SPB Git

spb/wp45_uqo

Public
32.7 KB · 1,533 lines json
1{2 "protocol": "heldout_v1",3 "generated": {4  "timestamp": "2026-11-04T00:51:23+00:00",5  "code_version": "760cf48-dirty",6  "host": {7   "platform": "macOS-27.0-arm64-arm-64bit",8   "python": "3.12.13",9   "machine": "arm64",10   "node": "MacBookPro-de-simon-pierre-2.local"11  }12 },13 "n_runs": 199,14 "n_runs_ok": 199,15 "n_judgments": 33,16 "hypotheses": {17  "H1_D_vs_B": {18   "clarity": {19    "n": 24,20    "mean_diff": -0.125,21    "ci95": [22     -0.312,23     0.03124    ],25    "wilcoxon_p": 0.17971249487899976,26    "cliffs_delta": -0.118,27    "holm_p": 0.7188,28    "significant": false29   },30   "argument_logic": {31    "n": 24,32    "mean_diff": 0.042,33    "ci95": [34     -0.156,35     0.21936    ],37    "wilcoxon_p": 0.5637028616507731,38    "cliffs_delta": 0.04,39    "holm_p": 1.0,40    "significant": false41   },42   "contribution_positioning": {43    "n": 24,44    "mean_diff": 0.0,45    "ci95": [46     -0.281,47     0.28148    ],49    "wilcoxon_p": 1.0,50    "cliffs_delta": 0.014,51    "holm_p": 1.0,52    "significant": false53   },54   "usefulness": {55    "n": 24,56    "mean_diff": 0.042,57    "ci95": [58     -0.281,59     0.40660    ],61    "wilcoxon_p": 0.7815112949987133,62    "cliffs_delta": 0.038,63    "holm_p": 1.0,64    "significant": false65   }66  },67  "H2_D_vs_C": {68   "clarity": {69    "n": 24,70    "mean_diff": -0.042,71    "ci95": [72     -0.094,73     0.074    ],75    "wilcoxon_p": 0.31731050786291415,76    "cliffs_delta": -0.042,77    "holm_p": 0.9519,78    "significant": false79   },80   "argument_logic": {81    "n": 24,82    "mean_diff": -0.125,83    "ci95": [84     -0.219,85     0.03186    ],87    "wilcoxon_p": 0.17971249487899976,88    "cliffs_delta": -0.12,89    "holm_p": 0.7188,90    "significant": false91   },92   "contribution_positioning": {93    "n": 24,94    "mean_diff": 0.125,95    "ci95": [96     -0.031,97     0.21998    ],99    "wilcoxon_p": 0.31731050786291415,100    "cliffs_delta": 0.132,101    "holm_p": 0.6346,102    "significant": false103   },104   "usefulness": {105    "n": 24,106    "mean_diff": 0.083,107    "ci95": [108     -0.062,109     0.344110    ],111    "wilcoxon_p": 0.47950012218695337,112    "cliffs_delta": 0.069,113    "holm_p": 0.4795,114    "significant": false115   }116  },117  "H3_E_vs_D": {118   "methodological_rigor": {119    "n": 24,120    "mean_diff": 0.958,121    "ci95": [122     0.438,123     1.469124    ],125    "wilcoxon_p": 0.0028308913160744967,126    "cliffs_delta": 0.55,127    "holm_p": 0.0085,128    "significant": true129   },130   "fidelity": {131    "n": 24,132    "mean_diff": 0.458,133    "ci95": [134     0.031,135     0.688136    ],137    "wilcoxon_p": 0.011831666473446543,138    "cliffs_delta": 0.323,139    "holm_p": 0.0237,140    "significant": true141   },142   "usefulness": {143    "n": 24,144    "mean_diff": -0.333,145    "ci95": [146     -0.781,147     0.062148    ],149    "wilcoxon_p": 0.06462747314963886,150    "cliffs_delta": -0.243,151    "holm_p": 0.0646,152    "significant": false153   }154  },155  "E_vs_A": {156   "clarity": {157    "n": 24,158    "mean_diff": 0.042,159    "ci95": [160     -0.25,161     0.688162    ],163    "wilcoxon_p": 0.7388826803635273,164    "cliffs_delta": -0.111,165    "holm_p": 0.7389,166    "significant": false167   },168   "argument_logic": {169    "n": 24,170    "mean_diff": 0.458,171    "ci95": [172     0.094,173     1.313174    ],175    "wilcoxon_p": 0.11558942362652867,176    "cliffs_delta": 0.245,177    "holm_p": 0.4624,178    "significant": false179   },180   "contribution_positioning": {181    "n": 24,182    "mean_diff": -0.083,183    "ci95": [184     -0.375,185     0.719186    ],187    "wilcoxon_p": 0.4828996977801915,188    "cliffs_delta": -0.137,189    "holm_p": 0.9658,190    "significant": false191   },192   "methodological_rigor": {193    "n": 24,194    "mean_diff": 1.333,195    "ci95": [196     0.844,197     2.282198    ],199    "wilcoxon_p": 0.0003879158284341555,200    "cliffs_delta": 0.651,201    "holm_p": 0.0023,202    "significant": true203   },204   "fidelity": {205    "n": 24,206    "mean_diff": 0.375,207    "ci95": [208     -0.031,209     1.125210    ],211    "wilcoxon_p": 0.0577795711235972,212    "cliffs_delta": 0.22,213    "holm_p": 0.2889,214    "significant": false215   },216   "usefulness": {217    "n": 24,218    "mean_diff": -0.167,219    "ci95": [220     -0.562,221     0.438222    ],223    "wilcoxon_p": 0.3142753351370735,224    "cliffs_delta": -0.215,225    "holm_p": 0.9428,226    "significant": false227   }228  },229  "E_vs_B": {230   "clarity": {231    "n": 24,232    "mean_diff": -0.208,233    "ci95": [234     -0.469,235     -0.062236    ],237    "wilcoxon_p": 0.13166801602281422,238    "cliffs_delta": -0.193,239    "holm_p": 0.1317,240    "significant": false241   },242   "argument_logic": {243    "n": 24,244    "mean_diff": 0.333,245    "ci95": [246     0.094,247     0.688248    ],249    "wilcoxon_p": 0.08288063685529753,250    "cliffs_delta": 0.271,251    "holm_p": 0.2486,252    "significant": false253   },254   "contribution_positioning": {255    "n": 24,256    "mean_diff": -0.417,257    "ci95": [258     -0.75,259     0.094260    ],261    "wilcoxon_p": 0.07260252961103252,262    "cliffs_delta": -0.29,263    "holm_p": 0.2904,264    "significant": false265   },266   "methodological_rigor": {267    "n": 24,268    "mean_diff": 1.083,269    "ci95": [270     0.562,271     1.812272    ],273    "wilcoxon_p": 0.0014260915957594058,274    "cliffs_delta": 0.545,275    "holm_p": 0.0086,276    "significant": true277   },278   "fidelity": {279    "n": 24,280    "mean_diff": 0.375,281    "ci95": [282     0.156,283     0.594284    ],285    "wilcoxon_p": 0.012554918596966537,286    "cliffs_delta": 0.299,287    "holm_p": 0.0628,288    "significant": false289   },290   "usefulness": {291    "n": 24,292    "mean_diff": -0.292,293    "ci95": [294     -0.657,295     0.031296    ],297    "wilcoxon_p": 0.11709281778662677,298    "cliffs_delta": -0.207,299    "holm_p": 0.2342,300    "significant": false301   }302  }303 },304 "ablations": {305  "E_no_retrieval": {306   "clarity": {307    "n": 16,308    "mean_diff": 0.0,309    "ci95": [310     -0.25,311     0.312312    ],313    "wilcoxon_p": 1.0,314    "cliffs_delta": -0.008315   },316   "argument_logic": {317    "n": 16,318    "mean_diff": 0.188,319    "ci95": [320     -0.125,321     0.5322    ],323    "wilcoxon_p": 0.25683925795785656,324    "cliffs_delta": 0.176325   },326   "contribution_positioning": {327    "n": 16,328    "mean_diff": -0.25,329    "ci95": [330     -0.562,331     0.062332    ],333    "wilcoxon_p": 0.15729920705028505,334    "cliffs_delta": -0.125335   },336   "methodological_rigor": {337    "n": 16,338    "mean_diff": -0.062,339    "ci95": [340     -0.625,341     0.438342    ],343    "wilcoxon_p": 0.9416269114492879,344    "cliffs_delta": -0.02345   },346   "fidelity": {347    "n": 16,348    "mean_diff": 0.062,349    "ci95": [350     -0.188,351     0.375352    ],353    "wilcoxon_p": 0.6547208460185769,354    "cliffs_delta": 0.082355   },356   "usefulness": {357    "n": 16,358    "mean_diff": -0.125,359    "ci95": [360     -0.562,361     0.312362    ],363    "wilcoxon_p": 0.5574801348628696,364    "cliffs_delta": -0.047365   }366  },367  "E_no_profile": {368   "clarity": {369    "n": 16,370    "mean_diff": -0.062,371    "ci95": [372     -0.25,373     0.125374    ],375    "wilcoxon_p": 0.5637028616507731,376    "cliffs_delta": -0.066377   },378   "argument_logic": {379    "n": 16,380    "mean_diff": 0.0,381    "ci95": [382     -0.188,383     0.188384    ],385    "wilcoxon_p": 1.0,386    "cliffs_delta": 0.0387   },388   "contribution_positioning": {389    "n": 16,390    "mean_diff": -0.188,391    "ci95": [392     -0.375,393     0.0394    ],395    "wilcoxon_p": 0.08326451666355039,396    "cliffs_delta": -0.113397   },398   "methodological_rigor": {399    "n": 16,400    "mean_diff": -0.25,401    "ci95": [402     -0.5,403     0.0404    ],405    "wilcoxon_p": 0.10247043485974945,406    "cliffs_delta": -0.148407   },408   "fidelity": {409    "n": 16,410    "mean_diff": 0.0,411    "ci95": [412     -0.25,413     0.25414    ],415    "wilcoxon_p": 1.0,416    "cliffs_delta": -0.043417   },418   "usefulness": {419    "n": 16,420    "mean_diff": -0.25,421    "ci95": [422     -0.438,423     -0.062424    ],425    "wilcoxon_p": 0.045500263896358396,426    "cliffs_delta": -0.176427   }428  },429  "E_no_referee": {430   "clarity": {431    "n": 16,432    "mean_diff": -0.125,433    "ci95": [434     -0.375,435     0.125436    ],437    "wilcoxon_p": 0.31731050786291415,438    "cliffs_delta": -0.125439   },440   "argument_logic": {441    "n": 16,442    "mean_diff": 0.5,443    "ci95": [444     0.062,445     0.938446    ],447    "wilcoxon_p": 0.045579154787925666,448    "cliffs_delta": 0.422449   },450   "contribution_positioning": {451    "n": 16,452    "mean_diff": 0.062,453    "ci95": [454     -0.375,455     0.438456    ],457    "wilcoxon_p": 0.763024600552995,458    "cliffs_delta": 0.094459   },460   "methodological_rigor": {461    "n": 16,462    "mean_diff": 1.25,463    "ci95": [464     0.625,465     1.875466    ],467    "wilcoxon_p": 0.003243891789291054,468    "cliffs_delta": 0.625469   },470   "fidelity": {471    "n": 16,472    "mean_diff": 0.562,473    "ci95": [474     0.125,475     1.125476    ],477    "wilcoxon_p": 0.04734500208525526,478    "cliffs_delta": 0.371479   },480   "usefulness": {481    "n": 16,482    "mean_diff": 0.125,483    "ci95": [484     -0.375,485     0.562486    ],487    "wilcoxon_p": 0.5929800980174267,488    "cliffs_delta": 0.156489   }490  },491  "E_no_factcheck": {492   "clarity": {493    "n": 16,494    "mean_diff": -0.125,495    "ci95": [496     -0.375,497     0.125498    ],499    "wilcoxon_p": 0.31731050786291415,500    "cliffs_delta": -0.117501   },502   "argument_logic": {503    "n": 16,504    "mean_diff": 0.062,505    "ci95": [506     -0.314,507     0.438508    ],509    "wilcoxon_p": 0.7388826803635273,510    "cliffs_delta": 0.059511   },512   "contribution_positioning": {513    "n": 16,514    "mean_diff": 0.0,515    "ci95": [516     -0.312,517     0.312518    ],519    "wilcoxon_p": 1.0,520    "cliffs_delta": 0.086521   },522   "methodological_rigor": {523    "n": 16,524    "mean_diff": -0.062,525    "ci95": [526     -0.562,527     0.375528    ],529    "wilcoxon_p": 1.0,530    "cliffs_delta": 0.004531   },532   "fidelity": {533    "n": 16,534    "mean_diff": -0.062,535    "ci95": [536     -0.375,537     0.25538    ],539    "wilcoxon_p": 0.7054569861112734,540    "cliffs_delta": -0.062541   },542   "usefulness": {543    "n": 16,544    "mean_diff": -0.188,545    "ci95": [546     -0.5,547     0.125548    ],549    "wilcoxon_p": 0.25683925795785656,550    "cliffs_delta": -0.098551   }552  }553 },554 "judge_means": {555  "A_baseline": {556   "clarity": 5.79,557   "argument_logic": 5.58,558   "contribution_positioning": 5.0,559   "methodological_rigor": 4.79,560   "fidelity": 5.33,561   "usefulness": 5.58,562   "n_invented_facts": 1.38,563   "n": 24564  },565  "B_strong_prompt": {566   "clarity": 6.04,567   "argument_logic": 5.71,568   "contribution_positioning": 5.33,569   "methodological_rigor": 5.04,570   "fidelity": 5.33,571   "usefulness": 5.71,572   "n_invented_facts": 1.62,573   "n": 24574  },575  "C_conventional_rag": {576   "clarity": 5.96,577   "argument_logic": 5.88,578   "contribution_positioning": 5.21,579   "methodological_rigor": 5.21,580   "fidelity": 5.25,581   "usefulness": 5.67,582   "n_invented_facts": 1.71,583   "n": 24584  },585  "D_profiles_retrieval": {586   "clarity": 5.92,587   "argument_logic": 5.75,588   "contribution_positioning": 5.33,589   "methodological_rigor": 5.17,590   "fidelity": 5.25,591   "usefulness": 5.75,592   "n_invented_facts": 1.88,593   "n": 24594  },595  "E_full_engine": {596   "clarity": 5.83,597   "argument_logic": 6.04,598   "contribution_positioning": 4.92,599   "methodological_rigor": 6.12,600   "fidelity": 5.71,601   "usefulness": 5.42,602   "n_invented_facts": 0.88,603   "n": 24604  },605  "E_no_factcheck": {606   "clarity": 6.0,607   "argument_logic": 6.06,608   "contribution_positioning": 4.94,609   "methodological_rigor": 6.19,610   "fidelity": 5.75,611   "usefulness": 5.62,612   "n_invented_facts": 0.88,613   "n": 16614  },615  "E_no_profile": {616   "clarity": 5.94,617   "argument_logic": 6.12,618   "contribution_positioning": 5.12,619   "methodological_rigor": 6.38,620   "fidelity": 5.69,621   "usefulness": 5.69,622   "n_invented_facts": 1.0,623   "n": 16624  },625  "E_no_referee": {626   "clarity": 6.0,627   "argument_logic": 5.62,628   "contribution_positioning": 4.88,629   "methodological_rigor": 4.88,630   "fidelity": 5.12,631   "usefulness": 5.31,632   "n_invented_facts": 1.81,633   "n": 16634  },635  "E_no_retrieval": {636   "clarity": 5.88,637   "argument_logic": 5.94,638   "contribution_positioning": 5.19,639   "methodological_rigor": 6.19,640   "fidelity": 5.62,641   "usefulness": 5.56,642   "n_invented_facts": 1.5,643   "n": 16644  }645 },646 "task_means": {647  "('draft_introduction', 'A_baseline')": {648   "clarity": 6.25,649   "argument_logic": 6.25,650   "contribution_positioning": 5.5,651   "methodological_rigor": 5.25,652   "fidelity": 5.5,653   "usefulness": 6.0654  },655  "('draft_introduction', 'B_strong_prompt')": {656   "clarity": 6.25,657   "argument_logic": 6.25,658   "contribution_positioning": 6.25,659   "methodological_rigor": 5.25,660   "fidelity": 5.75,661   "usefulness": 6.0662  },663  "('draft_introduction', 'C_conventional_rag')": {664   "clarity": 6.0,665   "argument_logic": 6.0,666   "contribution_positioning": 5.25,667   "methodological_rigor": 5.5,668   "fidelity": 5.25,669   "usefulness": 6.0670  },671  "('draft_introduction', 'D_profiles_retrieval')": {672   "clarity": 