{"as_of":"2026-08-19T14:49:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:fa6ffb4a4e7259d3b66f574a707dfd5d53c0fe7e1522c644304f48d600c0b2b8","coverage":[{"denominator":0,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":19,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":19,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-19T06:32:44.657259+00:00","state":"measured"},{"denominator":19,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":19,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-11T23:15:28.530737Z","state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":1,"source":"arxiv_reference","source_observed_at":"2026-08-05T02:28:24.338817Z","state":"measured"}],"external_citation_measurements":[{"count":9,"observed_at":"2026-08-05T02:28:24.338817Z","source":"arxiv_reference"}],"inbound":[{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-11T23:15:28.530737Z","title":"Large language models suffer from their own output: An analysis of the self-consuming training loop.arXiv preprint arXiv:2311.16822,","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2412.02674","last_updated":"2025-02-25T16:59:11Z","snapshot_observed_at":"2026-08-14T20:58:55.959220Z","submitted_at":"2024-12-03T18:47:26Z","title":"Mind the Gap: Examining the Self-Improvement Capabilities of Large Language Models","version":2},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-11T23:15:28.530737Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2412.02674"},"observation_digest":"sha256:141466f76d1d3ddb79319e0cd84fccddd246fc2a24126a0e7c1b3f9db2ab113b","observation_id":"d2c8edaf-45f1-49d0-9a5a-abaa4a59eb3e","resolution":{"observed_at":"2026-08-11T23:15:28.530737Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-11T15:02:39.477913Z","title":null,"venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2412.11385","last_updated":"2024-12-16T02:27:59Z","snapshot_observed_at":"2026-08-18T03:17:20.694563Z","submitted_at":"2024-12-16T02:27:59Z","title":"Why Does ChatGPT \"Delve\" So Much? Exploring the Sources of Lexical Overrepresentation in Large Language Models","version":1},"reference_index":4,"source":"arxiv_source","source_observed_at":"2026-08-11T15:02:39.477913Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2412.11385"},"observation_digest":"sha256:94788c9b79aada3f4f91dfece690f023394ec05d3ab5c3114ce51aaf0bacd327","observation_id":"37597351-7ede-412e-bcf7-547cb9633a1c","resolution":{"observed_at":"2026-08-11T15:02:39.477913Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-11T12:05:48.412401Z","title":"Large language models suffer from their own output: An analysis of the self-consuming training loop","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2412.14689","last_updated":"2025-05-28T05:22:58Z","snapshot_observed_at":"2026-08-15T08:21:08.511659Z","submitted_at":"2024-12-19T09:43:39Z","title":"How to Synthesize Text Data without Model Collapse?","version":3},"reference_index":13,"source":"arxiv_source","source_observed_at":"2026-08-11T12:05:48.412401Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2412.14689"},"observation_digest":"sha256:234697d0b925da22373ad382a5ece320bf98b5d901f503574ed094d2cfc48cb9","observation_id":"04670e5a-9ed3-4656-8cfa-114ded09bf9b","resolution":{"observed_at":"2026-08-11T12:05:48.412401Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-11T11:55:12.078470Z","title":"Large language models suffer from their own output: An analysis of the self-consuming training loop, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2412.14872","last_updated":"2025-05-19T12:48:01Z","snapshot_observed_at":"2026-08-15T16:07:43.509929Z","submitted_at":"2024-12-19T14:11:15Z","title":"Theoretical Proof that Auto-regressive Language Models Collapse when Real-world Data is a Finite Set","version":3},"reference_index":4,"source":"arxiv_source","source_observed_at":"2026-08-11T11:55:12.078470Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2412.14872"},"observation_digest":"sha256:0bdea96b43d0605452658127d1f58a6f504468804aeb8a78777ed492355b3b16","observation_id":"c8d65c2d-7c1f-4f28-a378-7f42cff081bf","resolution":{"observed_at":"2026-08-11T11:55:12.078470Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-11T05:04:51.850670Z","title":"Large language models suffer from their own output: An analysis of the self-consuming training loop.CoRR, abs/2311.16822, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2412.18148","last_updated":"2025-06-01T15:06:21Z","snapshot_observed_at":"2026-08-17T08:46:50.230603Z","submitted_at":"2024-12-24T04:04:54Z","title":"Are We in the AI-Generated Text World Already? Quantifying and Monitoring AIGT on Social