The reopening watchlist read for the first time: 0 of 10 candidate papers meet their condition, 8 of 10 watched decisions returned nothing, and not one of the ten titles names the gene it was retrieved for
also called watchlist adjudication · reopening conditions read · 0 of 10 reopen · the watcher has no reader
A script watches for new papers that would overturn decisions this project has made, and nobody had ever read what it found. Reading it: none of the ten papers it flagged actually bears on the decision it was flagged for, and none of them even mentions the gene in its title.
The same thing, for a scientist
`scripts/watch-reopening-conditions.py` exists because a cycle-23 audit found that four of seven struck target candidates rested on reasoning that fails on reading, and concluded: *"those reopening conditions have no owner."* The script is the owner for retrieval. **It has never had an owner for ADJUDICATION** - nothing reads what it returns. This cycle it returned a queue, and this is the reading. **The queue is smaller than the headline.** The script prints *"10 ACTIONABLE candidate paper(s) across 10 decisions"*, and **eight of the ten decisions returned zero papers**. The real queue is **7 papers for MEIS1 and 3 for KDM6A**. Every other watched decision - PRMT1, RBBP5, ASH2L, CTR9, CDC73, CCNT1, ELL2, MEF2C - returned nothing new since 2026-01-01. **Verdict: 0 of 10 reopen anything. Both decisions stand.** **MEIS1 is demoted, and its condition is specific:** *a demonstration that lowering MEIS1 PROTEIN is selectively lethal in KMT2A-rearranged cells, independently of HOXA9 programme readout*. The seven papers are **menin-inhibitor reviews and commentary** - *Targeting the Menin-KMT2A Axis in AML*, *Menin inhibitors as a treatment for acute leukemia: from bench to the clinic*, *Clinical Integration of Menin Inhibitors in AML*, and four more of the same shape, including one about **NPM1-mutated** AML. None reports a protein-level MEIS1 intervention. The demotion rests on a CRISPR result being the wrong instrument for a factor whose abundance is set post-translationally, and a review cannot touch that. **KDM6A is demoted, and its condition has two limbs:** a non-catalytic modality showing a KMT2A-rearranged-selective effect, or the DIRECTION of effect resolved within KMT2A-rearranged disease. The three papers are a menin-inhibitor commentary, a **mesothelin expression** survey, and a paper on **autoimmune cytopenias** and histone lysine *methyltransferase* mutations - KMT, not KDM. Neither limb is touched. **AND THE PRECISION PROBLEM IS THE FINDING WORTH KEEPING: not one of the ten titles contains the gene it was retrieved for.** Zero of seven name MEIS1; zero of three name KDM6A. The queries are broad enough to return the ambient menin-inhibitor review literature, which grows continuously, so this queue will refill every cycle with papers that cannot in principle satisfy the conditions. A watchlist that cries wolf monthly is one nobody reads - which is precisely the inattention failure the cycle-23 audit built it to prevent.
| script | scripts/watch-reopening-conditions.py |
|---|---|
| report | docs/watch-report.json |
| status | run and READ |
| run date | 2026-09-22 |
| window | papers published since 2026-01-01 |
| why this ran | the script's own docstring quotes the cycle-23 audit saying the reopening conditions 'have no owner'. The script owns retrieval; nothing owned adjudication until now. |
| THE PRECISION PROBLEM | NOT ONE of the ten titles contains the gene it was retrieved for - zero of seven name MEIS1, zero of three name KDM6A. The queries return the ambient menin-inhibitor review literature, which grows continuously, so this queue will refill every cycle with papers that cannot in principle satisfy the conditions. A watchlist that cries wolf monthly is one nobody reads, which is the exact inattention failure it was built to prevent. |
| absence class | measured-not-analysed |
scope — the numbers behind the chart
{
"decisions_watched": 10,
"decisions_returning_zero_new_papers": 8,
"candidate_papers": 10,
"papers_meeting_their_condition": 0
}decisions that returned nothing
[ "PRMT1", "RBBP5", "ASH2L", "CTR9", "CDC73", "CCNT1", "ELL2", "MEF2C" ]
MEIS1
{
"disposition": "demoted - UNCHANGED",
"condition": "a demonstration that lowering MEIS1 PROTEIN is selectively lethal in KMT2A-rearranged cells, independently of HOXA9 programme readout",
"papers": 7,
"verdict": "all seven are menin-inhibitor reviews or commentary and none reports a protein-level MEIS1 intervention; one is about NPM1-mutated AML. The demotion rests on a CRISPR knockout being the wrong instrument for a factor whose abundance is post-translationally set, and no review can touch that."
