SPB Git

spb/wp45_uqo

Public
32.6 KB · 1,533 lines json
1{2 "protocol": "heldout_v1",3 "generated": {4  "timestamp": "2026-11-04T00:02:26+00:00",5  "code_version": "d1aa8dc-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": 0.042,21    "ci95": [22     -0.406,23     0.28124    ],25    "wilcoxon_p": 0.7920689965469446,26    "cliffs_delta": 0.014,27    "holm_p": 1.0,28    "significant": false29   },30   "argument_logic": {31    "n": 24,32    "mean_diff": 0.333,33    "ci95": [34     -0.219,35     0.65636    ],37    "wilcoxon_p": 0.1120871657779208,38    "cliffs_delta": 0.203,39    "holm_p": 0.4483,40    "significant": false41   },42   "contribution_positioning": {43    "n": 24,44    "mean_diff": 0.083,45    "ci95": [46     -0.406,47     0.43848    ],49    "wilcoxon_p": 0.9665165964249498,50    "cliffs_delta": 0.023,51    "holm_p": 0.9665,52    "significant": false53   },54   "usefulness": {55    "n": 24,56    "mean_diff": 0.167,57    "ci95": [58     -0.5,59     0.68860    ],61    "wilcoxon_p": 0.5747134367649175,62    "cliffs_delta": 0.007,63    "holm_p": 1.0,64    "significant": false65   }66  },67  "H2_D_vs_C": {68   "clarity": {69    "n": 24,70    "mean_diff": 0.0,71    "ci95": [72     -0.281,73     0.2574    ],75    "wilcoxon_p": 1.0,76    "cliffs_delta": 0.005,77    "holm_p": 1.0,78    "significant": false79   },80   "argument_logic": {81    "n": 24,82    "mean_diff": 0.083,83    "ci95": [84     -0.438,85     0.46986    ],87    "wilcoxon_p": 0.7506546409575028,88    "cliffs_delta": 0.0,89    "holm_p": 1.0,90    "significant": false91   },92   "contribution_positioning": {93    "n": 24,94    "mean_diff": -0.208,95    "ci95": [96     -0.719,97     0.15698    ],99    "wilcoxon_p": 0.15271894732542726,100    "cliffs_delta": -0.13,101    "holm_p": 0.6109,102    "significant": false103   },104   "usefulness": {105    "n": 24,106    "mean_diff": 0.042,107    "ci95": [108     -0.5,109     0.438110    ],111    "wilcoxon_p": 0.9426906297302409,112    "cliffs_delta": -0.068,113    "holm_p": 1.0,114    "significant": false115   }116  },117  "H3_E_vs_D": {118   "methodological_rigor": {119    "n": 24,120    "mean_diff": 1.333,121    "ci95": [122     1.0,123     1.845124    ],125    "wilcoxon_p": 9.280186701164916e-05,126    "cliffs_delta": 0.677,127    "holm_p": 0.0003,128    "significant": true129   },130   "fidelity": {131    "n": 24,132    "mean_diff": 0.625,133    "ci95": [134     0.312,135     1.031136    ],137    "wilcoxon_p": 0.0019691496770990976,138    "cliffs_delta": 0.359,139    "holm_p": 0.0039,140    "significant": true141   },142   "usefulness": {143    "n": 24,144    "mean_diff": 0.25,145    "ci95": [146     -0.031,147     0.656148    ],149    "wilcoxon_p": 0.10880943004054569,150    "cliffs_delta": 0.167,151    "holm_p": 0.1088,152    "significant": false153   }154  },155  "E_vs_A": {156   "clarity": {157    "n": 24,158    "mean_diff": -0.042,159    "ci95": [160     -0.406,161     0.281162    ],163    "wilcoxon_p": 0.7815112949987133,164    "cliffs_delta": -0.059,165    "holm_p": 0.7815,166    "significant": false167   },168   "argument_logic": {169    "n": 24,170    "mean_diff": 0.917,171    "ci95": [172     0.406,173     1.438174    ],175    "wilcoxon_p": 0.0008241896037559798,176    "cliffs_delta": 0.516,177    "holm_p": 0.0041,178    "significant": true179   },180   "contribution_positioning": {181    "n": 24,182    "mean_diff": 0.5,183    "ci95": [184     0.125,185     0.907186    ],187    "wilcoxon_p": 0.018735985158564714,188    "cliffs_delta": 0.198,189    "holm_p": 0.0562,190    "significant": false191   },192   "methodological_rigor": {193    "n": 24,194    "mean_diff": 2.042,195    "ci95": [196     1.438,197     2.625198    ],199    "wilcoxon_p": 4.6023226073469e-05,200    "cliffs_delta": 0.847,201    "holm_p": 0.0003,202    "significant": true203   },204   "fidelity": {205    "n": 24,206    "mean_diff": 0.417,207    "ci95": [208     0.156,209     0.781210    ],211    "wilcoxon_p": 0.012419330651552268,212    "cliffs_delta": 0.281,213    "holm_p": 0.0497,214    "significant": true215   },216   "usefulness": {217    "n": 24,218    "mean_diff": 0.458,219    "ci95": [220     -0.094,221     1.031222    ],223    "wilcoxon_p": 0.08329941488171003,224    "cliffs_delta": 0.163,225    "holm_p": 0.1666,226    "significant": false227   }228  },229  "E_vs_B": {230   "clarity": {231    "n": 24,232    "mean_diff": -0.042,233    "ci95": [234     -0.406,235     0.25236    ],237    "wilcoxon_p": 0.8127423725657857,238    "cliffs_delta": -0.05,239    "holm_p": 0.8127,240    "significant": false241   },242   "argument_logic": {243    "n": 24,244    "mean_diff": 0.792,245    "ci95": [246     0.406,247     1.062248    ],249    "wilcoxon_p": 0.0002901870694228414,250    "cliffs_delta": 0.483,251    "holm_p": 0.0015,252    "significant": true253   },254   "contribution_positioning": {255    "n": 24,256    "mean_diff": 0.25,257    "ci95": [258     -0.125,259     0.688260    ],261    "wilcoxon_p": 0.2720394478957354,262    "cliffs_delta": 0.089,263    "holm_p": 0.5441,264    "significant": false265   },266   "methodological_rigor": {267    "n": 24,268    "mean_diff": 1.667,269    "ci95": [270     1.125,271     2.188272    ],273    "wilcoxon_p": 8.456524858492154e-05,274    "cliffs_delta": 0.741,275    "holm_p": 0.0005,276    "significant": true277   },278   "fidelity": {279    "n": 24,280    "mean_diff": 1.042,281    "ci95": [282     0.781,283     1.438284    ],285    "wilcoxon_p": 0.0003061844202490992,286    "cliffs_delta": 0.573,287    "holm_p": 0.0012,288    "significant": true289   },290   "usefulness": {291    "n": 24,292    "mean_diff": 0.417,293    "ci95": [294     -0.125,295     0.969296    ],297    "wilcoxon_p": 0.11972288596282951,298    "cliffs_delta": 0.149,299    "holm_p": 0.3592,300    "significant": false301   }302  }303 },304 "ablations": {305  "E_no_retrieval": {306   "clarity": {307    "n": 16,308    "mean_diff": -0.125,309    "ci95": [310     -0.375,311     0.125312    ],313    "wilcoxon_p": 0.31731050786291415,314    "cliffs_delta": -0.117315   },316   "argument_logic": {317    "n": 16,318    "mean_diff": 0.188,319    "ci95": [320     -0.062,321     0.438322    ],323    "wilcoxon_p": 0.17971249487899976,324    "cliffs_delta": 