--- title: "No independent source yet corroborates the NYC school-nudge RCT's zero-rejection/98%-acceptance figure; the only other document carrying it is the same authors' own ACM EC 2026 peer-reviewed version" type: "claim" status: "seedling" source_url: "https://erica-chiang.github.io/assets/files/chiang-ec26.pdf" source_sha: "d86182414184d8f55e859fd169aa334388e46d823152999c6ed44cfbc94a6327" source_author: "Erica Chiang, Kenny Peng, Rebecca Lichtenstein, Brielle McDaniel, Kristen O'Neil, Dejaunique Thomas, Lianna Wright, Jon Kleinberg, Eva Tardos, Nikhil Garg" source_date: "2026-10-06T00:00:00.000Z" source_title: "Personalized Recommendations Without Inducing Congestion: Mitigating Disparities in the NYC High School Match" source_venue: "ACM Conference on Economics and Computation (EC) 2026, author-hosted camera-ready PDF β€” tls: verified" source_quote: "In a randomized controlled trial, we find that 16.4% of treatment applicants ranked a recommended program, versus 10.5% of control applicants who ranked a program they would have been recommended (57% relative increase; 𝑝=0.011); 5.6% of treatment applicants matched to such a program, versus 3.3% of control applicants (71% relative increase; 𝑝=0.071); further, no treatment applicant was rejected from a recommended program." source_quote_2: "Furthermore, no treatment applicants were rejected from a program they were nudged to,1 and only one control applicant was rejected from a program they would have been nudged to." source_quote_5: "No treatment students were rejected from a program they were nudged to (out of 25 effective applicants, i.e., not counting applicants who listed a nudged program but were accepted to another program they ranked higher), and 1 control student was rejected from a program they would have been nudged to (out of 28 effective applicants), for an overall 98% acceptance rate." source_url_5: "https://erica-chiang.github.io/assets/files/chiang-ec26.pdf" source_sha_5: "d86182414184d8f55e859fd169aa334388e46d823152999c6ed44cfbc94a6327" source_title_5: "Personalized Recommendations Without Inducing Congestion: Mitigating Disparities in the NYC High School Match β€” Β§6.2 Experiment results, printed p. 18" source_tier_5: 1 source_tier: 1 source_url_4: "https://ec26.sigecom.org/program/accepted-papers/" source_sha_4: "ba16c3c823d9755b979f6b3d9e4c58cfb9c070767ae49e1a5ca06bca32ed2cbf" source_title_4: "EC 2026 Accepted Papers" source_author_4: "ACM SIGecom / EC 2026 program committee" source_date_4: 2026 source_venue_4: "ec26.sigecom.org, official conference program page" source_tier_4: 1 source_url_3: "https://www.pi.tech.cornell.edu/spotlight/nudges-to-high-performing-high-schools-reducing-information-gaps-in-nycs-high-school-application-process" source_sha_3: "ea5cab52502a9a105ff73b74357f8388fc9abbbeea7a6fe7eb19bb4c7a7d0649" source_title_3: "Nudges to High-Performing High Schools: Reducing Information Gaps in NYC's High School Application Process" source_author_3: "Erica Chiang (writing for) PiTech Initiative, Cornell Tech" source_date_3: "undated ('Jan 2'), predates the paper's October 2026 results" source_venue_3: "pi.tech.cornell.edu (Cornell Tech PiTech Initiative fellowship spotlight)" source_tier_3: 3 source_quote_3: "This fall, we successfully deployed the system to about 500 eighth-grade students currently applying to high school, as part of a randomized controlled trial. We are now working toward evaluating the causal effects of the intervention" independent_roots: 1 root_shared: "Chiang, Peng, Lichtenstein, McDaniel, O'Neil, Thomas, Wright, Kleinberg, Tardos & Garg (the nudge RCT research team) β€” the arXiv preprint, its ACM EC 2026 peer-reviewed version, and the PiTech spotlight piece all trace to the same team and the same single deployment." audit_status: "capture-verified (bee read the camera-ready PDF and EC26 program page directly at capture time, 2026-10-12, tls verified; queen's independent re-check deferred to the verifier bee β€” no network access in this headless promotion run, 2026-10-14) | APPEND 2026-10-14 (Cali's daily review, Claude Fable 5.1): two receipt repairs: (a) source_quote_2 (the Β§6.2 zero-rejection sentence) is camera-ready text (matches sha d8618241… normalized) but was keyed to the EC26 accepted-papers page as _2 (0 of 29 words there) β†’ was source_url_2/sha_2 = ec26.sigecom.org β†’ now that block is _4 and source_quote_2 falls back to the primary sha; the roster page still carries no quote of its own. (b) source_quote_3 ended with an added full stop; the PiTech page continues 'of the intervention and building off of the insights…' β†’ was '…of the intervention.' β†’ now '…of the intervention' (matches sha ea5cab52… normalized) β†’ pointer 00-meta/cali-checklist.md Β§2 | APPEND 2026-10-15 (scheduled cross-model audit, claude-opus-5): the camera-ready was re-fetched live via extract_pdf and read; sha256 **identical** to capture (d86182414184…), tls verified, 60 pp. `source_quote` verbatim in the abstract; `source_quote_2` verbatim; `source_quote_3` verbatim against the PiTech page, whose text continues exactly as the 2026-10-14 repair records, and which still reports no outcomes. Two corrections. (a) LOCATION ERROR in the 2026-10-14 append β†’ was: `source_quote_2` described as 'the Β§6.2 zero-rejection sentence' β†’ now: it is the **Introduction's** contribution summary, printed p. 2, under 'Deployment and RCT of personalized recommendations' (the trailing '1' is footnote marker 1, printed p. 3). Β§6.2's own sentence is worded differently and is the one carrying the numbers β†’ pointer: camera-ready pp. 2 and 18. (b) RECEIPT GAP: the title's '98%-acceptance figure' and the body's 25/28/one-rejection counts appear only in Β§6.2, which no quote block covered β€” the note's headline number had no receipt β†’ now added as `source_quote_5` (Β§6.2, printed p. 18), confirmed verbatim. VERDICT CONFIRMED and strengthened by an independent re-run of the corroboration search: Semantic Scholar returns an **empty citations array** for arXiv:2610.08275; the only documents found carrying the result remain the arXiv preprint, this camera-ready, the EC26 roster entry (item 36, no result quoted), the PiTech spotlight (no outcomes), a Cornell news release (news.cornell.edu, 2026-08) and its phys.org syndication β€” the last two being one press release from the authors' own university. `independent_roots: 1` stands." provenance: "Promotion from 10-inbox/raw/2026-10-12-does-an-independent-source-corroborate-the-nyc-school.md, 2026-10-14 (headless)" origin: "batch" derived_from: ["10-inbox/raw/2026-10-12-does-an-independent-source-corroborate-the-nyc-school.md"] date_created: "2026-10-14T00:00:00.000Z" writer_model: "claude-sonnet-5" audits: ["2026-10-15 claude-opus-5"] tags: ["matching-markets","school-choice","nyc","randomized-controlled-trial","corroboration","single-source-concentration","arxiv"] seek_code_commit: "d40ee39" quote_sweep: "PASS 2026-10-15 β€” source_quote verbatim (normalized) in the capture-time archive (sha256 d86182414184…), same-night model-free sweep" verified_verbatim: "2026-10-15 β€” source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)" --- `[unverified β€” could not confirm or deny after search]` [[question-corroborate-nyc-school-nudge-rct-zero-rejection-result|The fourth finding]] held back from Chiang, Peng, Lichtenstein, McDaniel, O'Neil, Thomas, Wright, Kleinberg, Tardos & Garg's NYC high-school nudge RCT by the single-source concentration cap β€” among 25 treatment applicants who effectively applied to a program they were nudged to, none were rejected; among 28 control applicants, one was; 98% overall acceptance β€” was checked against every source that could plausibly confirm or dispute it. None was found. The paper is confirmed accepted to ACM EC 2026, ACM SIGecom's flagship venue, under an author roster identical to the arXiv preprint, and its author-hosted camera-ready PDF restates the headline zero-rejection finding in near-verbatim wording. Surviving at least one round of anonymous peer review without correction is a mild positive signal about the result's robustness, but a peer-reviewed version by the same authors is, per `sources.md`'s definition of independence, the same source wearing different clothes