Exposing the Hidden Middle of Scientific Research

This direction is merged into another direction: methodological transparency. Its last version stays readable and citable here.

Version 1 of 112 papers19 conceptsfrozen 2026-10-04

Scientific progress is often presented as a clean, linear progression from hypothesis to discovery. However, the actual practice of research is a "black box" of messy iterations, failed experiments, and undocumented technical choices. While traditional scholarly communication focuses on the final "output"—the peer-reviewed paper—a growing body of research argues that we must instead document the entire research process lifecycle.

By making the "hidden middle" of research transparent, we can improve reproducibility, enable better AI reasoning, and move from simply sharing results to practicing "Science in the Open."

The Provenance Gap: Beyond Who to How

In structured knowledge bases, we often know who made a claim, but rarely how they arrived at it. Kody Moodley, Michel Dumontier, and colleagues argue in A model for capturing provenance of assertions about chemical substances that current platforms focus on authorship while omitting the generating process.

Without metadata describing specific experimental or analytical procedures, it is impossible for external researchers to evaluate the scientific rigor of an assertion. This lack of "procedural provenance" means that even when data is open, it is not necessarily verifiable.

The Reporting Discrepancy

Even when researchers attempt to be thorough, there is a significant gap between their ability to describe a study's background and their ability to document its technical execution. In a systematic review of public health dissertations, ShuangYang Dai and Xiaobin Zhou found that while introductory sections were well-handled, critical methodological details—such as how missing data was managed or whether sensitivity analyses were performed—were frequently omitted (Evaluation of the reporting quality of observational studies in Master of Public Health’s dissertations in China: a systematic review).

This suggests that the "meat" of research—the technical decisions that determine a study's robustness—is often the part least visible to the community.

Mapping the Research Ecosystem

To solve this, researchers are proposing new semantic frameworks to capture the "tacit knowledge" of the lab. Enrico Daga and Marilena Daquino describe a Research Ecosystem that uses GitHub-based annotations and knowledge graphs to map the interdependencies between data, software, and methodology.

The goal is to move beyond documenting final outputs to documenting the "data journey." This approach treats the intermediate steps—the data cleaning, the software dependencies, and the methodological rationales—as first-class scientific citizens that should be machine-readable and citable.

From Open Science to Science in the Open

A more radical approach to transparency is being pioneered by groups like EleutherAI. In EleutherAI: Going Beyond "Open Science" to "Science in the Open", Jason Phang, Stella Biderman, and others describe a model where the entire research process happens in public view (e.g., on Discord).

This "Science in the Open" approach captures:

  • Failed experiments: Paths that didn't work, which are almost never published.
  • Informal trade-offs: The "why" behind technical decisions that never make it into a formal paper.
  • Real-time debugging: The iterative process of refining a model or method.

The Challenge of Tacit Knowledge for AI

The importance of this "hidden middle" is most evident when we try to automate science. The CritPt benchmark, developed by Minhui Zhu and a large team of physicists, reveals that even advanced LLMs struggle with frontier research because they lack access to "insider knowledge."

Papers often omit tedious intermediate calculations and subtle assumptions. Because AI models are trained primarily on these polished final outputs, they fail to navigate the messy, multi-step reasoning required to conduct original research. Bridging this "reasoning gap" requires documenting the very things we currently leave out of our publications.

Conclusion

The move toward documenting the research process is a move toward a more robust and inclusive scientific enterprise. Whether through dedicated platforms for technical workflows like the NEUBIAS Gateway or semantic frameworks for the research lifecycle, the goal is the same: to stop treating the process of discovery as a black box and start treating it as the core of the scientific record.

Papers in this version

History & citationVersion 1 · 2026-10-04

Merged into another direction · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_exposing_the_hidden_middle_of_scientific_31b44709,
  title = {Exposing the Hidden Middle of Scientific Research},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/exposing-the-hidden-middle-of-scientific-research/rdr_31b4470929664daca6a29056338de2d5},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_exposing_the_hidden_middle_of_scientific_31b44709_v1,
  title = {Exposing the Hidden Middle of Scientific Research},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/exposing-the-hidden-middle-of-scientific-research/rdr_31b4470929664daca6a29056338de2d5/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}