Media","version":3},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-11T05:04:51.850670Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2412.18148"},"observation_digest":"sha256:f4360f3b52ce5810236a3c33dc00bb02037c62ffd70a570830b70e89368e56e7","observation_id":"6cbf3e27-7c68-41c6-8cdf-7eb0c9e00dd2","resolution":{"observed_at":"2026-08-11T05:04:51.850670Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-09T23:25:11.165137Z","title":null,"venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2501.18479","last_updated":"2025-01-30T16:51:44Z","snapshot_observed_at":"2026-08-16T17:03:27.469968Z","submitted_at":"2025-01-30T16:51:44Z","title":"Transformer Semantic Genetic Programming for Symbolic Regression","version":1},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-09T23:25:11.165137Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2501.18479"},"observation_digest":"sha256:babcb6717cfb50a5950f6a14db4d8d7eb8cc10b6ea883a1ea73f36733a014c44","observation_id":"288bc404-e1e1-4219-95e9-b1dd438e5c94","resolution":{"observed_at":"2026-08-09T23:25:11.165137Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-09T14:54:29.116977Z","title":"Large language models suffer from their own output: An analysis of the self-consuming training loop","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2502.01612","last_updated":"2025-02-13T05:32:54Z","snapshot_observed_at":"2026-08-18T05:44:18.840497Z","submitted_at":"2025-02-03T18:45:22Z","title":"Self-Improving Transformers Overcome Easy-to-Hard and Length Generalization Challenges","version":2},"reference_index":12,"source":"arxiv_source","source_observed_at":"2026-08-09T14:54:29.116977Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2502.01612"},"observation_digest":"sha256:01ccdd34bb8c2bd59c9aaa4de7a675aff4dc48a49981a0694984c1c9c55165fa","observation_id":"cb20644f-f92e-4437-8b47-d1ee2af1b2ba","resolution":{"observed_at":"2026-08-09T14:54:29.116977Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-06T23:12:18.868972Z","title":"Large language models suffer from their own output: An analysis of the self-consuming training loop.arXiv preprint arXiv:2311.16822, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2506.19262","last_updated":"2025-06-25T03:25:04Z","snapshot_observed_at":"2026-08-16T15:41:02.536314Z","submitted_at":"2025-06-24T02:44:58Z","title":"What Matters in LLM-generated Data: Diversity and Its Effect on Model Fine-Tuning","version":2},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-06T23:12:18.868972Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2506.19262"},"observation_digest":"sha256:7838c834a46c336aeb331e18d0c96378f25d97f1dcbaab5aec00347d75d1f174","observation_id":"6492b30d-3ec5-4f39-a21b-099f9cb35650","resolution":{"observed_at":"2026-08-06T23:12:18.868972Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-06T22:57:07.176261Z","title":"Briesch, D","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2506.20623","last_updated":"2025-07-09T08:24:09Z","snapshot_observed_at":"2026-08-18T12:09:59.279948Z","submitted_at":"2025-06-25T17:12:22Z","title":"Lost in Retraining: Roaming the Parameter Space of Exponential Families Under Closed-Loop Learning","version":2},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-06T22:57:07.176261Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2506.20623"},"observation_digest":"sha256:ec678a0e8805aff6f0825ae17241986e92ad2df0fdd48e31356d944176774153","observation_id":"30100e79-17d3-4daf-a61d-9c70d531b617","resolution":{"observed_at":"2026-08-06T22:57:07.176261Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-06T22:24:07.440741Z","title":null,"venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2506.21817","last_updated":"2025-06-26T23:44:24Z","snapshot_observed_at":"2026-08-18T22:50:24.116373Z","submitted_at":"2025-06-26T23:44:24Z","title":"Exploring the Structure of AI-Induced Language Change in Scientific English","version":1},"reference_index":7,"source":"arxiv_source","source_observed_at":"2026-08-06T22:24:07.440741Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2506.21817"},"observation_digest":"sha256:2679abcc10eeed8de63f4b87e5baa82fa6fb10943d8ce608d6d3a1a9adaf7dad","observation_id":"aa16beea-3c79-4a1e-9e3d-85a983158c79","resolution":{"observed_at":"2026-08-06T22:24:07.440741Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-06T10:23:57.646894Z","title":null,"venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2508.00238","last_updated":"2025-08-01T00:47:33Z","snapshot_observed_at":"2026-08-17T08:06:42.842425Z","submitted_at":"2025-08-01T00:47:33Z","title":"Model Misalignment and Language Change: Traces of AI-Associated Language in Unscripted Spoken