The Metascience Observatory @metascienceobs.bsky.social · 22/09/2026Seems like a solvable problem! Surprised someone hasn't already written a script and a wikibot to solve this issue. 001
The Metascience Observatory @metascienceobs.bsky.social · 22/09/2026it depends if you are on a university/institutional network/VPN where you have access. On the Harvard VPN, which Metascience Observatory founder Dan Elton has access to, we find we can download 70-85% of PDFs. Without that, it drops to 50-70%. Get rate is higher in biomedical where OA is more common 111
The Metascience Observatory @metascienceobs.bsky.social · 10/09/2026#6 Junjie Yu, Youyi Lu, et al. "miR-200b suppresses cell proliferation, migration and enhances chemosensitivity in prostate cancer by regulating Bmi-1" Oncology Reports (2014) 10.3892/or.2013.2897 pubmed: 24317363 submitted here: pubpeer.com/publications... 010
The Metascience Observatory @metascienceobs.bsky.social · 10/09/2026#5 cross-paper image reuse, submitted to pubpeer. The 2014 paper caption says the image shows "EGFP expression in pAcISUBC transfected BuMECs" pAcISUBC is a human insulin expression vector constructed in the 2014 paper that does not appear in 2013 paper. 10.1007/s11033-014-3464-3 PMID: 24969480 110
The Metascience Observatory @metascienceobs.bsky.social · 09/09/2026#4 International Journal of Molecular Sciences 13(7):8514-8539 (2012). 10.3390/ijms13078514 There is a duplicated region across cells that received different treatments. The fluorescence patterns are almost identical, although some spots on the 2nd row do not appear in the first row. 🧐🤔 110
The Metascience Observatory @metascienceobs.bsky.social · 09/09/2026#3 "Aberrant expression and constitutive activation of STAT3 in cervical carcinogenesis: implications in high-risk human papillomavirus infection" Molecular Cancer (2010) 10.1186/1476-4598-9-282 There is a duplicated region across two different staining conditions. pubpeer.com/publications... 110
The Metascience Observatory @metascienceobs.bsky.social · 09/09/2026#2 "Permissive role for mGlu1 metabotropic glutamate receptors in excitotoxic retinal degeneration" Neuroscience (2017) 10.1016/j.neuroscience.2017.09.005 There is a duplication across figures that correspond to 2 different mouse strains - C57BL/6J and BALB/c. (All these submitted to PubPeer) 110
The Metascience Observatory @metascienceobs.bsky.social · 09/09/2026THREAD OF IMAGE DUPLICATION FINDINGS (will be continually updated) #1 Nature Medicine 18, 892–901 (2012). 10.1038/nm.2772 The cells in Fig. 1 are described as from a human patient, while the matching cells in Fig. 6 are described as from a tumor grown subcutaneously in a mouse from human cells. 100
The Metascience Observatory @metascienceobs.bsky.social · 09/09/2026.. than services like reviewer3 or refine.ink which charge $30 or more per paper, and we think our system is more thorough than current commercial products when it comes to checking statistics & arithmetic. tagging @jtimmer.bsky.social on this thread along with image sleuth @mumumouse2.bsky.social 000
The Metascience Observatory @metascienceobs.bsky.social · 09/09/2026We think there would be a lot of value to the scientific community to have a free "harness" they can use w/ a Claude Code, Codex, or Grok Build plan. The full audit only costs about $0.25-$1 per paper depending on which subscription you use and the length/complexity of paper. That is cheaper.... 110
The Metascience Observatory @metascienceobs.bsky.social · 09/09/2026For more information on our Forensic Metascience Agent, visit metascienceobservatory.org/forensic-met... We are currently validating the system and hope to show that it adds value on top of naive prompts to ChatGPT.metascienceobservatory.orgThe Metascience Observatory | Analyzing Scientific ReproducibilityUsing AI to analyze every scientific paper to assess reproducibility, rigor, and fraud across all fields of science. 110
The Metascience Observatory @metascienceobs.bsky.social · 09/09/20260.5 % is well below @elisabethbik.bsky.social's 2016 finding of 3.8% in biomedical papers (journals.asm.org/doi/10.1128/...). Most likely we are not catching a lot. I'll be iterating on the system to try to improve it. We plan to eventually open source it so people can use it themselves! (3/3) 152
The Metascience Observatory @metascienceobs.bsky.social · 09/09/2026Metascience Observatory Founder @delton.bsky.social carefully went through the findings along with biology/neuroscience expert Greg Fitzgerald, PhD. We submitted 4 PubPeer comments so far and will post a 5th soon. It looks like final count will be 5/1000 or 0.5%. We'll now iterate on system.. (2/3) 150
The Metascience Observatory @metascienceobs.bsky.social · 09/09/2026We ran our simple image duplication detection system on 1,000 papers. The papers 1. have 1 author who has a image-related retraction according to @retractionwatch.com, 2. are not retracted, and 3. have no image-related @pubpeer.com comments. We use AI to screen findings & suppress false pos.. (1/3) 132
The Metascience Observatory @metascienceobs.bsky.social · 02/09/2026What do people think? Should scientists who release old raw data (ie from papers more than 2 years old) get preferential treatment by funders? Releasing datasets is especially valuable in the age of AI - agentic AI systems can be sleuths at scale, flagging things for more detailed human review. 010
The Metascience Observatory @metascienceobs.bsky.social · 01/09/2026Great news! You can now pip install our fetchPDF library! pip install fetchpdf it's the world's best command-line PDF-getter, using completely legal channels! 121
