components, explicit environments, clean execution, and persistent distribution. +STAMPED gives RSEs and researchers a shared vocabulary for the properties that make this difference: coherent research-object boundaries, exact identities, executable procedures, independent components, explicit environments, clean execution, and persistent distribution [1]. This poster grounds that vocabulary in a practical implementation, showing how RSE decisions about provenance, execution, validation, and distribution shape the reviewability and reuse of computational research across domains. The project also addresses the conference theme, “Advancing Science in the Age of AI,” by shifting attention from what AI can do to what evidence AI-assisted work leaves behind. From e4b6e4e8fa0adacb9500aec1cd36b04d0817be0f Mon Sep 17 00:00:00 2001 From: leej3 Date: Fri, 7 Aug 2026 11:57:45 -0400 Subject: [PATCH 17/17] Add FAIR back in Co-Authored-By: codex-cli 0.143.0 / gpt-5.6-sol --- 2026-usrse/stamped-abcd-poster-draft.md | 3 +++ 1 file changed, 3 insertions(+) diff --git a/2026-usrse/stamped-abcd-poster-draft.md b/2026-usrse/stamped-abcd-poster-draft.md index e353c14..f2b01e5 100644 --- a/2026-usrse/stamped-abcd-poster-draft.md +++ b/2026-usrse/stamped-abcd-poster-draft.md @@ -16,6 +16,7 @@ RSEs routinely encounter computational analyses whose code is available but whos This missing context makes analyses harder to review, maintain, transfer, and extend. AI-assisted tools can accelerate development, testing, and documentation, but their high-volume, nondeterministic output further complicates these tasks and heightens the need for explicit context and reviewable intermediate states. STAMPED (https://stamped-principles.org) defines seven properties for organizing these materials as a durable and more useful research object [1]. +The framework complements the established FAIR (Findable, Accessible, Interoperable, and Reusable) principles by focusing on the organization and execution of computational research objects [7]. This poster presents our application of STAMPED to an existing analysis presented at OHBM 2025 on age-dependent bias in cortical morphometry tools using Adolescent Brain Cognitive Development (ABCD) Study data (https://abcdstudy.org) [3]. We use the principles to guide improvements to the research object surrounding the analysis [2]. @@ -58,6 +59,8 @@ The authors reviewed the source evidence, edited the text, and remain responsibl 6. Center for Open Neuroscience; *con-duct: a lightweight wrapper for monitoring command execution*; RRID:SCR_025436; [repository](https://github.com/con/duct) +7. Wilkinson MD, Dumontier M, Aalbersberg IJ, et al.; *The FAIR Guiding Principles for scientific data management and stewardship*; *Scientific Data*; 2016;3:160018; [doi:10.1038/sdata.2016.18](https://doi.org/10.1038/sdata.2016.18) + ## Connection to Mission, Goals, & Interests of US-RSE Community Research software engineers help determine whether a computational result is merely produced once or becomes a durable scientific contribution.