Visible, Trackable, Forkable: Opening the Process of Science

Sergey V. Samsonau · 2026

Read Visible, Trackable, Forkable: Opening the Process of Science by Sergey V. Samsonau (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-10932.txt.

Introduction

Samsonau argues science should open its process, not its outputs. Current practice shows conclusions but hides their making: private work, polished papers, errors punished by retraction, not amendment. Two decades of reform opened papers, data, code, notebooks, and protocols, yet the unit stayed finished, never the thinking process: shifting questions, dead ends, and shifts. Opening the process would speed progress, widen access, raise trust, and scale quality checks. The workflow needs visible, trackable, forkable work, with changes recorded, and branching open from any point.

Catalog record

CREATORSergey V. Samsonau
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39782
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.10932

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Visible, Trackable, Forkable: Opening the Process of Science Sergey V. Samsonau Abstract The way science is currently practiced shows conclusions but hides how they were reached. Researchers work privately, polish their results, publish a finished paper, and defend it. Errors are punished by retraction rather than corrected by amendment. Alternative directions are pursued through competing papers with no shared history. The reasoning, the dead ends, the trade-offs, the corrections: everything that would let others understand how a conclusion was reached is invisible. Two decades of open science reform have addressed this by opening specific artifacts: papers, data, code, notebooks, protocols. Each is valuable, but the unit remains a finished product. None opens the thinking process itself: the evolving sequence of questions, interpretations, dead ends, and direction changes that constitutes the actual scientific contribution. This paper argues that opening the process of science (not just its outputs) would produce a step change in the speed of scientific progress, the accessibility of scientific reasoning, the trustworthiness of scientific claims, and the scalability of scientific quality assurance. We identify three properties the workflow needs: visible (the process is open, not just the product), trackable (every change is recorded and attributable), and forkable (anyone can branch from any point with shared history preserved). A visible, trackable flow is inherently verifiable: by humans, by automated tools, by AI agents. Software development adopted this flow decades ago, and the results (faster correction, broader contribution, maintained quality at scale) demonstrate the opportunity for science. Keywords: open science, scientific workflow, version control, reproducibility, process transparency, scientific reform 1 Introduction Two fundamentally different workflows exist for organizing complex intellectual work. In the first, practitioners work privately for months or years, producing a finished artifact (comprehensive, polished, internally consistent). They submit it to gatekeepers who evaluate it. If accepted, it enters the permanent record, essentially unmodifiable. If an error is later found, the artifact is retracted: removed from the record, with the author’s name attached to the failure. Others who wish to take the work in a different direction must start from scratch. Contribution is measured by the number of finished artifacts produced and how often others reference them. In the second, work is continuously recorded. Every change, every decision, every correction is documented. Others can see what has been done, report problems, propose corrections, and receive credit for their contributions. If an error is made, it is fixed; the history shows both the error and the correction, and no one pretends the error did not happen. If someone disagrees with a decision made at step 47, they can take a complete copy of the work up to step 46 and pursue their alternative, with the full shared history preserved. Contribution is measured by what was actually done, not only by finished products shipped. The first is how science is currently practiced. The second is how software development works. The difference is not a matter of tools or platforms. It is a difference in flow : in the basic sequence of how work is done, how errors are handled, how contributions are tracked, and how disagreements are resolved. This paper argues that science’s workflow lacks three specific properties that software’s workflow has, and that these missing properties (not peer review failures, not perverse incentives, not individual negligence) are the structural root of the problems science currently faces. The three properties are: Visible. The process of research is open, not just the finished product. Currently, science publishes results but hides the work that produced them: the failed approaches, the corrected errors, the methodological trade-offs, the reasoning behind decisions. When people say “open science,” they mean open outputs. The process remains closed. Trackable. Every change, every decision, every correction is recorded and attributable, at any granularity, from a single correction to a complete analysis. Currently, science records only one unit: the finished paper. Forkable. Anyone can branch from any point in the work to pursue a different direction, with the shared history preserved. Currently, pursuing an alternative requires starting over, and the relationship between competing approaches is opaque. A flow with these three properties is inherently verifiable. If the process is visible and every step is tracked, then anyone (a peer reviewer, an automated tool, an AI agent) can check the work at any point. Verification becomes a natural activity within the flow, not a separate gatekeeping event applied to a finished artifact months after publication. The key insight is that verification does not require a separate property; it requires visibility and trackability. You cannot check what you cannot see, and you cannot see what is not recorded. These three properties are ordered by dependency. Visibility is the most fundamental: you cannot track or fork a process that is hidden. Trackability enables forkability (you can only branch meaningfully from a point you can identify in a recorded history). Together, they constitute a different flow, a different way of working that produces different outcomes. Related work. A substantial literature has addressed opening specific artifacts of the research process. Open access made papers freely readable. The FAIR principles [ 10 ] established standards for data findability, accessibility, interoperability, and reusability. Open Notebook Science proposed making the laboratory notebook (raw data, protocols, experimental records) publicly available in real time. The TOP Guidelines [ 6 ] formalized transparency standards for design, data, code, and materials. Munafò et al. [ 5 ] provided the most comprehensive catalogue of threats to reproducibility (HARKing, publication bias, p-hacking, low statistical power) and proposed reforms across methods, reporting, incentives, and evaluation, but each reform operates within the existing workflow: adding checkpoints, improving reporting, opening specific artifacts. Lyon [ 4 ] identified transparency as a “third dimension” of open science beyond access and participation. Ram [ 8 ] argued that version control systems like Git can improve reproducibility and transparency in research. Registered Reports pre-register methodology before data collection. Thibault et al. [ 9 ] envision a “truly collaborative research ecosystem” and come closest to the present argument, noting that version control and forking could maintain transparent relationships between evolving research items; but their examples are protocols and code, still finished artifacts under version control rather than the thinking process itself. These contributions share a common frame: they ask for more artifacts, or earlier artifacts, or more open artifacts, but the unit remains a finished product. Open access opens the paper. Open data opens the dataset. Open code opens the scripts. Registered Reports open the methodology before results are known. Even Open Notebook Science opens the laboratory record: what was done, what reagents were used, what measurements were taken. All of these are valuable. None of them open the thinking process itself. The present paper argues for something different: opening the full thinking process. Scientists are not technicians following established procedures; their primary contribution is original thinking: forming questions, interpreting observations, constructing theories, designing methods, building original tools and instruments to study nature in new ways, making connections across domains, changing direction when evidence demands it. It is this evolving sequence of thought that constitutes the actual scientific contribution, and it is precisely this that the current workflow hides. Not just what data were collected, but what new observations arrived this week

Read the complete text