TOMORROW'S PUBLISHER — FORECAST REGISTER AND CLAIM AUDIT 27 October 2026 | Ivan Massow Publishers must share the cost of catching fake research All five estimates below are authenticated final Noah automatic quick forecasts. No standalone forecast_this jobs, replacement targets or host-selected forecast horizons were commissioned. The probabilities are modelled event estimates, not observed prevalence, detector accuracy, or adoption shares. Returned uncertainty intervals are model ranges, not a guaranteed interval or a sample confidence interval. READER SELECTION The article uses mass_fraud_retraction, false_flag_reversal and screening_roi_evidence in their relevant sections. The two hub_live_adopter targets are retained here but omitted from the narrative because their exact-service first-use baseline is unresolved and their restricted roster is not an informative general adoption benchmark. RUP already describes using simultaneous-submission checks. Royal Society's existing Hub use includes this type of shared information; the exact production baseline by service at every named publisher is not established. The omission is about target eligibility and business relevance, not low probabilities. No omitted probability was replaced or recast as industry stagnation. PROGRAMME nabdl_programme_run_492dba3a00bc55492ad405496cf84e645221358972b5b143 Trend job: nabdl_job_309e1643fc565d01e7a49ed80ae1a7ede0b5f644e27761fefe584daaf2de5413 Automatic discovery/forecasts job: nabdl_job_fd95e237812ce27c2260a550fce8e0195e042c77743c9c3593c21a4f98a52713 Research note: Noah completed all 36 planned research checks across 27 distinct lines of enquiry; the detailed method and source trail are retained in the Claim audit. FULL ORIGINAL REGISTER 1. By 2027-03-31, will at least one of Royal Society Publishing, Taylor & Francis or Rockefeller University Press begin production use, after 2026-10-27, of STM Integrity Hub cross-publisher duplicate-submission matching to screen live manuscript submissions? Origin: Noah automatic quick forecast, authenticated final v2 reassessment. YES: 19%; exact binary NO complement: 81%. Returned YES uncertainty interval: [6,38]%. Resolution date: 2027-03-31 YES condition: By 2027-03-31, at least one named publisher that had not previously used the matching service in production has begun screening live submissions with STM Integrity Hub cross-publisher duplicate-submission matching. NO condition: By 2027-03-31, none of the named publishers has newly begun production screening of live submissions with STM Integrity Hub cross-publisher duplicate-submission matching. Event/publication distinction: The event is production screening of live submissions; dated STM Hub or publisher implementation records verify deployment but are not themselves the deployment. Run reference: a280be0b20a1c2eb217a187a0a991fe1155ea02fc7a3e6d7979587fdd0e898b9 Target digest: dd80ca51e8d3219689ca290607a71bbc307461c250279a88c072f1049d129b0e Final receipt: 2c88ccf524689d451c25e68abaf6e6aae5fdd8694cd5da9d7c38e3320fcd983c Retained native receipt: 793913681903f5e95998dfb3d98333c35e1ca287c88cc860359dc73ca4887d39 Status: ready; estimator executed: true 2. By 2027-10-27, will at least one of Royal Society Publishing, Taylor & Francis or Rockefeller University Press begin production use, after 2026-10-27, of STM Integrity Hub cross-publisher duplicate-submission matching to screen live manuscript submissions? Origin: Noah automatic quick forecast, authenticated final v2 reassessment. YES: 30%; exact binary NO complement: 70%. Returned YES uncertainty interval: [10,55]%. Resolution date: 2027-10-27 YES condition: By 2027-10-27, at least one named publisher that had not previously used the matching service in production has begun screening live submissions with STM Integrity Hub cross-publisher duplicate-submission matching. NO condition: By 2027-10-27, none of the named publishers has newly begun production screening of live submissions with STM Integrity Hub cross-publisher duplicate-submission matching. Event/publication distinction: The event is production screening of live submissions; dated STM Hub or publisher implementation records verify deployment but are not themselves the deployment. Run reference: 61358b6b72c0e5506a8d6f622205e694ba6ea59ad5ad8a678c5a60c432999fd7 Target digest: 8ff8ba701c5a968d497b43a1c298e42e5ea0b9d97916c2e2941704154cb16b6c Final receipt: d86534c5f1219d5bd0a66266bf9ba0bf10af6689356342fe36d25a1eb9b6da84 Retained native receipt: 1ae7709af188e98f5119d75e52b8a08efbe6a6ff30362d8d39afb336de4c2ce9 Status: ready; estimator executed: true 3. By 2027-10-27, will STM Integrity Hub, Royal Society Publishing, Taylor & Francis or Rockefeller University Press publish a quantitative controlled or before-and-after evaluation finding that pre-publication integrity screening reduced either total editorial labor or cost per submission after investigations and appeals, or the rate of screened papers later retracted for fraud, with cohort definitions and denominators? Origin: Noah automatic quick forecast, authenticated final v2 reassessment. YES: 22%; exact binary NO complement: 