{ "schema_version": 1, "builder_plan": { "scenario": "A workflow needs specialist review, but extra agents may add more cost than useful work.", "change": "Test the proposed division of work against a single-agent implementation.", "baseline": "One agent with the same tools and an equal total token and time budget.", "steps": [ "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.", "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.", "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.", "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test." ], "measure": "Accepted outputs, coordination failures, wall-clock time, and total cost across all agents.", "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation." }, "builder_review": null, "arxiv_id": "2609.17464", "title": "Decomposition Buys Integrity, Not Yield", "summary": "The paper replaces the folklore rationale for splitting tasks across agent trees with a retention model showing each tier discards a fixed fraction of leaf findings, so it argues against default hierarchical decomposition rather than proposing a new method. Adoption of its guidance costs nothing in infrastructure but forgoes decomposition's real benefits: bounding root-context exposure to N^(1/k) items and lower billing at very large finding counts. What can fail is applying the population-level parameters (delta=0.34, C=0.571, mu=0.939) to a workload whose item granularity, tool boundaries, or brief quality differ, misjudging the rare cases where delegation genuinely pays.", "rationale": "The paper offers decision guidance, not an adoptable technique, and its central finding is that delegation is warranted in only a small minority of sessions, which a small team can apply directly by keeping agents flat unless finding counts are large.", "technique": "Analysis of multi-agent task decomposition yield", "family": "orquestracao_multiagente", "recommendation": "observar", "published": "2026-09-15", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2609.17464", "independent_implementations": 0, "total_implementations": 1, "stars": 1, "citations": 0, "claims": [ { "claim_id": "claim-01", "statement": "The paper replaces the folklore rationale for splitting tasks across agent trees with a retention model showing each tier discards a fixed fraction of leaf findings, so it argues against default hierarchical decomposition rather than proposing a new method. Adoption of its guidance costs nothing in infrastructure but forgoes decomposition's real benefits: bounding root-context exposure to N^(1/k) items and lower billing at very large finding counts. What can fail is applying the population-level parameters (delta=0.34, C=0.571, mu=0.939) to a workload whose item granularity, tool boundaries, or brief quality differ, misjudging the rare cases where delegation genuinely pays.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.17464", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "A depth-k decomposition yields C^k N^(1-delta) findings with measured delta=0.34 and per-tier C*mu=0.536; flat is optimal for yield, and at equal spend two tiers only overtake flat beyond 403 findings, with 0.7% to 11.3% of production sessions benefiting from delegation versus 7.8% that delegate.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.17464", "source_page": null, "source_excerpt": "" } ], "exposure_map": [ { "dimension": "quality", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compute", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "latency", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "operations", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compatibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "security", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "data_and_training", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "reproducibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" } ], "risks": [], "minimum_test": [], "open_questions": [], "independent_tests": [], "lab_reproductions": [], "report_available": false, "validation_tier": "unknown", "evidence_tier": "unknown", "infrastructure_basis": "unknown", "equations": [], "equations_status": "not_fetched", "core_kind": "", "equations_fetched_at": "", "equations_source_url": "", "equations_source_sha256": "", "exposures_reviewed_at": "", "exposures_source_sha256": "" }