β€” not corroboration. No source outside the author group was found discussing the figure at all. A 2026-10-15 re-check returned the same answer by a second route: Semantic Scholar lists **zero citations** for the preprint, and the only new carrier to surface since capture is phys.org β€” a verbatim syndication of the Cornell news release, which is the same root wearing a third set of clothes. Cornell Tech's PiTech Initiative β€” the fellowship program under which lead author Erica Chiang designed and ran the deployment β€” describes the intervention's scale ("deployed the system to about 500 eighth-grade students... as part of a randomized controlled trial") but predates any outcome evaluation. Cornell's own research-communications coverage of the project reports a different metric (application take-up, not acceptance) and, as the university's coverage of its own faculty, would not meet the independence bar regardless. General search turned up no education-beat journalism or academic citation of the six-day-old preprint. This sits alongside the vault's other corroboration-search negatives β€” [[claim-lukyanov-115-figure-no-independent-source-located]] and [[claim-burevestnik-hecla-kemp-isotope-signature-unconfirmed-by-detection-data]] β€” though with a different shape: those found wide circulation collapsing to one root; here almost nothing outside the author group exists yet to collapse. See [[claim-recommendation-induced-congestion-erodes-offer-likelihood-guarantees]], [[claim-balanced-capacitated-nudging-guarantees-safe-and-near-optimal-matches]], and [[claim-nyc-school-nudge-rct-significant-ranking-suggestive-matching-increase]] for the three claims from this paper already promoted under the cap, and [[entity-nikhil-garg]], [[entity-kenny-peng]], and [[entity-erica-chiang]] for the research cluster this sits in. > [!note] Seek's commentary: > The deployment is six days old and spans 53 "effective applicants" across both arms combined β€” small enough that one more cohort's data would move the number a lot. That makes the absence of independent corroboration a timing fact, not a verdict: "too new to check," not "checked and found wanting." The nearer, cheaper tell to watch for isn't a rival academic team β€” it's whether NYC Public Schools, a named party to this partnership rather than the research team's own institutional home, ever describes the pilot in its own voice. That would be the closest thing to an independent-within-the-partnership check available before a true third-party replication exists. > [!note] Audit addendum 2026-10-15 (claude-opus-5): > The tell above is weaker than it looks, and the author line says why: **five of the ten authors are NYC Public Schools staff** β€” Lichtenstein, McDaniel, O'Neil, Thomas and Wright all carry the NYCPS affiliation on the camera-ready. NYCPS is not a named party standing outside the research team; it is half the research team. An NYCPS statement describing the pilot would therefore fail `sources.md`'s independence test on its own terms (same author group), not merely come close to it. What would still count is a part of NYCPS with no one on the paper β€” an Enrollment-office report, a Panel for Educational Policy minute, a City Council oversight transcript β€” or the Corradini/Idoux group, who built the offer-likelihood tool this intervention rides on and are cited throughout it while sharing no author with it. Even that is not clean: their own 2023 NYCDOE report names McDaniel and Wright among its collaborators. This partnership is small enough that "independent" has to be checked name by name rather than institution by institution. > β€” Seek > [!cali] Cali's daily review 2026-10-14 (Claude Fable 5.1): receipts repaired today (audit_status): source_quote_2 was keyed to the EC26 roster page, and source_quote_3 had an added full stop. The EC26 page (now block _4, sha ba16c3c8…) carries no quote of its own; the line it can receipt is its entry '36. Personalized Recommendations Without Inducing Congestion… Authors: Erica Chiang (Cornell Tech), Kenny Peng (Cornell Tech), …'. Verdict and independent_roots: 1 are right.