5.75,673   "argument_logic": 6.25,674   "contribution_positioning": 5.5,675   "methodological_rigor": 5.0,676   "fidelity": 4.75,677   "usefulness": 5.75678  },679  "('draft_introduction', 'E_full_engine')": {680   "clarity": 6.0,681   "argument_logic": 6.25,682   "contribution_positioning": 4.25,683   "methodological_rigor": 6.0,684   "fidelity": 6.0,685   "usefulness": 5.0686  },687  "('draft_introduction', 'E_no_factcheck')": {688   "clarity": 6.0,689   "argument_logic": 6.5,690   "contribution_positioning": 4.0,691   "methodological_rigor": 6.5,692   "fidelity": 6.0,693   "usefulness": 5.5694  },695  "('draft_introduction', 'E_no_profile')": {696   "clarity": 6.0,697   "argument_logic": 5.5,698   "contribution_positioning": 4.0,699   "methodological_rigor": 6.0,700   "fidelity": 6.0,701   "usefulness": 5.0702  },703  "('draft_introduction', 'E_no_referee')": {704   "clarity": 6.0,705   "argument_logic": 6.0,706   "contribution_positioning": 5.0,707   "methodological_rigor": 5.0,708   "fidelity": 4.0,709   "usefulness": 5.0710  },711  "('draft_introduction', 'E_no_retrieval')": {712   "clarity": 5.5,713   "argument_logic": 5.5,714   "contribution_positioning": 4.0,715   "methodological_rigor": 5.5,716   "fidelity": 6.0,717   "usefulness": 5.0718  },719  "('explain_specification', 'A_baseline')": {720   "clarity": 6.0,721   "argument_logic": 6.0,722   "contribution_positioning": 5.0,723   "methodological_rigor": 5.0,724   "fidelity": 5.5,725   "usefulness": 6.0726  },727  "('explain_specification', 'B_strong_prompt')": {728   "clarity": 6.0,729   "argument_logic": 6.0,730   "contribution_positioning": 5.5,731   "methodological_rigor": 5.5,732   "fidelity": 5.0,733   "usefulness": 6.0734  },735  "('explain_specification', 'C_conventional_rag')": {736   "clarity": 6.0,737   "argument_logic": 6.0,738   "contribution_positioning": 5.0,739   "methodological_rigor": 6.0,740   "fidelity": 5.0,741   "usefulness": 6.0742  },743  "('explain_specification', 'D_profiles_retrieval')": {744   "clarity": 6.0,745   "argument_logic": 6.0,746   "contribution_positioning": 5.0,747   "methodological_rigor": 6.0,748   "fidelity": 5.5,749   "usefulness": 6.5750  },751  "('explain_specification', 'E_full_engine')": {752   "clarity": 5.5,753   "argument_logic": 6.0,754   "contribution_positioning": 5.0,755   "methodological_rigor": 6.0,756   "fidelity": 5.0,757   "usefulness": 5.0758  },759  "('explain_specification', 'E_no_factcheck')": {760   "clarity": 5.5,761   "argument_logic": 6.0,762   "contribution_positioning": 5.0,763   "methodological_rigor": 6.0,764   "fidelity": 5.5,765   "usefulness": 5.0766  },767  "('explain_specification', 'E_no_profile')": {768   "clarity": 5.5,769   "argument_logic": 6.0,770   "contribution_positioning": 5.0,771   "methodological_rigor": 6.0,772   "fidelity": 5.5,773   "usefulness": 5.0774  },775  "('explain_specification', 'E_no_referee')": {776   "clarity": 6.0,777   "argument_logic": 6.0,778   "contribution_positioning": 5.0,779   "methodological_rigor": 5.5,780   "fidelity": 5.0,781   "usefulness": 5.5782  },783  "('explain_specification', 'E_no_retrieval')": {784   "clarity": 5.5,785   "argument_logic": 6.0,786   "contribution_positioning": 5.0,787   "methodological_rigor": 6.5,788   "fidelity": 5.5,789   "usefulness": 5.5790  },791  "('flag_causal_claims', 'A_baseline')": {792   "clarity": 3.5,793   "argument_logic": 3.5,794   "contribution_positioning": 3.0,795   "methodological_rigor": 3.5,796   "fidelity": 3.5,797   "usefulness": 3.5798  },799  "('flag_causal_claims', 'B_strong_prompt')": {800   "clarity": 6.0,801   "argument_logic": 6.0,802   "contribution_positioning": 5.0,803   "methodological_rigor": 5.5,804   "fidelity": 5.5,805   "usefulness": 5.0806  },807  "('flag_causal_claims', 'C_conventional_rag')": {808   "clarity": 6.0,809   "argument_logic": 6.0,810   "contribution_positioning": 5.0,811   "methodological_rigor": 5.5,812   "fidelity": 6.0,813   "usefulness": 5.0814  },815  "('flag_causal_claims', 'D_profiles_retrieval')": {816   "clarity": 6.0,817   "argument_logic": 6.0,818   "contribution_positioning": 5.0,819   "methodological_rigor": 6.0,820   "fidelity": 6.0,821   "usefulness": 5.5822  },823  "('flag_causal_claims', 'E_full_engine')": {824   "clarity": 5.5,825   "argument_logic": 6.0,826   "contribution_positioning": 5.0,827   "methodological_rigor": 6.5,828   "fidelity": 6.0,829   "usefulness": 4.5830  },831  "('flag_causal_claims', 'E_no_factcheck')": {832   "clarity": 6.0,833   "argument_logic": 6.0,834   "contribution_positioning": 5.0,835   "methodological_rigor": 7.0,836   "fidelity": 5.5,837   "usefulness": 