{"_canonicalization":{"envelope_id":"axm_ + sha256(envelope minus {signature, axiom_id, anchors})","envelope_signature":"ed25519(envelope minus {signature, axiom_id})","json":"sort_keys=True, separators=(',',':'), ensure_ascii=False, allow_nan=False, utf-8","leaf_hash":"sha256(0x00 || canonical_json(envelope_full))","seal_signature":"ed25519(seal minus {signature, sig_algorithm})"},"axiom_id":"axm_51e8ea4c2bbf395e2bd0f11883aef68e59bbb80f5ec7402701a7adcc6ab2bcdd","bitcoin_anchor":{"bitcoin_attestations":[],"calendar_attestations":[],"ots_url":"","stamped_at":"","status":"pending_next_stamp"},"envelope":{"anchors":[{"chain":"crovia.axiom_graph","height":0,"merkle_proof":"spider_arxiv_retraction_v1","root_at_anchor":"spider_arxiv_retraction_v1"}],"axiom_id":"axm_51e8ea4c2bbf395e2bd0f11883aef68e59bbb80f5ec7402701a7adcc6ab2bcdd","axiom_type":"AX.OBS","body":{"axiom_subtype":"research.arxiv_retraction.v1","category":"research","fingerprint":"87db8319548821e74dff59af0029c6918266795d927df341d98f35b2ef89aca2","published":"2026-05-25T15:57:11Z","receipt_hash":"87db8319548821e74dff59af0029c6918266795d927df341d98f35b2ef89aca2","schema":"spider.research.arxiv_retraction.v1","spider":"arxiv_retraction","spider_record":{"axiom_subtype":"research.arxiv_retraction.v1","category":"research","decision_hint":"POSITIVE","envelope_target":"AX.OBS","fingerprint":"87db8319548821e74dff59af0029c6918266795d927df341d98f35b2ef89aca2","observed_at":"2026-05-27T05:54:28.994851Z","parent_run_hash":"720fc40e3ea752b8ce1040c0176e81d5391dc3cf5bb440c5371e2a287193224f","published":"2026-05-25T15:57:11Z","runtime_version":"0.1.0","schema":"spider.research.arxiv_retraction.v1","source_status":200,"source_url":"http://export.arxiv.org/api/query?search_query=cat:cs.CL+AND+%28abs:withdrawn+OR+abs:retracted%29&max_results=20&sortBy=submittedDate&sortOrder=descending","spider":"arxiv_retraction","summary_excerpt":"We document an empirical phenomenon in chain-of-thought and ReAct agents driven by ten large language models from seven architecture families: meaning-bearing perturbations (e.g., paraphrase, synonym) alter final answers more often than presentation perturbations (e.g., formatting, reordering) of comparable severity. Across 68 cells spanning GSM8K, MATH, and HotpotQA (1,530 originals and $\\sim$11,150 variants), the inconsistency gap averages +19.69 pp after severity matching (paired $t=9.58$, $p<0.0001$), with 64/68 cells positive. The gap survives four severity-proxy audits and remains significant when excluding qwen models (+11.10 pp, $p<0.0001$). Several stress tests fail honestly: cluster-bootstrap significance disappears under stricter assumptions, tractability contrasts do not replicate, cross-architecture generator swaps break per-cell rankings, and a second LLM judge yields only moderate agreement ($κ=0.50$).\n  We then validate the headline effect on a fully held-out 11th model","title":"When Do LLM Agents Treat Surface Noise Differently from Semantic Noise? A 68-Cell Measurement Study with a Held-Out Trace-Level Validation","url":"https://arxiv.org/pdf/2605.25981v2","vendor":"arxiv"},"summary":"We document an empirical phenomenon in chain-of-thought and ReAct agents driven by ten large language models from seven architecture families: meaning-bearing perturbations (e.g., paraphrase, synonym) alter final answers more often than presentation perturbations (e.g., formatting, reordering) of comparable severity. Across 68 cells spanning GSM8K, MATH, and HotpotQA (1,530 originals and $\\sim$11,150 variants), the inconsistency gap averages +19.69 pp after severity matching (paired $t=9.58$, $p<0.0001$), with 64/68 cells positive. The gap survives four severity-proxy audits and remains significant when excluding qwen models (+11.10 pp, $p<0.0001$). Several stress tests fail honestly: cluster-bootstrap significance disappears under stricter assumptions, tractability contrasts do not replicate, cross-architecture generator swaps break per-cell rankings, and a second LLM judge yields only moderate agreement ($κ=0.50$).