Living research benchmark
Paper grades
Two complementary views of eligible research papers: how well the manuscript is executed and how strong, consequential, and testable its central concept may be.
Literature debate and conflict-discussion papers are excluded from grading, paper counts, comparisons, and trend lines.
Live publication feedConnecting…
Zenodo paper citations
Loading daily citation statistics…
DataCite-indexed incoming citations, deduplicated across each paper’s versions. A work citing two paper records contributes once to each record’s count. Separate deposits remain separate records, including matching titles. Citation coverage is incomplete: zero means no indexed citations were found, and counts may differ from Zenodo or Google Scholar.
Daily observations begin October 9, 2026. Missing collection dates remain gaps; newly tracked records contribute from their first observation. Citation corrections can lower totals.
Citations and citing works for each paper
| Paper record | Indexed citations | Citing works |
|---|
Daily citation totals
| Collection date | Paper records | Indexed citations |
|---|
Download per-paper daily citation history (JSON) · About the citation source
Readership over time
Zenodo views & downloads
Loading daily Zenodo statistics…
ViewsDownloads
Each point sums Zenodo's displayed, cumulative all-version counters across the listed papers on that collection date. New papers can increase the totals. Downloads can include supplementary files. Open circles indicate a partial inventory. Lines connect verified observations across unavailable dates; no missing daily values are estimated. Unavailable dates are labeled on the chart and in Daily totals. Tracking begins October 9, 2026. This section follows all tracked papers independently of the grade filters.
Views and downloads for each paper
| Paper | Views | Downloads |
|---|
Daily totals
| Collection date | Papers | Views | Downloads | Inventory coverage |
|---|
Researcher publications
UMDF / POLG researchers
Researcher publications
UMDF / POLG researchers
This category follows publications by researchers with documented support from UMDF or the POLG Foundation. It includes the stated bibliography scope across each researcher's career. Researcher support does not establish that a grant funded an individual paper. Papers retain their original benchmark group and grades.
Each paper requires grades from ChatGPT, Claude and Gemini. Each reviewer grades manuscript quality and concept impact; the plot uses the two averages across reviewers. A paper counts as complete only when all three reviewers have graded both measures.
Loading researcher coverage…
Researcher roster and funding sources
Publication dates
All dates included. Filters the table and every plot.
Previous paper
Manuscript quality
Editorial execution, claim-evidence alignment, methodology, reproducibility, reporting integrity, and communication. This score asks whether the paper is ready and defensible as a scientific artifact.
New paper
Concept impact
Concept integrity, potential consequence if confirmed, and validation readiness. This score prioritizes ideas for testing; it does not estimate truth, clinical efficacy, or future awards.
Concept progress over time
Mito Map concept impact by week
Show or hide papers by supported researchers in this plot. Papers linked to both foundations appear while either is checked. Other papers stay visible. Researcher support does not establish funding for an individual paper.
This toggle changes the cohort's horizontal comparison line. Average uses all available cohort concept scores in the date range. Top 5 uses the five papers with the highest average of quality and concept scores. The weekly trend stays the same.
Average uses all available journal concept scores in the date range. Top 5 uses the five high-quality journal papers with the highest average of manuscript quality and concept impact, requiring both grades. Only the journal comparison line changes. Fewer papers are used when fewer qualify.
Each point is the mean concept-impact score for papers published that week; labels show the number of graded papers. Dashed lines show each series' cumulative mean. General Mito Map publications and POLG/TWNK focus-generated publications have separate series; focused papers are light green. Comparison lines remain descriptive because cohorts and reviewer counts differ.
How concept impact is calculated
40% Concept integrity40% Potential consequence20% Validation readiness
The three axes are combined as a weighted geometric score, so a severe weakness cannot be fully offset by strength elsewhere. Hard caps address non-falsifiability, absent discriminating predictions, central evidence mismatch, decisive contradictions, unsupported clinical-impact claims, unbounded generalization, and pure relabeling. Historical records generally use the frozen v1.2 rubric with three model reviews; some newly published records have fewer reviewers, so the live trend is descriptive rather than a formal estimate of improvement.
Progress over time
Mito Map manuscript quality by week
Show or hide papers by supported researchers in this plot. Papers linked to both foundations appear while either is checked. Other papers stay visible. Researcher support does not establish funding for an individual paper.
This toggle changes the cohort's horizontal comparison line. Average uses all available cohort quality scores in the date range. Top 5 uses the five papers with the highest average of quality and concept scores. The weekly trend stays the same.
Average uses all available journal quality scores in the date range. Top 5 uses the five high-quality journal papers with the highest average of manuscript quality and concept impact, requiring both grades. Only the journal comparison line changes. Fewer papers are used when fewer qualify.
Each point is the mean manuscript-quality score for papers published that week; labels show the number of graded papers. Dashed lines show each series' cumulative mean. General Mito Map publications and POLG/TWNK focus-generated publications have separate series; focused papers are light green. Horizontal lines are descriptive means of graded members; pending papers contribute no score.
How manuscript quality is calculated
30% Claim–evidence alignment25% Methodological validity20% Reproducibility and transparency15% Reporting integrity10% Communication clarity
The intended v3 score is a weighted 0–100 assessment of scientific readiness. Unsupported central claims or invalid central analyses cap it at 59; untraceable central data or undisclosed central methods cap it at 69; a major unresolved integrity mismatch caps it at 74. Earlier benchmark records use this frozen rubric and three model reviews. Some new live publications currently inherit their artifact-bound eight-domain publication-QA score and should be treated as provisional until they receive the same v3 review protocol.
