{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:SVJR56AFLUSQCWA6Q3K4ME2NJI","short_pith_number":"pith:SVJR56AF","schema_version":"1.0","canonical_sha256":"95531ef8055d2501581e86d5c6134d4a3796d5bf9260fb5231af35a05bc9d9e6","source":{"kind":"arxiv","id":"2605.20204","version":1},"attestation_state":"computed","paper":{"title":"RealUserSim: Bridging the Reality Gap in Agent Benchmarking via Grounded User Simulation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.HC","authors_text":"Huan Wang, Jielin Qiu, Juntao Tan, Liangwei Yang, Ming Zhu, Rithesh Murthy, Shelby Heinecke, Silvio Savarese, Wenting Zhao","submitted_at":"2026-04-07T19:42:36Z","abstract_excerpt":"LLM-based user simulation is the primary mechanism for end-to-end agent evaluation, yet simulated users are poor proxies for real humans: unconstrained LLM defaults produce a Formalism Ceiling (style match rates of 6-8% against real users), while hand-crafted behavioral directives trigger Directive Amplification, where models hyper-interpret instructions into unnatural behavioral extremes that vary dramatically across simulator models. We present RealUserSim, the first user simulation framework grounded in real behavioral data. From 14,000+ authentic human-LLM conversations (WildChat), we extr"},"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":"2605.20204","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.HC","submitted_at":"2026-04-07T19:42:36Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"fb55cb039f9369043961e9dee1ee8f86164eed8387882caf140540c8842094d0","abstract_canon_sha256":"45ae21df8e42a989e73c1a490f57b134378476e18039e193ff5182ce924f8eec"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T00:04:21.004585Z","signature_b64":"pmzox9WLuIz0VWgc+US8CEbZ9STSFNy8GzJRV6VRe7yyzgi01wBHHxcqzQHuQynQQOJLcCF1UAabZRxbBitGAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"95531ef8055d2501581e86d5c6134d4a3796d5bf9260fb5231af35a05bc9d9e6","last_reissued_at":"2026-05-21T00:04:21.003815Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T00:04:21.003815Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RealUserSim: Bridging the Reality Gap in Agent Benchmarking via Grounded User Simulation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.HC","authors_text":"Huan Wang, Jielin Qiu, Juntao Tan, Liangwei Yang, Ming Zhu, Rithesh Murthy, Shelby Heinecke, Silvio Savarese, Wenting Zhao","submitted_at":"2026-04-07T19:42:36Z","abstract_excerpt":"LLM-based user simulation is the primary mechanism for end-to-end agent evaluation, yet simulated users are poor proxies for real humans: unconstrained LLM defaults produce a Formalism Ceiling (style match rates of 6-8% against real users), while hand-crafted behavioral directives trigger Directive Amplification, where models hyper-interpret instructions into unnatural behavioral extremes that vary dramatically across simulator models. We present RealUserSim, the first user simulation framework grounded in real behavioral data. From 14,000+ authentic human-LLM conversations (WildChat), we extr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.20204","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/2605.20204/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"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":"2605.20204","created_at":"2026-05-21T00:04:21.003939+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.20204v1","created_at":"2026-05-21T00:04:21.003939+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.20204","created_at":"2026-05-21T00:04:21.003939+00:00"},{"alias_kind":"pith_short_12","alias_value":"SVJR56AFLUSQ","created_at":"2026-05-21T00:04:21.003939+00:00"},{"alias_kind":"pith_short_16","alias_value":"SVJR56AFLUSQCWA6","created_at":"2026-05-21T00:04:21.003939+00:00"},{"alias_kind":"pith_short_8","alias_value":"SVJR56AF","created_at":"2026-05-21T00:04:21.003939+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/SVJR56AFLUSQCWA6Q3K4ME2NJI","json":"https://pith.science/pith/SVJR56AFLUSQCWA6Q3K4ME2NJI.json","graph_json":"https://pith.science/api/pith-number/SVJR56AFLUSQCWA6Q3K4ME2NJI/graph.json","events_json":"https://pith.science/api/pith-number/SVJR56AFLUSQCWA6Q3K4ME2NJI/events.json","paper":"https://pith.science/paper/SVJR56AF"},"agent_actions":{"view_html":"https://pith.science/pith/SVJR56AFLUSQCWA6Q3K4ME2NJI","download_json":"https://pith.science/pith/SVJR56AFLUSQCWA6Q3K4ME2NJI.json","view_paper":"https://pith.science/paper/SVJR56AF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.20204&json=true","fetch_graph":"https://pith.science/api/pith-number/SVJR56AFLUSQCWA6Q3K4ME2NJI/graph.json","fetch_events":"https://pith.science/api/pith-number/SVJR56AFLUSQCWA6Q3K4ME2NJI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SVJR56AFLUSQCWA6Q3K4ME2NJI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SVJR56AFLUSQCWA6Q3K4ME2NJI/action/storage_attestation","attest_author":"https://pith.science/pith/SVJR56AFLUSQCWA6Q3K4ME2NJI/action/author_attestation","sign_citation":"https://pith.science/pith/SVJR56AFLUSQCWA6Q3K4ME2NJI/action/citation_signature","submit_replication":"https://pith.science/pith/SVJR56AFLUSQCWA6Q3K4ME2NJI/action/replication_record"}},"created_at":"2026-05-21T00:04:21.003939+00:00","updated_at":"2026-05-21T00:04:21.003939+00:00"}