78%. Returned YES uncertainty interval: [10,38]%. Resolution date: 2027-10-27 YES condition: By 2027-10-27, a named organization has published a qualifying quantitative comparison with defined cohorts and denominators showing lower all-in editorial labor or cost, or fewer later fraud-related retractions, for screened submissions. NO condition: By 2027-10-27, no named organization has published a qualifying quantitative comparison showing either specified outcome. Event/publication distinction: The forecast is the publication of decision-useful evidence; underlying cost or fraud reduction could occur without satisfying this transparency event. Run reference: 81b7d947b0274d2ba206808cdd4cdfa52628697c59ec88f183f77751837a59f0 Target digest: 76cb9564ce0ac69b9e29c375c2758458774fbf786200c9360309b6e9c545df4d Final receipt: 8af6583dfe082bd1609ac0a6f4370bde9f2311dc63a4516cf1cea4f0bbb12a52 Retained native receipt: 3338e5281d3dec632e96f56503cc7bb4d6348c0b78ed86b94fbb722ea38e0bf4 Status: ready; estimator executed: true 4. By 2027-10-27, will a UK or European scholarly publisher publicly document a case in which an automated pre-publication integrity flag materially contributed to holding or rejecting a manuscript and the publisher later reversed that decision after determining through human review that the flag was materially erroneous? Origin: Noah automatic quick forecast, authenticated final v2 reassessment. YES: 26%; exact binary NO complement: 74%. Returned YES uncertainty interval: [10,48]%. Resolution date: 2027-10-27 YES condition: By 2027-10-27, a qualifying publisher has publicly documented an erroneous automated integrity flag, its material role in an initial hold or rejection, and a reversal following human review. NO condition: By 2027-10-27, no UK or European scholarly publisher has publicly documented all elements of such a false-positive reversal. Event/publication distinction: The forecast concerns public documentation of a reversal, because disclosure is the observable reputational and governance signal; undisclosed reversals do not satisfy it. Run reference: d5d635ee04a2c88a6496cbe89cbbe1119326671681c4a1552bbbed9c06e7be35 Target digest: fab9f4060241dc1e75c154aac2a0b82d6856dbc0978814ec10e10883dab61214 Final receipt: 6df03da61e103999e9d5587585f9042dad8c3037959637f239ce4585149d701f Retained native receipt: 0bcfb780741cfee6162692c6ffb1f7247b09fb2c850bf5c9a97049ba8915b528 Status: ready; estimator executed: true 5. By 2027-10-27, will a UK- or Europe-headquartered scholarly publisher carry out a coordinated retraction of at least 100 articles attributed principally to paper-mill activity, fabricated peer review, or manipulated images or data? Origin: Noah automatic quick forecast, authenticated final v2 reassessment. YES: 62%; exact binary NO complement: 38%. Returned YES uncertainty interval: [40,78]%. Resolution date: 2027-10-27 YES condition: By 2027-10-27, one qualifying publisher has retracted at least 100 articles in a coordinated action principally attributed to one or more specified research-integrity failures. NO condition: By 2027-10-27, no qualifying publisher has carried out a coordinated retraction meeting the 100-article threshold and stated grounds. Event/publication distinction: The event is the retraction of the articles; publisher notices, Crossref records and Retraction Watch reporting provide evidence that the coordinated action occurred. Run reference: 580cfae0471e92828931fc8b37758c303e3cf51f1d06e5e45deffc3b840f3b2d Target digest: 9b6d8e1a806aa1ce06e8cdf23bd39b8eb78ffe3aa9d13ccfa8e9bdc7d089aaf4 Final receipt: f1408e501b62f54501faebaa34a1fc3b265ed38851d12ed8e0387d43d5581b39 Retained native receipt: 8359b9f26f7071c7ff79aad46b6aaf70d995ebf9c3e0927d0d40dd417512a3ff Status: ready; estimator executed: true INTERPRETATION The19% and30% targets share the same after27October2026 first-production-use definition and exact named roster, with March and October2027 deadlines. They must not be treated as industry adoption probabilities or proof that shared checking is uncommon. Existing use cannot resolve a first-use-after-cutoff target. No exact crossing date or later probability was inferred. The22% target requires a favorable quantitative controlled or before-and-after evaluation with defined cohorts and denominators. For the work/cost branch, investigations and appeals count; for the retraction branch, the rate of subsequent fraud-related retractions among screened papers matters. The forecast concerns such evidence being published by one of the four named organisations, not the probability screening works. Its78% exact NO complement is absence of that qualifying publication by the same deadline, not absence of benefits. The26% target requires a publisher's public account linking an erroneous automated pre-publication flag to a material hold/rejection and a human-reviewed reversal. It is not a26% false-positive rate or a prediction that all or most authors will be treated wrongly. The74% exact NO complement is absence of a qualifying documented case, not proof errors never occur. The62% target concerns one coordinated action of at least100 articles by a