{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.31689", "kind": "paper", "url": "https://arxiv.org/abs/2606.31689"}], "authors": ["Yufeng Lin", "Jialu Zhang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.31689", "source_url": "https://arxiv.org/abs/2606.31689"}, "categories": ["Code & Software", "General AI", "cs.SE"], "description": {"en": "ScratchWorld is an offline diagnostic benchmark for world models, using Scratch projects as executable worlds. It evaluates next-state prediction, long-horizon tracking, causal attribution, and counterfactual prediction with replay-verified transitions."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.31689", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.31689", "title": "ScratchWorld"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["University of Waterloo"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2606.31689", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2606.31689", "source_url": "https://openalex.org/W7166814293"}, "key": "claire-radar:2606.31689", "modality": null, "name": "ScratchWorld", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_scratchworld_04c08f86", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "4ccb71783a415c3744132331c190c8e0d7bee59093bc5710a9daec350b92764e", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.31689"}, "publisher": null, "released": "2026-06-30", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.31689", "source_url": "https://arxiv.org/abs/2606.31689"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-31689", "source": "claire_radar", "source_benchmark_id": "2606.31689", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Code & Software", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2606.31689", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "unavailable"}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "20ec2b91a1225ff4419b7bcc3e7d4df8f4a8ece6b5522b0356cb2d3390f43c93", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "ScratchWorld is an offline diagnostic benchmark for world models, using Scratch projects as executable worlds. It evaluates next-state prediction, long-horizon tracking, causal attribution, and counterfactual prediction with replay-verified transitions.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce ScratchWorld, an offline diagnostic benchmark that treats Scratch projects as executable worlds and uses a pinned Scratch VM to produce replay-verified transitions, hidden variables, causal traces, and counterfactual outcomes."}, "familyId": "bmf_0395c14a6789", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_scratchworld_04c08f86", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.31689", "paper": "https://arxiv.org/abs/2606.31689", "pdf": "https://arxiv.org/pdf/2606.31689", "project": null, "report": "https://arxiv.org/abs/2606.31689"}, "metrics": [], "motivation": "World-model evaluations often score a predicted future by overlap with a target state or observation.", "name": "ScratchWorld", "oneLine": "ScratchWorld is an offline diagnostic benchmark for world models, using Scratch projects as executable worlds. It evaluates next-state prediction, long-horizon tracking, causal attribution, and counterfactual prediction with replay-verified transitions.", "paperTitle": "ScratchWorld: Evaluating If World Models Compute Executable Consequences", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.31689", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-30", "source": {"authors": ["Yufeng Lin", "Jialu Zhang"], "categories": ["cs.SE"], "id": "2606.31689", "title": "ScratchWorld: Evaluating If World Models Compute Executable Consequences", "type": "arxiv", "url": "https://arxiv.org/abs/2606.31689"}, "sourceUpdatedAt": "2026-06-30T14:02:25Z", "topics": ["cs.SE"], "venueAttempts": [], "whyItMatters": "Provides a new evaluation paradigm for world models that avoids confounds like overlap with persistent state. Enables direct comparison of model capabilities on executable consequences."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
