{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7PSRDRVIFUCQ4RGPDL2ZRFTJ4R","short_pith_number":"pith:7PSRDRVI","schema_version":"1.0","canonical_sha256":"fbe511c6a82d050e44cf1af5989669e47e53a8eb0a3e73f21bf120805a73b5b5","source":{"kind":"arxiv","id":"2605.15761","version":1},"attestation_state":"computed","paper":{"title":"A Unified Perturbation Framework for Analyzing Leaderboard Stability and Manipulation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Amir-Hossein Karimi, Hosna Oyarhoseini, Jimmy Lin","submitted_at":"2026-05-15T09:21:33Z","abstract_excerpt":"Evaluation leaderboards such as LMArena play a central role in benchmarking large language models by aggregating pairwise human preferences into model rankings, yet the robustness of these rankings remains poorly understood. We present a unified perturbation framework for analyzing Bradley-Terry leaderboards under structured data modifications using influence-based approximations. Our framework studies three match-level perturbations -- Drop, Add, and Flip -- together with player removal, and evaluates their effects on top-k membership, global ranking consistency via Kendall's tau, and confide"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.15761","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-15T09:21:33Z","cross_cats_sorted":[],"title_canon_sha256":"a62c7b9dc0c8600cf90d3ccb8937680f251faecbae6d5fc7f2c8d872eded21de","abstract_canon_sha256":"105e9a22547ee43022a7a04b736fde54cd556baebbccf9ae714542cabcdbf80d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:01:16.827670Z","signature_b64":"OsaVekoYATDPY0X3keC/TyNvfkeA0xg2XNdfSEXys1QKg6XcWRkJNIAM7FWfY1Nevh95TSMSEMU0IpNOz/azBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fbe511c6a82d050e44cf1af5989669e47e53a8eb0a3e73f21bf120805a73b5b5","last_reissued_at":"2026-05-20T00:01:16.827020Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:01:16.827020Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Unified Perturbation Framework for Analyzing Leaderboard Stability and Manipulation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Amir-Hossein Karimi, Hosna Oyarhoseini, Jimmy Lin","submitted_at":"2026-05-15T09:21:33Z","abstract_excerpt":"Evaluation leaderboards such as LMArena play a central role in benchmarking large language models by aggregating pairwise human preferences into model rankings, yet the robustness of these rankings remains poorly understood. We present a unified perturbation framework for analyzing Bradley-Terry leaderboards under structured data modifications using influence-based approximations. Our framework studies three match-level perturbations -- Drop, Add, and Flip -- together with player removal, and evaluates their effects on top-k membership, global ranking consistency via Kendall's tau, and confide"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.15761","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15761/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-19T17:33:48.763431Z","status":"skipped","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T17:21:55.955091Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"27cbd802131d70fec3a5c691dd163ff9ea8faa6066fbb2c5d5036ce017d4595c"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.15761","created_at":"2026-05-20T00:01:16.827116+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.15761v1","created_at":"2026-05-20T00:01:16.827116+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15761","created_at":"2026-05-20T00:01:16.827116+00:00"},{"alias_kind":"pith_short_12","alias_value":"7PSRDRVIFUCQ","created_at":"2026-05-20T00:01:16.827116+00:00"},{"alias_kind":"pith_short_16","alias_value":"7PSRDRVIFUCQ4RGP","created_at":"2026-05-20T00:01:16.827116+00:00"},{"alias_kind":"pith_short_8","alias_value":"7PSRDRVI","created_at":"2026-05-20T00:01:16.827116+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R","json":"https://pith.science/pith/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R.json","graph_json":"https://pith.science/api/pith-number/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R/graph.json","events_json":"https://pith.science/api/pith-number/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R/events.json","paper":"https://pith.science/paper/7PSRDRVI"},"agent_actions":{"view_html":"https://pith.science/pith/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R","download_json":"https://pith.science/pith/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R.json","view_paper":"https://pith.science/paper/7PSRDRVI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.15761&json=true","fetch_graph":"https://pith.science/api/pith-number/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R/graph.json","fetch_events":"https://pith.science/api/pith-number/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R/action/storage_attestation","attest_author":"https://pith.science/pith/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R/action/author_attestation","sign_citation":"https://pith.science/pith/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R/action/citation_signature","submit_replication":"https://pith.science/pith/7PSRDRVIFUCQ4RGPDL2ZRFTJ4R/action/replication_record"}},"created_at":"2026-05-20T00:01:16.827116+00:00","updated_at":"2026-05-20T00:01:16.827116+00:00"}