The Metascience Observatory @metascienceobs.bsky.social · 01/09/2026Another Dan Ariely paper contains tampered (fraudulent) data, and it looks like Ariely was responsible (the file says "last saved by Ariely" and the only other author, Wertenbroch, claims he never had access to the dataset) datacolada.org/138datacolada.org[138] Artificial Deadlines (Part 1): Evidence of Fraud in an Influential Study About ProcrastinationA new paper in Psychological Science (.htm) reports a failure to replicate Study 2 of Ariely and Wertenbroch’s influential article entitled, “Procrastination, Deadlines, and Performance: Self-Contr… 011
The Metascience Observatory @metascienceobs.bsky.social · 28/08/2026AI improved a lot last year. We've run on 40 papers with Opus 5 & so far as we've been reviewing the results we have found zero errors. There are definitely things it misses though -- we want it to understand every single number and sanity check it if possible. Would cost a fortune to have human do. 000
The Metascience Observatory @metascienceobs.bsky.social · 28/08/2026We currently use Opus 5 or Grok 4.6 which have extremely low false positive rates. Unfortunately many critics have not updated on how good AI has gotten in the last year. Much is tractable now that was not very tractable just a year ago. We are focusing on validation to help convince AI skeptics. 000
The Metascience Observatory @metascienceobs.bsky.social · 27/08/2026Three different AIs were asked to predict how AI might impact science. This was one of the scenarios they arrived at - "The Slop Singularity". (ai-will-do-science.surge.sh) We don't think it will get quite so bad, but something along these lines is happening. AI will help in the end. 000
The Metascience Observatory @metascienceobs.bsky.social · 27/08/2026AI for metascience data extraction, AI for peer review, AI for forensic auditing, AI for rigor-checking, AI for sleuthing, AI for replication, AI for lit review..... today's AI is making so much possible now! Apply to attend this online conference for free here: forrt.org/ai-metascien...forrt.orgFORRT AI in Metascience Online Conference | FORRTRegister for the FORRT AI in Metascience Online Conference, 28–30 September 2026. Online and free to attend. 023
The Metascience Observatory @metascienceobs.bsky.social · 27/08/2026AI for metascience is booming, and @lukaswallrich.bsky.social and the huge volunteer team at the superb @forrt.bsky.social project are hosting this very timely online conference. Metascience Observatory Founder @delton.bsky.social will be speaking Sept 28-30 along w/ 33 other speakers. Link 👇 121
The Metascience Observatory @metascienceobs.bsky.social · 24/08/2026metascienceobservatory.org/forensic-met...metascienceobservatory.orgThe Metascience Observatory | Analyzing Scientific ReproducibilityUsing AI to analyze every scientific paper to assess reproducibility, rigor, and fraud across all fields of science. 000
The Metascience Observatory @metascienceobs.bsky.social · 24/08/2026The fraud problem in science, summarized. This is part of our motivation for building the Forensic Metascience Agent, a AI agent orchestration system designed to enable a thorough audit of a scientific paper, including images, tables/data, statistics, and supplementary files. (🔗 w/ more info below) 511
The Metascience Observatory @metascienceobs.bsky.social · 24/08/2026We believe PubPeer can re-invigorate scientific self-criticism. “PubPeer exists & thrives precisely because of a frustration, shared by its initiators and its users, at the failures of traditional scientific institutions to administer well-organised scepticism.”-Wytske et al. osf.io/preprints/me... 142
The Metascience Observatory @metascienceobs.bsky.social · 06/08/2026The NIH recently released a RFI seeking public input on new ways to measure and reward scientific impact. The agency is evaluating whether traditional metrics such as publication counts and citation indexes adequately capture the value of biomedical research. => grants.nih.gov/grants/guide...grants.nih.govNOT-OD-26-087: Request for Information on Measuring and Rewarding Scientific ImpactNIH Funding Opportunities and Notices in the NIH Guide for Grants and Contracts: Request for Information on Measuring and Rewarding Scientific Impact NOT-OD-26-087. NIH 011
The Metascience Observatory @metascienceobs.bsky.social · 05/08/2026Yesterday we published a detailed report about what we've learned so far from our replication database about reproducibility and correlates of reproducibility. => open.substack.com/pub/moreisdi... 021
Reposted by The Metascience ObservatoryLukas Röseler @aufdroeseler.bsky.social · 10/10/2025ReplicationResearch.org is now open for submissions! Submit replications and reproductions from many different fields, as well as conceptual contributions. With diamond OA, open and citable peer review reports, and reproducibility checks, we push the boundaries of open and fair publishing. 39354
Reposted by The Metascience ObservatoryDan Elton @delton.bsky.social · 20/06/2026I've created a new @metascienceobs.bsky.social "Bird's Eye Review" dashboard for human clinical trials on antiviral nasal sprays. Check it out ➡️ metascienceobservatory.org/birds-eye-re... Let me know what you think! There's also a #LongCOVID dashboard too: metascienceobservatory.org/birds-eye-re... 132
The Metascience Observatory @metascienceobs.bsky.social · 29/01/2026Read our first newsletter on Substack: metascienceobservatory.substack.com/p/metascienc... You can subscribe on Substack or on our conventional mailing list: metascienceobservatory.eo.page/kdntvmetascienceobservatory.substack.comMetascience Observatory Newsletter #1Happy New Year from the Metascience Observatory! This newsletter will go out either bimonthly or quarterly. 031