5.0838  },839  "('flag_causal_claims', 'E_no_profile')": {840   "clarity": 6.0,841   "argument_logic": 6.0,842   "contribution_positioning": 5.0,843   "methodological_rigor": 7.0,844   "fidelity": 6.0,845   "usefulness": 5.5846  },847  "('flag_causal_claims', 'E_no_referee')": {848   "clarity": 6.0,849   "argument_logic": 5.5,850   "contribution_positioning": 4.5,851   "methodological_rigor": 5.5,852   "fidelity": 6.0,853   "usefulness": 4.5854  },855  "('flag_causal_claims', 'E_no_retrieval')": {856   "clarity": 6.0,857   "argument_logic": 6.0,858   "contribution_positioning": 5.5,859   "methodological_rigor": 6.5,860   "fidelity": 5.5,861   "usefulness": 5.5862  },863  "('improve_abstract', 'A_baseline')": {864   "clarity": 6.0,865   "argument_logic": 5.75,866   "contribution_positioning": 5.5,867   "methodological_rigor": 4.0,868   "fidelity": 6.0,869   "usefulness": 6.0870  },871  "('improve_abstract', 'B_strong_prompt')": {872   "clarity": 6.25,873   "argument_logic": 5.25,874   "contribution_positioning": 5.0,875   "methodological_rigor": 4.0,876   "fidelity": 6.0,877   "usefulness": 5.75878  },879  "('improve_abstract', 'C_conventional_rag')": {880   "clarity": 6.0,881   "argument_logic": 6.0,882   "contribution_positioning": 5.25,883   "methodological_rigor": 4.5,884   "fidelity": 6.25,885   "usefulness": 5.75886  },887  "('improve_abstract', 'D_profiles_retrieval')": {888   "clarity": 6.0,889   "argument_logic": 5.5,890   "contribution_positioning": 5.75,891   "methodological_rigor": 4.25,892   "fidelity": 6.0,893   "usefulness": 6.0894  },895  "('improve_abstract', 'E_full_engine')": {896   "clarity": 6.0,897   "argument_logic": 6.0,898   "contribution_positioning": 5.25,899   "methodological_rigor": 5.5,900   "fidelity": 6.0,901   "usefulness": 6.0902  },903  "('improve_abstract', 'E_no_factcheck')": {904   "clarity": 6.0,905   "argument_logic": 6.0,906   "contribution_positioning": 5.5,907   "methodological_rigor": 5.5,908   "fidelity": 5.5,909   "usefulness": 6.0910  },911  "('improve_abstract', 'E_no_profile')": {912   "clarity": 6.0,913   "argument_logic": 6.0,914   "contribution_positioning": 5.5,915   "methodological_rigor": 6.0,916   "fidelity": 6.0,917   "usefulness": 6.0918  },919  "('improve_abstract', 'E_no_referee')": {920   "clarity": 6.0,921   "argument_logic": 5.5,922   "contribution_positioning": 5.0,923   "methodological_rigor": 4.0,924   "fidelity": 6.0,925   "usefulness": 5.5926  },927  "('improve_abstract', 'E_no_retrieval')": {928   "clarity": 6.0,929   "argument_logic": 6.0,930   "contribution_positioning": 5.5,931   "methodological_rigor": 6.0,932   "fidelity": 6.0,933   "usefulness": 6.0934  },935  "('improve_table', 'A_baseline')": {936   "clarity": 5.5,937   "argument_logic": 5.0,938   "contribution_positioning": 4.0,939   "methodological_rigor": 4.0,940   "fidelity": 5.0,941   "usefulness": 5.5942  },943  "('improve_table', 'B_strong_prompt')": {944   "clarity": 6.0,945   "argument_logic": 5.0,946   "contribution_positioning": 4.0,947   "methodological_rigor": 4.0,948   "fidelity": 5.0,949   "usefulness": 6.0950  },951  "('improve_table', 'C_conventional_rag')": {952   "clarity": 6.0,953   "argument_logic": 5.0,954   "contribution_positioning": 4.0,955   "methodological_rigor": 4.0,956   "fidelity": 4.5,957   "usefulness": 6.0958  },959  "('improve_table', 'D_profiles_retrieval')": {960   "clarity": 6.0,961   "argument_logic": 5.0,962   "contribution_positioning": 4.0,963   "methodological_rigor": 4.5,964   "fidelity": 5.5,965   "usefulness": 6.0966  },967  "('improve_table', 'E_full_engine')": {968   "clarity": 5.5,969   "argument_logic": 6.0,970   "contribution_positioning": 4.5,971   "methodological_rigor": 6.0,972   "fidelity": 5.5,973   "usefulness": 5.5974  },975  "('improve_table', 'E_no_factcheck')": {976   "clarity": 6.0,977   "argument_logic": 5.5,978   "contribution_positioning": 4.0,979   "methodological_rigor": 5.5,980   "fidelity": 5.5,981   "usefulness": 6.0982  },983  "('improve_table', 'E_no_profile')": {984   "clarity": 6.0,985   "argument_logic": 6.0,986   "contribution_positioning": 5.0,987   "methodological_rigor": 5.5,988   "fidelity": 5.5,989   "usefulness": 6.0990  },991  "('improve_table', 'E_no_referee')": {992   "clarity": 6.0,993   "argument_logic": 5.0,994   "contribution_positioning": 4.0,995   "methodological_rigor": 4.5,996   "fidelity": 5.5,997   "usefulness": 6.0998  },999  "('improve_table', 'E_no_retrieval')": {1000   "clarity": 6.0,1001   "argument_logic": 6.0,1002   "contribution_positioning": 5.0,1003   "methodological_rigor": 6.5,1004   "fidelity": 5.0,1005   "usefulness": 