The hottest Reproducibility Substack posts right now

And their main takeaways
Category
Top Science Topics

Science Has a Major Fraud Problem

Common Sense with Bari Weiss • 2323 implied HN points • 10 Mar 26
🔬 Science Reproducibility
  1. Fraudulent and manipulated data happen in science, and even high-profile papers and labs can be wrong or dishonest.
  2. Online forums and independent sleuths—including students and journalists—are playing a big role in finding and exposing bad science that institutions missed.
  3. Academic incentives and prestige often protect powerful researchers and can punish those who insist on honest, rigorous work, making it harder to fix the problem.

One Weird Trick to Get Significant Results

Cremieux Recueil • 295 implied HN points • 13 Mar 26
🔬 Science Reproducibility
  1. Researchers often split samples and hunt for subgroups where effects become significant, but reporting subgroup "wins" without testing interactions or accounting for low power produces misleading, likely fluke results.
  2. The functional medicine trial example shows clear red flags: inconsistent numbers, bad or post-hoc preregistration, incorrect power/sample-size math, undisclosed conflicts, non-ITT analyses, and unreported/misused subgroup tests with weak measures.
  3. These practices make findings fragile and hard to replicate, so studies need proper prospective registration, correct power calculations, transparent reporting (including interaction tests), multiple-comparisons control, and shared data to be trustworthy.

Inventing the Methods Section

Asimov Press • 380 implied HN points • 12 Jan 26
🔬 Science Reproducibility
  1. Over time, methods went from practical, detailed recipes to short, sidelined Methods sections, and that shift makes many experiments hard or slow to reproduce.
  2. A lot of essential lab know-how is tacit and doesn’t fit cleanly into text, so videos, protocol repositories, and supplements help but face sustainability and credit problems and still treat methods as second-class outputs.
  3. Fixing this requires new infrastructure (versioning, executable protocols, automation, recorded workflows, cloud labs) and changing incentives so people are rewarded for sharing and improving methods, not just for novel results.

Brief Notes on Scientific Critique

Cremieux Recueil • 434 implied HN points • 27 Dec 25
🔬 Science Reproducibility
  1. Make sure your criticism is correct: check the data, run the needed analyses, and only accuse or declare problems when you can justify them.
  2. Focus on meaningful, relevant issues that actually change conclusions — don’t list hypotheticals; quantify or demonstrate how a confound or error would affect the results.
  3. Be generous and contextual: assume good faith, ask for clarification or contact authors privately when fixable, and build enough domain knowledge to notice real problems instead of relying on rote one‑liners.

Update on Metascience-Related Language in the NIH Budget

The Good Science Project • 59 implied HN points • 24 Jan 26
🏥 Health Politics Reproducibility
  1. Lawmakers barred NIH and other agencies from changing how negotiated indirect cost rates are calculated or pursuing rulemaking to alter the 2017 approach, while asking agencies to discuss transparency improvements and consider models like FAIR.
  2. The bill encourages expanding person-focused grants (like R35/MIRA) and boosting support for early-career researchers, but it rejected a House proposal for $100M in replication funding and only asks NIH to encourage and brief the committee on replication efforts.
  3. Committees directed NIH to tackle high article-processing charges, promote alternatives to animal research, allow international subawards for clinical trials, and reduce administrative burden, while saying any major NIH restructuring must follow statutory notice rules.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

Back and Forth on the Value of Replication

The Good Science Project • 48 implied HN points • 29 Jan 26
🔬 Science Reproducibility
  1. Replicating studies early usually gives much bigger returns because it can stop entire lines of follow-on work from chasing a wrong result, though some older papers that still drive current research can also be worth replicating.
  2. Citation counts are an imperfect measure of influence, and once a paper's findings are deeply embedded across many follow-on studies, a single replication may not undo that influence—so sometimes it's higher impact to replicate key descendant papers instead of only the original.
  3. The impact of replication can be increased by changing incentives and communication: funders and journals can publicize replication results, link them to original papers, and adjust funding or citation expectations to make replications matter more.

Making Science More Reproducible

Briefly Bio • 158 implied HN points • 18 Jul 24
🔬 Science Reproducibility
  1. Reproducibility in science is a big issue, with many experiments failing to be duplicated. This creates a challenge for scientists trying to build on each other's work.
  2. Clear and detailed documentation of scientific processes is crucial. When scientists share their methods well, it helps others replicate results more easily.
  3. Using technology like structured documentation can improve transparency in research. This way, scientists can better understand what happened in an experiment and learn from it.

Reproducibility starts with good communication

Briefly Bio • 59 implied HN points • 22 Jul 24
🔬 Science Reproducibility
  1. Good communication is key to making experiments reproducible. If scientists can't share their methods clearly, it’s hard for others to repeat their work.
  2. Many methods of communication in science are outdated and not effective. This can lead to misunderstandings and mistakes in reproducing experiments.
  3. Tools that help streamline communication in science can improve reproducibility. A better system can replace complicated and unclear documentation.

Recent Metascience News/Links

The Good Science Project • 40 implied HN points • 18 Dec 25
🔬 Science Reproducibility
  1. Even though we spend much more on science and R&D than in the past, the bottleneck for economic growth is often our ability to translate discoveries into marketable products, not a shortage of new ideas.
  2. Research funding and review rules are shifting: NSF is allowing fewer outside reviews and giving program managers more discretion, and NIH has removed the old requirement to get advance permission for very large grant applications.
  3. Reproducibility and data-quality problems keep appearing in areas like crystallography, and analysts caution against treating measures like “variance explained” as if they directly show a variable’s causal impact.

A Good Science Manifesto

The Good Science Project • 26 implied HN points • 11 Dec 25
🔬 Science Reproducibility
  1. Science funding should prioritize producing reliable, useful knowledge and reward being right, supporting both risky exploratory work and goal-oriented projects.
  2. Funders must cut heavy administrative burdens and require open sharing of data and methods so others can verify and build on results quickly.
  3. The funding system should be more flexible and diverse: experiment with new funding models, provide stable support for infrastructure and staff scientists, and distribute support more evenly across career stages.

The reproducibility crisis and what it means for the NIH

Vinay Prasad's Observations and Thoughts • 111 implied HN points • 12 Feb 25
🔬 Science Reproducibility
  1. Many scientific experiments don't give the same results when repeated. It's like trying to bake a cake and it flops even though the recipe seemed good.
  2. Scientists often face pressure to produce results, which can lead to mistakes or even dishonesty in their work. They tend to focus more on getting results than on finding the truth.
  3. There's a big need for change in how science operates. Even if some discoveries are useful, there's a concern about whether research is really delivering reliable results.

How to Improve Both Scientific Innovation and Reproducibility at Once

The Good Science Project • 14 implied HN points • 04 Nov 24
🔬 Science Reproducibility
  1. Science struggles with two main issues: not being able to reproduce results and not being as innovative as before. Many studies can't be repeated successfully, which raises concerns about their reliability.
  2. To boost both reproducibility and innovation, it's important to encourage sharing of failed experiments and null results. This would help scientists take risks and avoid only publishing positive outcomes.
  3. Creating 'Red Teams' in science can challenge current beliefs and assumptions. These groups would actively work to test and potentially disprove existing theories, fostering better scientific inquiry.