UK/Europe-headquartered scholarly publisher on the specified grounds. An aggregate annual total, separate unrelated actions, recommended but unexecuted withdrawals, or a US-headquartered publisher alone does not satisfy it. The38% NO complement and[22,60]% complementary interval negate exactly that same event. No named publisher is accused of a future event. CLAIM CHECKS Royal Society: roughly9% concerns one journal's2025 suspected-paper-mill rejection share; not all journals or established fraud. Its2026 comment supplies no comparable denominator, so no2026 rate is inferred. RUP: five journals; local image observations and shared checks. No claim that all publishers have the same exposure. IEEE: case statuses and reporting periods preserved; these are allegations and resolutions, not a fraud census. Pilot, expansion and planned rollout remain distinct. Hub operator:125,000 monthly manuscript checks,40 publishers,7 submission systems in a June2026 conference abstract. Throughput is not a detection success rate. Taylor & Francis: automated screening plus specialist investigation. The May2023 tool announcement is historical, not a2026 launch. STM report: qualitative industry-commissioned interviews,18 experts/13 organisations; no sector-wide cost-saving or precision estimate. Illustrative two-journal example is explicitly hypothetical. Budgeting, author fairness and measurement recommendations are editorial judgements. RESEARCH LIMITS Retained matches on sports integrity, policing, fiction publishing and corporate transactions do not establish scholarly screening efficacy. Personal allegation stories were not used. Archive volume, country shares, sentiment and support/falsification lane labels are not causal findings or probabilities. Primary sources and the relevant IEEE source retrieved by Noah were checked directly. SOURCES CHECKED27OCTOBER2026 Investing in integrity — Royal Society Publishing Published: 2026-10-15 | Checked: 2026-10-27T05:03:42.392Z https://royalsociety.org/blog/2026/10/investing-in-integrity/ How AI Is Reshaping Scientific Publishing — Rockefeller University Published: 2026-10-26 | Checked: 2026-10-27T05:03:42.398Z https://www.rockefeller.edu/news/40794-how-ai-is-reshaping-scientific-publishing/ Misconduct: Taylor & Francis editorial policies — Taylor & Francis Published: undated or rolling page; reporting period specified in source | Checked: 2026-10-27T05:03:42.398Z https://authorservices.taylorandfrancis.com/editorial-policies/misconduct/ New report documents publisher investment in research integrity infrastructure — STM Published: 2026-01-13 | Checked: 2026-10-27T05:03:42.398Z https://stm-assoc.org/new-report-documents-publisher-investment-in-research-integrity-infrastructure/ STM Integrity Hub: infrastructural change through new collaborations — Joris van Rossum / FORCE2026 Published: 2026-06-05 | Checked: 2026-10-27T05:03:42.398Z https://ink.library.smu.edu.sg/force2026/Programme/agenda/54/ STM Integrity Hub — STM Published: undated or rolling page; reporting period specified in source | Checked: 2026-10-27T05:03:42.398Z https://stm-assoc.org/what-we-do/strategic-areas/research-integrity/integrity-hub/ Safeguarding Scholarly Communication: Publisher Practices to Uphold Research Integrity — Research Consulting / STM Published: undated or rolling page; reporting period specified in source | Checked: 2026-10-27T05:03:42.398Z https://s3.eu-west-2.amazonaws.com/stm.offloadmedia/wp-content/uploads/2026/01/07113200/STM-Research-Integrity-Storytelling_Final-Updated-01-May-2026-1.pdf Taylor & Francis Welcomes the Launch of STM’s Paper Mill Detection Tool — Taylor & Francis Published: 2023-05-16 | Checked: 2026-10-27T05:04:46.411Z https://newsroom.taylorandfrancisgroup.com/stm-paper-mill-detection-tool/ IEEE Publishing Ethics Team Upholds Research Integrity — IEEE Spectrum / IEEE Publishing Ethics Published: undated or rolling page; reporting period specified in source | Checked: 2026-10-27T05:21:17.375Z https://spectrum.ieee.org/ieee-publishing-ethics-research-integrity Research Integrity:2026half-year update — Springer Nature Published: undated or rolling page; reporting period specified in source | Checked: 2026-10-27T05:21:17.376Z https://www.springernature.com/gp/advancing-discovery/research-integrity POST-DRAFT DISCOVERY One bounded follow-up completed as job nabdl_job_b02a4a57fa9424aa01b9c7f5d8ae48efa796f28c6bd607a554ead00a5367fc4b. Six paired checks returned no admitted matched evidence for cohort cleanup shares, case-resolution movement or downstream status propagation; no rate or causal finding was invented. A retained contextual lead about AI and unreliable research was followed to the primary TRACES preprint and used as a scoped tangent. TRACES CHECK Rodionov and Assylbekov,11August2026,arXiv preprint.42selected papers,30models,10runs. Agent-style IFR-a0.93±0.004 treats engagement rather than refusal as failure, including engagement with recognition/warnings. This differs from its interactive measure and from live retrieval. No claim the93%applies to allAIscience or proves correctionmetadatawouldfixit. https://arxiv.org/abs/2608.11415