Mito Map through time
Quality and concept profile
Show or hide papers by supported researchers in this plot. Papers linked to both foundations appear while either is checked. Other papers stay visible. Researcher support does not establish funding for an individual paper.
Select or unselect the categories above to show or hide their points, reference lines and trails in this plot. All categories are selected by default.
Average shows a random sample of up to 20 papers from the current review cohort, plus the separately selected POLG additions, all with both grades; its dashed means use all available cohort scores in the date range. The sample stays stable during this visit. Top 5 shows only the five papers with the highest average of manuscript quality and concept impact, and both dashed means use those same papers. Fewer papers appear when fewer have both grades. Other selected categories stay visible.
Average includes all high-quality journal papers and their available grades in the date range. Top 5 shows the five journal papers with the highest average of manuscript quality and concept impact, requiring both grades; both pink dashed means use those same papers. Fewer papers appear when fewer qualify. The UMDF / POLG control remains independent; papers in both cohorts must satisfy both selections.
Loading Mito Map paper history…
Selected non-Mito Map benchmark points stay visible throughout playback. Mito Map papers animate oldest first when this plot is in view; playback pauses when you scroll away. Earlier points stay visible. The light green POLG/TWNK points identify papers produced by the specialized focus-generation process and remain separate from the original general Mito Map category. Older papers are classified by recorded publication focus, not by gene names in their titles. The pale mint trail connects recent general Mito Map papers only; the light green trail connects recent POLG/TWNK papers only, so no connection crosses between generation processes. The larger point marks the latest Mito Map paper. Papers awaiting concept review are named in the history but are not yet plotted; concept-only papers appear as diamonds. Missing grades break the trail. Undated papers appear last. Black dashed lines show the Nobel cohort means; pink dashed lines show the high-quality journal cohort means for manuscript quality and concept impact. Dark yellow points show the displayed sample or top five from the current 100 selected mitochondria-related publications by UMDF / POLG supported researchers, together with the separately reviewed POLG additions. This review cohort is separate from Mito Map. Pending papers stay in the table until scores are available. Matching dark yellow dashed horizontal and vertical lines show its concept and quality means for all date-visible available scores in Average mode or the same five displayed papers in Top 5 mode. Existing grades remain visible while the six-review update is pending, and papers without scores stay in the table. Completed reviews update automatically. Hover or focus a mark for its title and scores.
Complete record
Paper-level grades
Page 1 of 1
| ID | Published | Paper | Group | Quality | Concept | Benchmark status |
|---|---|---|---|---|---|---|
| Loading live grades… | ||||||
How to read these grades
Three additional POLG-supported researcher papers were selected separately and reviewed with the same six-assessment protocol. They extend the plotted comparison without changing the original fixed sample. The original review selection contains the 50 newest mitochondria-related verified UMDF / POLG supported-researcher publications and 50 randomly sampled older papers. With owner authorization, unresolved full-text holds may be replaced by other eligible papers with verified complete public manuscripts. The original selection and each substitution are retained in the queue audit. Researcher support does not establish direct grant funding for each paper. It obtains independent Gemini, ChatGPT/Codex and Claude assessments of both manuscript quality and concept impact. A paper is marked complete only after all six reviews pass validation. Each completed grade is the mean of its three reviewer scores under the fixed rubric; expand the review count in the table to see those scores. Missing public full text remains on hold until recovered or replaced by an audited eligible manuscript. These model assessments are provisional, and normalized-text review does not verify PDF layout or reproduce experiments.
Scores are descriptive measurements under fixed candidate rubrics. Cohorts differ by field, publication era, study type, and validation stage. Model reviewers may share training data and recognizable papers may not be fully blinded. Sensitivity ranges quantify dependence on scoring choices, not the probability that a concept is true.
Lower-ranked journal controls use the fixed 2024 classification: Q4 in at least one relevant subject category. The five additional mitochondrial primary studies were chosen before grading from Physical Activity and Nutrition and its former title, Journal of Exercise Nutrition & Biochemistry. Original journal titles and publication dates are retained; the 2024 classification applies to that continuing journal lineage, rather than establishing its quartile in each paper’s publication year. Both scores for these additions average independent ChatGPT, Claude and Gemini reviews of complete public text.
The additional Nobel-linked and high-quality-journal reference papers use two manuscript-quality reviews and one structured concept-impact review per paper; reviewer counts and source hashes are retained in the data. They were included only when complete machine-readable PMC full text passed the eligibility check. Nobel linkage identifies research related to the awarded discovery; it does not mean that an individual paper won a prize. These expanded cohorts are purposive and unmatched, so their means remain descriptive.
UMDF / POLG researcher membership is a separate author-support category and can overlap the existing groups. DOI, PMID and PMCID matches connect publications to existing grades without adding duplicate papers. Bibliographic or abstract metadata alone never supplies a grade. Category means use only unique graded members; the roster, author matches, support sources and pending full-text coverage remain visible for review.
Newly published Mito Map papers appear automatically only after human publication approval and a passing artifact-bound professionalism gate. Before publication, the approved final PDF receives both the manuscript-quality grade and the separate concept-impact grade; the concept review cannot use the manuscript-quality score as an input. The stored PDF hash binds both grades to the approved artifact, and publication is blocked if either grade is absent or stale.