{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HJZZBNXVRTCYCKBLBYWGH4ZQBY","short_pith_number":"pith:HJZZBNXV","schema_version":"1.0","canonical_sha256":"3a7390b6f58cc581282b0e2c63f3300e0eacee10f2bbe121ea062a8544ed540f","source":{"kind":"arxiv","id":"2606.24335","version":1},"attestation_state":"computed","paper":{"title":"Ill-Posed by Design: Probing Evidence Use in VLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Boaz Meivar, Eli Schwartz, Shai Avidan, Shaked Perek, Shani Shvartzman","submitted_at":"2026-06-23T09:20:09Z","abstract_excerpt":"Counterfactual analysis is widely used to study evidence use in vision-language models, but its diagnostic value is limited on well-posed tasks: when several cues independently support the same answer, removing one may not change the prediction. We propose monocular metric object-size estimation as an ill-posed diagnostic setting for evidence selection: because physical size cannot be determined from a single uncalibrated image, models must rely on imperfect cues category priors, target appearance, local context, apparent image size, and scene geometry. We assemble Metric VQA ($10{,}813$ dimen"},"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":"2606.24335","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-06-23T09:20:09Z","cross_cats_sorted":[],"title_canon_sha256":"697edfe2e86edd65fd3fec4906543a478ff6cf21e43d4979377b6138fcc94081","abstract_canon_sha256":"d1815a00db7d71c2dd2a1d84b4647440a0f1695a3b5268a346ba9e6cdf9585e6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-24T01:14:50.437727Z","signature_b64":"FM6L44lYQFueItQJA16SCWRj4gh9VIMOrtehxEcFUqVoaZTXbKQl1EmglQegkoJM1JTUdCF4VnrmuZFTXAtnAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3a7390b6f58cc581282b0e2c63f3300e0eacee10f2bbe121ea062a8544ed540f","last_reissued_at":"2026-06-24T01:14:50.437202Z","signature_status":"signed_v1","first_computed_at":"2026-06-24T01:14:50.437202Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Ill-Posed by Design: Probing Evidence Use in VLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Boaz Meivar, Eli Schwartz, Shai Avidan, Shaked Perek, Shani Shvartzman","submitted_at":"2026-06-23T09:20:09Z","abstract_excerpt":"Counterfactual analysis is widely used to study evidence use in vision-language models, but its diagnostic value is limited on well-posed tasks: when several cues independently support the same answer, removing one may not change the prediction. We propose monocular metric object-size estimation as an ill-posed diagnostic setting for evidence selection: because physical size cannot be determined from a single uncalibrated image, models must rely on imperfect cues category priors, target appearance, local context, apparent image size, and scene geometry. We assemble Metric VQA ($10{,}813$ dimen"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.24335","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/2606.24335/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":"2606.24335","created_at":"2026-06-24T01:14:50.437262+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.24335v1","created_at":"2026-06-24T01:14:50.437262+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.24335","created_at":"2026-06-24T01:14:50.437262+00:00"},{"alias_kind":"pith_short_12","alias_value":"HJZZBNXVRTCY","created_at":"2026-06-24T01:14:50.437262+00:00"},{"alias_kind":"pith_short_16","alias_value":"HJZZBNXVRTCYCKBL","created_at":"2026-06-24T01:14:50.437262+00:00"},{"alias_kind":"pith_short_8","alias_value":"HJZZBNXV","created_at":"2026-06-24T01:14:50.437262+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/HJZZBNXVRTCYCKBLBYWGH4ZQBY","json":"https://pith.science/pith/HJZZBNXVRTCYCKBLBYWGH4ZQBY.json","graph_json":"https://pith.science/api/pith-number/HJZZBNXVRTCYCKBLBYWGH4ZQBY/graph.json","events_json":"https://pith.science/api/pith-number/HJZZBNXVRTCYCKBLBYWGH4ZQBY/events.json","paper":"https://pith.science/paper/HJZZBNXV"},"agent_actions":{"view_html":"https://pith.science/pith/HJZZBNXVRTCYCKBLBYWGH4ZQBY","download_json":"https://pith.science/pith/HJZZBNXVRTCYCKBLBYWGH4ZQBY.json","view_paper":"https://pith.science/paper/HJZZBNXV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.24335&json=true","fetch_graph":"https://pith.science/api/pith-number/HJZZBNXVRTCYCKBLBYWGH4ZQBY/graph.json","fetch_events":"https://pith.science/api/pith-number/HJZZBNXVRTCYCKBLBYWGH4ZQBY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HJZZBNXVRTCYCKBLBYWGH4ZQBY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HJZZBNXVRTCYCKBLBYWGH4ZQBY/action/storage_attestation","attest_author":"https://pith.science/pith/HJZZBNXVRTCYCKBLBYWGH4ZQBY/action/author_attestation","sign_citation":"https://pith.science/pith/HJZZBNXVRTCYCKBLBYWGH4ZQBY/action/citation_signature","submit_replication":"https://pith.science/pith/HJZZBNXVRTCYCKBLBYWGH4ZQBY/action/replication_record"}},"created_at":"2026-06-24T01:14:50.437262+00:00","updated_at":"2026-06-24T01:14:50.437262+00:00"}