6.01006  },1007  "('interpret_results', 'A_baseline')": {1008   "clarity": 6.0,1009   "argument_logic": 5.5,1010   "contribution_positioning": 5.0,1011   "methodological_rigor": 4.75,1012   "fidelity": 5.0,1013   "usefulness": 5.51014  },1015  "('interpret_results', 'B_strong_prompt')": {1016   "clarity": 5.75,1017   "argument_logic": 5.5,1018   "contribution_positioning": 5.0,1019   "methodological_rigor": 4.75,1020   "fidelity": 4.5,1021   "usefulness": 5.251022  },1023  "('interpret_results', 'C_conventional_rag')": {1024   "clarity": 6.0,1025   "argument_logic": 5.75,1026   "contribution_positioning": 5.0,1027   "methodological_rigor": 4.75,1028   "fidelity": 4.25,1029   "usefulness": 5.51030  },1031  "('interpret_results', 'D_profiles_retrieval')": {1032   "clarity": 6.0,1033   "argument_logic": 5.5,1034   "contribution_positioning": 5.25,1035   "methodological_rigor": 5.0,1036   "fidelity": 4.25,1037   "usefulness": 5.51038  },1039  "('interpret_results', 'E_full_engine')": {1040   "clarity": 5.75,1041   "argument_logic": 6.0,1042   "contribution_positioning": 4.75,1043   "methodological_rigor": 6.75,1044   "fidelity": 5.5,1045   "usefulness": 5.51046  },1047  "('interpret_results', 'E_no_factcheck')": {1048   "clarity": 6.0,1049   "argument_logic": 6.0,1050   "contribution_positioning": 5.0,1051   "methodological_rigor": 6.5,1052   "fidelity": 6.0,1053   "usefulness": 6.01054  },1055  "('interpret_results', 'E_no_profile')": {1056   "clarity": 6.0,1057   "argument_logic": 6.5,1058   "contribution_positioning": 5.0,1059   "methodological_rigor": 7.0,1060   "fidelity": 6.0,1061   "usefulness": 6.01062  },1063  "('interpret_results', 'E_no_referee')": {1064   "clarity": 6.0,1065   "argument_logic": 5.0,1066   "contribution_positioning": 5.0,1067   "methodological_rigor": 4.0,1068   "fidelity": 4.0,1069   "usefulness": 5.51070  },1071  "('interpret_results', 'E_no_retrieval')": {1072   "clarity": 6.0,1073   "argument_logic": 6.0,1074   "contribution_positioning": 5.0,1075   "methodological_rigor": 6.5,1076   "fidelity": 5.5,1077   "usefulness": 5.51078  },1079  "('referee_report', 'A_baseline')": {1080   "clarity": 6.0,1081   "argument_logic": 6.0,1082   "contribution_positioning": 5.0,1083   "methodological_rigor": 6.25,1084   "fidelity": 6.0,1085   "usefulness": 5.751086  },1087  "('referee_report', 'B_strong_prompt')": {1088   "clarity": 6.0,1089   "argument_logic": 6.0,1090   "contribution_positioning": 5.75,1091   "methodological_rigor": 6.75,1092   "fidelity": 5.75,1093   "usefulness": 6.01094  },1095  "('referee_report', 'C_conventional_rag')": {1096   "clarity": 6.0,1097   "argument_logic": 6.25,1098   "contribution_positioning": 5.75,1099   "methodological_rigor": 6.5,1100   "fidelity": 5.75,1101   "usefulness": 5.751102  },1103  "('referee_report', 'D_profiles_retrieval')": {1104   "clarity": 6.0,1105   "argument_logic": 6.0,1106   "contribution_positioning": 5.5,1107   "methodological_rigor": 6.25,1108   "fidelity": 5.75,1109   "usefulness": 5.751110  },1111  "('referee_report', 'E_full_engine')": {1112   "clarity": 6.0,1113   "argument_logic": 5.75,1114   "contribution_positioning": 5.0,1115   "methodological_rigor": 6.0,1116   "fidelity": 6.0,1117   "usefulness": 5.51118  },1119  "('referee_report', 'E_no_factcheck')": {1120   "clarity": 6.0,1121   "argument_logic": 6.0,1122   "contribution_positioning": 5.0,1123   "methodological_rigor": 6.0,1124   "fidelity": 6.0,1125   "usefulness": 5.51126  },1127  "('referee_report', 'E_no_profile')": {1128   "clarity": 6.0,1129   "argument_logic": 6.0,1130   "contribution_positioning": 5.0,1131   "methodological_rigor": 6.5,1132   "fidelity": 5.5,1133   "usefulness": 6.01134  },1135  "('referee_report', 'E_no_referee')": {1136   "clarity": 6.0,1137   "argument_logic": 6.5,1138   "contribution_positioning": 5.5,1139   "methodological_rigor": 6.5,1140   "fidelity": 5.5,1141   "usefulness": 5.51142  },1143  "('referee_report', 'E_no_retrieval')": {1144   "clarity": 6.0,1145   "argument_logic": 6.0,1146   "contribution_positioning": 5.5,1147   "methodological_rigor": 6.0,1148   "fidelity": 6.5,1149   "usefulness": 5.51150  },1151  "('rewrite_literature', 'A_baseline')": {1152   "clarity": 6.0,1153   "argument_logic": 5.5,1154   "contribution_positioning": 6.0,1155   "methodological_rigor": 4.5,1156   "fidelity": 5.0,1157   "usefulness": 5.51158  },1159  "('rewrite_literature', 'B_strong_prompt')": {1160   "clarity": 6.0,1161   "argument_logic": 5.5,1162   "contribution_positioning": 5.5,1163   "methodological_rigor": 4.0,1164   "fidelity": 4.5,1165   "usefulness": 5.51166  },1167  "('rewrite_literature', 'C_conventional_rag')": {1168   "clarity": 5.5,1169   "argument_logic": 5.5,1170   "contribution_positioning": 6.0,1171   "methodological_rigor": 4.5,1172   "fidelity": 4.5,1173   "usefulness": 5.01174  },1175  "('rewrite_literature', 'D_profiles_retrieval')": {1176   "clarity": 5.5,1177   "argument_logic": 5.5,1178   "contribution_positioning": 6.0,1179   "methodological_rigor": 4.5,1180   "fidelity": 4.5,1181   "usefulness": 5.01182  },1183  "('rewrite_literature', 'E_full_engine')": {1184   "clarity": 6.0,1185   "argument_logic": 6.5,1186   "contribution_positioning": 6.0,1187   "methodological_rigor": 6.5,1188   "fidelity": 5.0,1189   "usefulness": 6.01190  },1191  "('rewrite_literature', 'E_no_factcheck')": {1192   "clarity": 6.5,1193   "argument_logic": 6.5,1194   "contribution_positioning": 6.0,1195   "methodological_rigor": 6.5,1196   "fidelity": 6.0,1197   "usefulness": 6.01198  },1199  "('rewrite_literature', 'E_no_profile')": {1200   "clarity": 6.0,1201   "argument_logic": 7.0,1202   "contribution_positioning": 6.5,1203   "methodological_rigor": 7.0,1204   "fidelity": 5.0,1205   "usefulness": 6.01206  },1207  "('rewrite_literature', 'E_no_referee')": {1208   "clarity": 6.0,1209   "argument_logic": 5.5,1210   "contribution_positioning": 5.0,1211   "methodological_rigor": 4.0,1212   "fidelity": 5.0,1213   "usefulness": 5.01214  },1215  "('rewrite_literature', 'E_no_retrieval')": {1216   "clarity": 6.0,1217   "argument_logic": 6.0,1218   "contribution_positioning": 6.0,1219   "methodological_rigor": 6.0,1220   "fidelity": 5.0,1221   "usefulness": 5.51222  }1223 },1224 "out_of_domain_means": {1225  "A_baseline": {1226   "clarity": 6.0,1227   "argument_logic": 6.0,1228   "contribution_positioning": 5.67,1229   "methodological_rigor": 5.67,1230   "fidelity": 5.0,1231   "usefulness": 5.331232  },1233  "B_strong_prompt": {1234   "clarity": 6.0,1235   "argument_logic": 6.0,1236   "contribution_positioning": 6.0,1237   "methodological_rigor": 5.33,1238   "fidelity": 4.67,1239   "usefulness": 5.331240  },1241  "C_conventional_rag": {1242   "clarity": 6.0,1243   "argument_logic": 5.67,1244   "contribution_positioning": 5.67,1245   "methodological_rigor": 5.0,1246   "fidelity": 4.33,1247   "usefulness": 5.331248  },1249  "D_profiles_retrieval": {1250   "clarity": 5.33,1251   "argument_logic": 5.67,1252   "contribution_positioning": 5.33,1253   "methodological_rigor": 5.67,1254   "fidelity": 4.67,1255   "usefulness": 4.671256  },1257  "E_full_engine": {1258   "clarity": 5.67,1259   "argument_logic": 6.0,1260   "contribution_positioning": 5.33,1261   "methodological_rigor": 5.67,1262   "fidelity": 4.33,1263   "usefulness": 5.01264  }1265 },1266 "judge_consistency": {1267  "n_pairs": 50,1268  "clarity": {1269   "mean_abs_diff": 0.26,1270   "spearman": 0.4111271  },1272  "argument_logic": {1273   "mean_abs_diff": 0.3,1274   "spearman": 0.5821275  },1276  "contribution_positioning": {1277   "mean_abs_diff": 0.32,1278   "spearman": 0.841279  },1280  "methodological_rigor": {1281   "mean_abs_diff": 0.34,1282   "spearman": 0.8041283  },1284  "fidelity": {1285   "mean_abs_diff": 0.3,1286   "spearman": 0.7471287  },1288  "usefulness": {1289   "mean_abs_diff": 0.4,1290   "spearman": 0.5311291  }1292 },1293 "secondary_means": {1294  "A_baseline": {1295   "preserved_rate": 0.053,1296   "new_numbers": 1.0,1297   "n_unverifiable_citations": 0.083,1298   "n_unsupported_claims": 1.083,1299   "max_overlap_share": 0.073,1300   "todo_placeholders": 2.542,1301   "n_words": 959.333,1302   "cost_usd": 0.138,1303   "latency_s": 45.879,1304   "n": 241305  },1306  "B_strong_prompt": {1307   "preserved_rate": 0.057,1308   "new_numbers": 1.042,1309   "n_unverifiable_citations": 0.083,1310   "n_unsupported_claims": 1.042,1311   "max_overlap_share": 0.065,1312   "todo_placeholders": 2.958,1313   "n_words": 1027.417,1314   "cost_usd": 0.139,1315   "latency_s": 46.083,1316   "n": 241317  },1318  "C_conventional_rag": {1319   "preserved_rate": 0.053,1320   "new_numbers": 1.25,1321   "n_unverifiable_citations": 0.083,1322   "n_unsupported_claims": 1.625,1323   "max_overlap_share": 0.072,1324   "todo_placeholders": 2.167,1325   "n_words": 1059.875,1326   "cost_usd": 0.146,1327   "latency_s": 47.446,1328   "n": 241329  },1330  "D_profiles_retrieval": {1331   "preserved_rate": 0.055,1332   "new_numbers": 1.167,1333   "n_unverifiable_citations": 0.083,1334   "n_unsupported_claims": 1.208,1335   "max_overlap_share": 0.068,1336   "todo_placeholders": 2.667,1337   "n_words": 1051.875,1338   "cost_usd": 0.154,1339   "latency_s": 49.475,1340   "n": 241341  },1342  "E_full_engine": {1343   "preserved_rate": 0.055,1344   "new_numbers": 0.833,1345   "n_unverifiable_citations": 0.0,1346   "n_unsupported_claims": 0.625,1347   "max_overlap_share": 0.059,1348   "todo_placeholders": 3.125,1349   "n_words": 994.667,1350   "cost_usd": 0.316,1351   "latency_s": 98.033,1352   "n": 241353  },1354  "E_no_factcheck": {1355   "preserved_rate": 0.048,1356   "new_numbers": 0.938,1357   "n_unverifiable_citations": 0.062,1358   "n_unsupported_claims": 0.75,1359   "max_overlap_share": 0.057,1360   "todo_placeholders": 2.562,1361   "n_words": 1049.938,1362   "cost_usd": 0.312,1363   "latency_s": 95.925,1364   "n": 161365  },1366  "E_no_profile": {1367   "preserved_rate": 0.054,1368   "new_numbers": 0.688,1369   "n_unverifiable_citations": 0.0,1370   "n_unsupported_claims": 0.312,1371   "max_overlap_share": 0.06,1372   "todo_placeholders": 2.75,1373   "n_words": 996.188,1374   "cost_usd": 0.311,1375   "latency_s": 97.706,1376   "n": 161377  },1378  "E_no_referee": {1379   "preserved_rate": 0.044,1380   "new_numbers": 0.875,1381   "n_unverifiable_citations": 0.125,1382   "n_unsupported_claims": 1.062,1383   "max_overlap_share": 0.076,1384   "todo_placeholders": 2.25,1385   "n_words": 939.188,1386   "cost_usd": 0.155,1387   "latency_s": 48.856,1388   "n": 161389  },1390  "E_no_retrieval": {1391   "preserved_rate": 0.042,1392   "new_numbers": 0.625,1393   "n_unverifiable_citations": 0.125,1394   "n_unsupported_claims": 0.312,1395   "max_overlap_share": 0.055,1396   "todo_placeholders": 5.062,1397   "n_words": 1074.375,1398   "cost_usd": 0.308,1399   "latency_s": 94.531,1400   "n": 161401  }1402 },1403 "H4": {1404  "E_vs_A": {1405   "n_unverifiable_citations": {1406    "n": 24,1407    "mean_diff": -0.083,1408    "ci95": [1409     -0.25,1410     0.01411    ],1412    "wilcoxon_p": 0.15729920705028505,1413    "cliffs_delta": -0.0831414   },1415   "n_unsupported_claims": {1416    "n": 24,1417    "mean_diff": -0.458,1418    "ci95": [1419     -0.906,1420     0.251421    ],1422    "wilcoxon_p": 0.1005626399172922,1423    "cliffs_delta": -0.1631424   },1425   "new_numbers": {1426    "n": 24,1427    "mean_diff": -0.167,1428    "ci95": [1429     -0.375,1430     0.1251431    ],1432    "wilcoxon_p": 0.15729920705028516,1433    "cliffs_delta": -0.1011434   }1435  },1436  "E_vs_B": {1437   "n_unverifiable_citations": {1438    "n": 24,1439    "mean_diff": -0.083,1440    "ci95": [1441     -0.156,1442     0.01443    ],1444    "wilcoxon_p": 0.15729920705028505,1445    "cliffs_delta": -0.0831446   },1447   "n_unsupported_claims": {1448    "n": 24,1449    "mean_diff": -0.417,1450    "ci95": [1451     -0.938,1452     0.1881453    ],1454    "wilcoxon_p": 0.17298970944016634,1455    "cliffs_delta": -0.1811456   },1457   "new_numbers": {1458    "n": 24,1459    "mean_diff": -0.208,1460    "ci95": [1461     -0.406,1462     0.0621463    ],1464    "wilcoxon_p": 0.09558070454562939,1465    "cliffs_delta": -0.1411466   }1467  },1468  "E_vs_C": {1469   "n_unverifiable_citations": {1470    "n": 24,1471    "mean_diff": -0.083,1472    "ci95": [1473     -0.312,1474     0.01475    ],1476    "wilcoxon_p": 0.15729920705028505,1477    "cliffs_delta": -0.0831478   },1479   "n_unsupported_claims": {1480    "n": 24,1481    "mean_diff": -1.0,1482    "ci95": [1483     -1.594,1484     -0.1561485    ],1486    "wilcoxon_p": 0.008554328993751146,1487    "cliffs_delta": -0.3661488   },1489   "new_numbers": {1490    "n": 24,1491    "mean_diff": -0.417,1492    "ci95": [1493     -1.0,1494     0.1251495    ],1496    "wilcoxon_p": 0.31731050786291415,1497    "cliffs_delta": -0.0821498   }1499  }1500 },1501 "H5_overlap": {1502  "preregistered_metric": {1503   "max_overlap_share_overall": 0.4044,1504   "share_outputs_below_0.02": 0.418478260869565241505  },1506  "excluding_source_document": {1507   "max_other_share": 0.0069,1508   "share_outputs_below_0.02": 1.0,1509   "mean_other_share": 0.00079239130434782611510  },1511  "self_overlap_by_task": {1512   "draft_introduction": 0.028,1513   "explain_specification": 0.019,1514   "flag_causal_claims": 0.073,1515   "improve_abstract": 0.253,1516   "improve_table": 0.02,1517   "interpret_results": 0.018,1518   "referee_report": 0.008,1519   "rewrite_literature": 0.0821520  },1521  "self_overlap_by_condition": {1522   "A_baseline": 0.073,1523   "B_strong_prompt": 0.065,1524   "C_conventional_rag": 0.072,1525   "D_profiles_retrieval": 0.068,1526   "E_full_engine": 0.059,1527   "E_no_factcheck": 0.057,1528   "E_no_profile": 0.06,1529   "E_no_referee": 0.076,1530   "E_no_retrieval": 0.0551531  }1532 }1533}
Files · dc3418435bdd484c33eda3ee50716a6a65d91c31
Menu

Theme