{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:QMSCD6VHU4SX5CHEYOPLIXMLOV","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"8e9bcd3eac2595cb3e6fb34babd54c1afa0cc28571e93f62cd856124630e9ae8","cross_cats_sorted":["stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-09-27T00:32:44Z","title_canon_sha256":"633cff32c20d18e56d3d1a9594d9b42a3ee70e1ae1e23c5ac27d3a870360a9db"},"schema_version":"1.0","source":{"id":"2209.13085","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2209.13085","created_at":"2026-07-05T10:25:01Z"},{"alias_kind":"arxiv_version","alias_value":"2209.13085v2","created_at":"2026-07-05T10:25:01Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2209.13085","created_at":"2026-07-05T10:25:01Z"},{"alias_kind":"pith_short_12","alias_value":"QMSCD6VHU4SX","created_at":"2026-07-05T10:25:01Z"},{"alias_kind":"pith_short_16","alias_value":"QMSCD6VHU4SX5CHE","created_at":"2026-07-05T10:25:01Z"},{"alias_kind":"pith_short_8","alias_value":"QMSCD6VH","created_at":"2026-07-05T10:25:01Z"}],"graph_snapshots":[{"event_id":"sha256:a686ad141ce8999fa9cd6f6412c88c90cb3a3197270cd3c8e9ec34a9417c0764","target":"graph","created_at":"2026-07-05T10:25:01Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2209.13085/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We provide the first formal definition of reward hacking, a phenomenon where optimizing an imperfect proxy reward function leads to poor performance according to the true reward function. We say that a proxy is unhackable if increasing the expected proxy return can never decrease the expected true return. Intuitively, it might be possible to create an unhackable proxy by leaving some terms out of the reward function (making it \"narrower\") or overlooking fine-grained distinctions between roughly equivalent outcomes, but we show this is usually not the case. A key insight is that the linearity o","authors_text":"David Krueger, Dmitrii Krasheninnikov, Joar Skalse, Nikolaus H. R. Howe","cross_cats":["stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-09-27T00:32:44Z","title":"Defining and Characterizing Reward Hacking"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2209.13085","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:18eea9d53a49da0ff3c2843dd4c39d8dec5839176746ae7ba10d6e681ef6533b","target":"record","created_at":"2026-07-05T10:25:01Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"8e9bcd3eac2595cb3e6fb34babd54c1afa0cc28571e93f62cd856124630e9ae8","cross_cats_sorted":["stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-09-27T00:32:44Z","title_canon_sha256":"633cff32c20d18e56d3d1a9594d9b42a3ee70e1ae1e23c5ac27d3a870360a9db"},"schema_version":"1.0","source":{"id":"2209.13085","kind":"arxiv","version":2}},"canonical_sha256":"832421faa7a7257e88e4c39eb45d8b7559c3214563b1a6ff8c01189bde9c8d77","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"832421faa7a7257e88e4c39eb45d8b7559c3214563b1a6ff8c01189bde9c8d77","first_computed_at":"2026-07-05T10:25:01.876864Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:25:01.876864Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"pJkwNXiwfEMhhjlhuyRg214BbHUUCBGkdldFq9LbZRfybUY5GLMAFlZ+m+7zb5CYWmE4o4G+/PJypzwlRLciDA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:25:01.877366Z","signed_message":"canonical_sha256_bytes"},"source_id":"2209.13085","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:18eea9d53a49da0ff3c2843dd4c39d8dec5839176746ae7ba10d6e681ef6533b","sha256:a686ad141ce8999fa9cd6f6412c88c90cb3a3197270cd3c8e9ec34a9417c0764"],"state_sha256":"ef1ab343fba16ef7c1aceb9d8dd1d594bd9658613feaf7b694e65b07aa5d80ca"}