\n  We then validate the headline effect on a fully held-out 11th model","title":"When Do LLM Agents Treat Surface Noise Differently from Semantic Noise? A 68-Cell Measurement Study with a Held-Out Trace-Level Validation","vendor":"arxiv"},"confidence":{"method":"deterministic"},"decision":"POSITIVE","issued_at":"2026-05-27T05:54:28Z","notes":"Spider arxiv_retraction (research) research.arxiv_retraction.v1","object":{"captured_by":"crovia.spider.arxiv_retraction","primary_source_url":"https://arxiv.org/pdf/2605.25981v2"},"predecessors":[],"schema":"crovia.axiom.v1","signature":"ed25519:92b9808209414da168652668231280a3057eece304d87d5e73fdefc60d688891b2c83e1b3d9ccd3febadf242aa84b2df15d3359782c0027c791fe356a52f0c03","signer":"crovia.substrate","subject":{"observed_at":"2026-05-27T05:54:28Z","source_collector":"spider:arxiv_retraction","target_id":"https://arxiv.org/pdf/2605.25981v2"},"tsa":{"authority":"crovia.substrate.bootstrap","rfc3161_token":"{\"kind\":\"crovia.bootstrap.tsa\",\"source_jsonl\":\"/opt/crovia/spider/data/research/arxiv_retraction_v1.jsonl\",\"source_seal_merkle_root\":\"spider_arxiv_retraction_v1\",\"upgrade_path\":\"Sessione H \\u2014 OpenTimestamps weekly anchor\"}"},"zk_mode":"clear","zk_proof":null},"ledger":{"leaf_hash":"a174f9fb7fcfeb84e927beafd9ba6e3f696b3cc67dd637653043f965374b2199","leaf_index":154188,"ledger_path":"/opt/crovia/substrate/axiom_ledger.jsonl"},"merkle_proof":{"hash_alg":"sha256","leaf_prefix":"0x00","node_prefix":"0x01","odd_leaf_rule":"duplicate_last","path":[{"sibling":"ce3efaa83950a5472da7101763398e7584c026402a611860ecbbbd36bc053cbc","side":"right"},{"sibling":"be38e6350c845a28be49df2de3d2c1ddf61bc43ff1277cd22ea4dc1dac90acb9","side":"right"},{"sibling":"3ad9602b94ab06e8d2226fadf1608c2c30c8b04586762d4fe5409d5c9f53dc6a","side":"left"},{"sibling":"5b33a8cfe795367039d390db04cf78dc50f73b37380743b189aa454555c1e968","side":"left"},{"sibling":"5f8e3d4713950a1cb15aab7073d1cbb6fa0576e03cb9296c57159ed0d5902e64","side":"right"},{"sibling":"a3ba54fe8e040e1c203bf5a5a607fe65b67789d20798276e891bed4598092c6e","side":"right"},{"sibling":"f7895620ccd5f3b9d5289d1a5b162d7ba3e006f63da154bf7f21da4ee8fcdbee","side":"left"},{"sibling":"4c6415e17f9ebc4ec5e0807b87126247631b24ecb1b83c28326cfd1c39fa3e3c","side":"right"},{"sibling":"19816d1fbcece31f77e724fd986c3733094934ae2d89c9404204d74bcc1aa1b3","side":"right"},{"sibling":"3769e896ad5ac9922f96e71b9de538b8401eb0c58e836b6591711f1b3dd16f5b","side":"left"},{"sibling":"3ff61a0e7c1cc6d586251f700f29bbb51578814704b03a56cdc61ae3a9230dc8","side":"right"},{"sibling":"816f233274bb10f5a122aac086a0c8c697b78fec67a4af55190bb596b7506fab","side":"left"},{"sibling":"d415e6939aee710631f5062799379b547d2c3e3d9a68f263bbb5a693285ab2ca","side":"left"},{"sibling":"80d20c7621442c173ec547537d49d0af63d04ede6970c58d240953c55f2f8fd5","side":"right"},{"sibling":"35ca36cee447f0ef7064a25d55f59357c66901e31427729c4c1d8b14aa8adb6c","side":"left"},{"sibling":"8a208d550c505b81b7763d35e4d0f0fa4c68ee82ee7bb710f4bfd3749fd75e81","side":"right"},{"sibling":"471a58ac8b631d71c780279b860e0219c98b9c6c0645138a19a46d6b36bc53f4","side":"right"},{"sibling":"d841ad93efda0869e5eb97678f348f03f5caab4353e05ff4bf18f47fb945b822","side":"left"}]},"schema":"crovia.axiom_proof.v1","seal":{"first_collector_run_id":"","first_receipt_hash":"","jsonl_path":"/opt/crovia/substrate/axiom_ledger.jsonl","key_id":"430895f101d38164","last_collector_run_id":"","last_receipt_hash":"","leaf_count":154238,"merkle_root":"6c6001511bb1d7ead0158e5fac89f235d72d2f1e0a91a147f4db9c564221ee51","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","run_id":"hourly_json_retrofit_20260527T063701Z","schema":"crovia.seal.v1","seal_family_version":"crovia-seal-family/1","seal_kind":"substrate_batch","sealed_at":"2026-05-27T06:37:35Z","sig_algorithm":"ed25519","signature":"69877b57da0ed2d8eeab6bca5394d02f2fd2d0db5ee3cc266c28e0ebb6a7af56b1da3474ebaf9cb78638d8c56c83cc9147cb72d18be30ad684e31054c7d34203","signer_version":"1.1.0"},"trust_root":{"key_id":"430895f101d38164","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","signature_algorithm":"ed25519","url":"/registry/canon/TRUST_ROOT.md"},"verifier":{"spec":"/registry/canon/AXIOM_RECEIPT_v1.md","url":"/v/axm_51e8ea4c2bbf395e2bd0f11883aef68e59bbb80f5ec7402701a7adcc6ab2bcdd"}}