English","version":1},"reference_index":16,"source":"arxiv_source","source_observed_at":"2026-08-06T10:23:57.646894Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2508.00238"},"observation_digest":"sha256:8b6a509f166988d8188b1183ffbe9c552c453e8cf06ca9ba979fd120270b128b","observation_id":"93203c79-89e5-4243-80c5-b5445b5f3150","resolution":{"observed_at":"2026-08-06T10:23:57.646894Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":"2311.16822","doi":"10.48550/arxiv.2311.16822","metadata_source":"arxiv_reference","pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"arXiv preprint , volume =","venue":"arXiv (Cornell University)","work_id":"e2c05930-ca41-44b9-927b-bfc969a6f408","year":2023},"citing_paper":{"arxiv_id":"2511.09416","last_updated":"2026-04-30T13:00:21Z","snapshot_observed_at":"2026-08-13T22:43:52.944756Z","submitted_at":"2025-11-12T15:32:50Z","title":"Transformer Semantic Genetic Programming for d-dimensional Symbolic Regression Problems","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-05-17T23:12:49.289629Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2511.09416"},"observation_digest":"sha256:d59a1245bc60a119d8b28c0f768c2bf83cc771f7e6aa257fb14b0cf4b49d5fb5","observation_id":"a3d80535-11ba-4d41-838d-668bc292f6b5","resolution":{"observed_at":"2026-05-17T23:15:26.797407Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"crossref_status_cache"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"openalex_status_cache"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":"2311.16822","doi":"10.48550/arxiv.2311.16822","metadata_source":"arxiv_reference","pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"arXiv preprint , volume =","venue":"arXiv (Cornell University)","work_id":"e2c05930-ca41-44b9-927b-bfc969a6f408","year":2023},"citing_paper":{"arxiv_id":"2605.20151","last_updated":"2026-05-19T17:41:09Z","snapshot_observed_at":"2026-08-12T23:02:11.872447Z","submitted_at":"2026-05-19T17:41:09Z","title":"When Does Model Collapse Occur in Structured Interactive Learning?","version":1},"reference_index":40,"source":"arxiv_source","source_observed_at":"2026-05-20T06:31:36.248669Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2605.20151"},"observation_digest":"sha256:8c0d71d123e534cd28ec7b4926c415cfd62bce4a3f60b457b9581dd8da15275e","observation_id":"831523d4-9212-40be-8ba6-f00e26ef0ebf","resolution":{"observed_at":"2026-05-20T06:33:05.622872Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"crossref_status_cache"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"openalex_status_cache"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":"2311.16822","doi":"10.48550/arxiv.2311.16822","metadata_source":"arxiv_reference","pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"arXiv preprint , volume =","venue":"arXiv (Cornell University)","work_id":"e2c05930-ca41-44b9-927b-bfc969a6f408","year":2023},"citing_paper":{"arxiv_id":"2605.20279","last_updated":"2026-08-15T18:09:26Z","snapshot_observed_at":"2026-08-19T11:14:49.314383Z","submitted_at":"2026-05-19T04:41:39Z","title":"The Economics of Model Collapse: Equilibrium, Welfare, and Optimal Provenance Subsidies in Synthetic Data Markets","version":1},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-05-21T02:09:27.275578Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2605.20279"},"observation_digest":"sha256:071bb498aa1323d64a69e440df6447fa1e1ec9bc0e170d30826010ccf03990ff","observation_id":"e7ecc8da-39e6-45f7-bd1c-f165d5915104","resolution":{"observed_at":"2026-05-21T02:13:56.563756Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"crossref_status_cache"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"openalex_status_cache"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":"2311.16822","doi":"10.48550/arxiv.2311.16822","metadata_source":"arxiv_reference","pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"arXiv preprint , volume =","venue":"arXiv (Cornell University)","work_id":"e2c05930-ca41-44b9-927b-bfc969a6f408","year":2023},"citing_paper":{"arxiv_id":"2605.20602","last_updated":"2026-05-20T01:44:47Z","snapshot_observed_at":"2026-07-06T23:31:08.671011Z","submitted_at":"2026-05-20T01:44:47Z","title":"Self-Training Doesn't Flatten Language -- It Restructures It: Surface Markers Amplify While Deep Syntax Dies","version":1},"reference_index":3,"source":"arxiv_source","source_observed_at":"2026-05-21T05:46:07.176408Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2605.20602"},"observation_digest":"sha256:9c30a52b4281e2abf4fc06cdec5fb8b776fc1bcbb2593a21cc4dc0e028bebd81","observation_id":"ebe232d6-fd0f-44cd-9e96-c3ea0617472f","resolution":{"observed_at":"2026-05-21T05:49:41.099042Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"crossref_status_cache"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"openalex_status_cache"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":"2311.16822","doi":"10.48550/arxiv.2311.16822","metadata_source":"arxiv_reference","pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"arXiv preprint , volume =","venue":"arXiv (Cornell University)","work_id":"e2c05930-ca41-44b9-927b-bfc969a6f408","year":2023},"citing_paper":{"arxiv_id":"2605.23054","last_updated":"2026-05-21T21:36:26Z","snapshot_observed_at":"2026-08-19T14:24:53.411347Z","submitted_at":"2026-05-21T21:36:26Z","title":"Model Collapse as Cultural Evolution","version":1},"reference_index":9,"source":"arxiv_source","source_observed_at":"2026-05-25T05:30:31.504863Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2605.23054"},"observation_digest":"sha256:170abbe8c3fb7496a9b95f8f539099942371cfc71df4c60ed4a106ebd03c5f2a","observation_id":"646dc64f-2811-41fe-9107-0380d2c1a65b","resolution":{"observed_at":"2026-05-25T05:35:23.369425Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"crossref_status_cache"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"openalex_status_cache"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":"2311.16822","doi":"10.48550/arxiv.2311.16822","metadata_source":"arxiv_reference","pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"arXiv preprint , volume =","venue":"arXiv (Cornell University)","work_id":"e2c05930-ca41-44b9-927b-bfc969a6f408","year":2023},"citing_paper":{"arxiv_id":"2606.00334","last_updated":"2026-05-29T20:19:49Z","snapshot_observed_at":"2026-08-15T14:31:48.355360Z","submitted_at":"2026-05-29T20:19:49Z","title":"Isolating LLM Lexical Bias: A Curation-Free Triangulated Metric for Preference-Stage Learning","version":1},"reference_index":29,"source":"arxiv_source","source_observed_at":"2026-06-28T22:04:42.187726Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2606.00334"},"observation_digest":"sha256:ec3876cb6463c232f62efeb1212de64697ae342b56797246be9e31315f632629","observation_id":"9c5bf529-d30c-40c7-b5b7-81c8a3abdcdd","resolution":{"observed_at":"2026-07-01T19:46:10.746390Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"crossref_status_cache"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"openalex_status_cache"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":"2311.16822","doi":"10.48550/arxiv.2311.16822","metadata_source":"arxiv_reference","pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"arXiv preprint , volume =","venue":"arXiv (Cornell University)","work_id":"e2c05930-ca41-44b9-927b-bfc969a6f408","year":2023},"citing_paper":{"arxiv_id":"2606.28438","last_updated":"2026-06-26T07:35:43Z","snapshot_observed_at":"2026-08-03T00:04:55.169727Z","submitted_at":"2026-06-26T07:35:43Z","title":"When AI Reviews Its Own Code: Recursive Self-Training Collapse in Code LLMs","version":1},"reference_index":150,"source":"arxiv_source","source_observed_at":"2026-06-30T01:29:42.919461Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2606.28438"},"observation_digest":"sha256:ebcc02792646f6780da15adba7e2454fb569d3b164c850eb040006e0a1914630","observation_id":"c90858a4-0ad9-4293-b1be-d37af326c69a","resolution":{"observed_at":"2026-06-30T01:34:09.463145Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"crossref_status_cache"},{"observed_at":"2026-05-25T16:23:20.922906+00:00","source":"openalex_status_cache"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.16822","snapshot_observed_at":"2026-08-01T12:54:30.522763Z","title":null,"venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.19292","last_updated":"2026-07-21T17:02:37Z","snapshot_observed_at":"2026-08-14T07:56:28.626682Z","submitted_at":"2026-07-21T17:02:37Z","title":"The safety failures we are not instrumenting: a perspective on hidden safety-critical challenges in modern AI systems","version":1},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-01T12:54:30.522763Z"},"links":{"cited_paper":"/paper/2311.16822","citing_paper":"/paper/2607.19292"},"observation_digest":"sha256:37708cf4cd4eb302a49e7609b32f6e84ff4f0a898b269ff3cdd0beede10bcc36","observation_id":"03798236-80b7-4d94-bd33-09cdb664a487","resolution":{"observed_at":"2026-08-01T12:54:30.522763Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"links":{"evidence":"/evidence","html":"/paper/2311.16822/citation-record","integrity":"/paper/2311.16822/integrity","json":"/paper/2311.16822/citation-record.json","paper":"/paper/2311.16822"},"outbound":[],"paper":{"arxiv_id":"2311.16822","last_updated":"2024-06-17T07:07:30Z","latest_version":2,"primary_category":"cs.LG","snapshot_observed_at":"2026-08-19T08:52:42.052171Z","submitted_at":"2023-11-28T14:36:43Z","title":"Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop"},"reference_resolution":{"displayed":0,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":0,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":0},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"thesis":"As of 19 August 2026, this Paper Citation Record lists 0 of 0 outbound references and 19 inbound Pith citation observations for arXiv:2311.16822."}