}KDM6A
{
"disposition": "demoted - UNCHANGED",
"condition": "either a non-catalytic modality with a KMT2A-rearranged-selective effect, or the direction of effect resolved within KMT2A-rearranged disease specifically",
"papers": 3,
"verdict": "a menin-inhibitor commentary, a mesothelin expression survey, and a paper on autoimmune cytopenias and histone lysine METHYLTRANSFERASE mutations - KMT, not KDM. Neither limb of the condition is touched."
}- This is a decision-hygiene audit and nominates nothing clinically. Its only bearing on an intervention is conservative: MEIS1 and KDM6A remain demoted as therapeutic candidates for KMT2A-rearranged disease, and nothing published since January changes that. A title-level pass is not a literature review, and neither demotion should be quoted as settled on the strength of this entry.
- Should the MEIS1 and KDM6A queries be narrowed to require the gene symbol in the title or abstract? As written they return the ambient menin-inhibitor review literature, and none of the ten retrieved titles names its gene. The risk of narrowing is that a paper reporting a MEIS1 protein-level result under a broader title would be missed, which is the failure the current breadth is buying.
- Who reads this next cycle? The script has run before and its output had never been adjudicated until now, which is the same inattention failure at one remove - the cycle-23 audit gave the conditions an owner for retrieval and not for reading.
- Do the eight silent decisions deserve the same scrutiny as the two noisy ones? PRMT1, RBBP5, ASH2L, CTR9, CDC73, CCNT1, ELL2 and MEF2C returned nothing, which could mean the literature is quiet or could mean their queries are too narrow - the opposite failure to MEIS1's. Nobody has checked which.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
scripts/watch-reopening-conditions.py, run 2026-09-22 over papers published since 2026-01-01, watched 10 struck, demoted or redefined target decisions, each carrying a defined reopening condition and a PubMed query. Eight returned no new papers: PRMT1, RBBP5, ASH2L, CTR9, CDC73, CCNT1, ELL2 and MEF2C. MEIS1 returned 7 and KDM6A 3. All ten were read at title level and none satisfies the condition it was retrieved for: the MEIS1 set consists of menin-inhibitor reviews and commentary with no protein-level MEIS1 intervention, and the KDM6A set is a menin-inhibitor commentary, a mesothelin expression survey and a paper on histone lysine methyltransferase mutations in autoimmune cytopenias. No title among the ten contains the gene symbol it was retrieved for.
What this evidence cannot show
THIS IS A TITLE-LEVEL ADJUDICATION AND NOTHING MORE. No abstract was read and no full text was opened, so a paper whose title is a review heading could in principle contain a primary protein-level MEIS1 result - reviews occasionally report new data, and a title cannot exclude it. The correct reading is that none of the ten is PROMISING enough to open, not that none of them could possibly qualify. The window is also fixed at 2026-01-01 by the script, so anything earlier is out of scope by construction.
Evidence that argues against this (2)
- A reader with a stake in reopening MEIS1 would say a title-level pass is exactly how a real result gets missed, and they would be right that the asymmetry favours the status quo: it costs one line to dismiss a paper and an hour to read one. The defence is that the demotion is recorded with its reason and its reopening condition rather than as a verdict, so the cost of being wrong here is a delay, not a permanent strike - but that defence is only as good as the next person who runs this script.
- Finding zero reopenings is the outcome that most flatters the existing decisions, and the same coordinator made several of them. An independent reading would be worth more than this one, and the ten PMIDs are listed explicitly in the citation so that it can be done without rerunning anything.
adjudicates the reopening conditions this register defines; 0 of 10 candidate papers meet theirs, and both demotions stand
demotion unchanged: all seven watchlist papers are menin-inhibitor reviews, none a protein-level MEIS1 intervention
demotion unchanged: none of three papers touches either limb of its reopening condition