0.141325   },326   "contribution_positioning": {327    "n": 16,328    "mean_diff": -0.062,329    "ci95": [330     -0.25,331     0.125332    ],333    "wilcoxon_p": 0.5637028616507731,334    "cliffs_delta": -0.047335   },336   "methodological_rigor": {337    "n": 16,338    "mean_diff": 0.125,339    "ci95": [340     -0.125,341     0.375342    ],343    "wilcoxon_p": 0.4142161782425253,344    "cliffs_delta": 0.109345   },346   "fidelity": {347    "n": 16,348    "mean_diff": 0.062,349    "ci95": [350     -0.312,351     0.438352    ],353    "wilcoxon_p": 0.7388826803635273,354    "cliffs_delta": 0.051355   },356   "usefulness": {357    "n": 16,358    "mean_diff": 0.0,359    "ci95": [360     -0.25,361     0.25362    ],363    "wilcoxon_p": 1.0,364    "cliffs_delta": 0.02365   }366  },367  "E_no_profile": {368   "clarity": {369    "n": 16,370    "mean_diff": -0.188,371    "ci95": [372     -0.562,373     0.125374    ],375    "wilcoxon_p": 0.31731050786291415,376    "cliffs_delta": -0.121377   },378   "argument_logic": {379    "n": 16,380    "mean_diff": 0.062,381    "ci95": [382     -0.25,383     0.375384    ],385    "wilcoxon_p": 0.7054569861112734,386    "cliffs_delta": 0.008387   },388   "contribution_positioning": {389    "n": 16,390    "mean_diff": -0.125,391    "ci95": [392     -0.438,393     0.188394    ],395    "wilcoxon_p": 0.4142161782425253,396    "cliffs_delta": -0.07397   },398   "methodological_rigor": {399    "n": 16,400    "mean_diff": 0.062,401    "ci95": [402     -0.312,403     0.5404    ],405    "wilcoxon_p": 0.763024600552995,406    "cliffs_delta": 0.02407   },408   "fidelity": {409    "n": 16,410    "mean_diff": 0.0,411    "ci95": [412     -0.438,413     0.438414    ],415    "wilcoxon_p": 1.0,416    "cliffs_delta": 0.008417   },418   "usefulness": {419    "n": 16,420    "mean_diff": -0.062,421    "ci95": [422     -0.312,423     0.188424    ],425    "wilcoxon_p": 0.6547208460185769,426    "cliffs_delta": -0.031427   }428  },429  "E_no_referee": {430   "clarity": {431    "n": 16,432    "mean_diff": -0.125,433    "ci95": [434     -0.438,435     0.188436    ],437    "wilcoxon_p": 0.4142161782425253,438    "cliffs_delta": -0.062439   },440   "argument_logic": {441    "n": 16,442    "mean_diff": 0.688,443    "ci95": [444     0.312,445     1.125446    ],447    "wilcoxon_p": 0.00838632096179826,448    "cliffs_delta": 0.43449   },450   "contribution_positioning": {451    "n": 16,452    "mean_diff": 0.25,453    "ci95": [454     -0.062,455     0.562456    ],457    "wilcoxon_p": 0.15729920705028516,458    "cliffs_delta": 0.09459   },460   "methodological_rigor": {461    "n": 16,462    "mean_diff": 1.625,463    "ci95": [464     1.0,465     2.188466    ],467    "wilcoxon_p": 0.0011302166601564717,468    "cliffs_delta": 0.781469   },470   "fidelity": {471    "n": 16,472    "mean_diff": 0.5,473    "ci95": [474     0.125,475     0.814476    ],477    "wilcoxon_p": 0.02092133533779403,478    "cliffs_delta": 0.289479   },480   "usefulness": {481    "n": 16,482    "mean_diff": 0.438,483    "ci95": [484     0.0,485     0.938486    ],487    "wilcoxon_p": 0.08326451666355039,488    "cliffs_delta": 0.246489   }490  },491  "E_no_factcheck": {492   "clarity": {493    "n": 16,494    "mean_diff": 0.0,495    "ci95": [496     -0.438,497     0.438498    ],499    "wilcoxon_p": 1.0,500    "cliffs_delta": 0.027501   },502   "argument_logic": {503    "n": 16,504    "mean_diff": 0.125,505    "ci95": [506     -0.25,507     0.438508    ],509    "wilcoxon_p": 0.47950012218695337,510    "cliffs_delta": 0.094511   },512   "contribution_positioning": {513    "n": 16,514    "mean_diff": 0.125,515    "ci95": [516     -0.125,517     0.375518    ],519    "wilcoxon_p": 0.31731050786291415,520    "cliffs_delta": 0.066521   },522   "methodological_rigor": {523    "n": 16,524    "mean_diff": 0.25,525    "ci95": [526     0.0,527     0.5528    ],529    "wilcoxon_p": 0.10247043485974945,530    "cliffs_delta": 0.113531   },532   "fidelity": {533    "n": 16,534    "mean_diff": 0.0,535    "ci95": [536     -0.25,537     0.25538    ],539    "wilcoxon_p": 1.0,540    "cliffs_delta": -0.008541   },542   "usefulness": {543    "n": 16,544    "mean_diff": 0.125,545    "ci95": [546     -0.188,547     0.438548    ],549    "wilcoxon_p": 0.47950012218695337,550    "cliffs_delta": 0.102551   }552  }553 },554 "judge_means": {555  "A_baseline": {556   "clarity": 5.67,557   "argument_logic": 5.04,558   "contribution_positioning": 4.12,559   "methodological_rigor": 4.17,560   "fidelity": 5.83,561   "usefulness": 4.83,562   "n_invented_facts": 0.79,563   "n": 24564  },565  "B_strong_prompt": {566   "clarity": 5.67,567   "argument_logic": 5.17,568   "contribution_positioning": 4.38,569   "methodological_rigor": 4.54,570   "fidelity": 5.21,571   "usefulness": 4.88,572   "n_invented_facts": 1.54,573   "n": 24574  },575  "C_conventional_rag": {576   "clarity": 5.71,577   "argument_logic": 5.42,578   "contribution_positioning": 4.67,579   "methodological_rigor": 4.58,580   "fidelity": 5.29,581   "usefulness": 5.0,582   "n_invented_facts": 1.33,583   "n": 24584  },585  "D_profiles_retrieval": {586   "clarity": 5.71,587   "argument_logic": 5.5,588   "contribution_positioning": 4.46,589   "methodological_rigor": 4.88,590   "fidelity": 5.62,591   "usefulness": 5.04,592   "n_invented_facts": 1.08,593   "n": 24594  },595  "E_full_engine": {596   "clarity": 5.62,597   "argument_logic": 5.96,598   "contribution_positioning": 4.62,599   "methodological_rigor": 6.21,600   "fidelity": 6.25,601   "usefulness": 5.29,602   "n_invented_facts": 0.42,603   "n": 24604  },605  "E_no_factcheck": {606   "clarity": 5.56,607   "argument_logic": 5.81,608   "contribution_positioning": 4.44,609   "methodological_rigor": 5.94,610   "fidelity": 6.12,611   "usefulness": 5.19,612   "n_invented_facts": 0.62,613   "n": 16614  },615  "E_no_profile": {616   "clarity": 5.75,617   "argument_logic": 5.88,618   "contribution_positioning": 4.69,619   "methodological_rigor": 6.12,620   "fidelity": 6.12,621   "usefulness": 5.38,622   "n_invented_facts": 0.69,623   "n": 16624  },625  "E_no_referee": {626   "clarity": 5.69,627   "argument_logic": 5.25,628   "contribution_positioning": 4.31,629   "methodological_rigor": 4.56,630   "fidelity": 5.62,631   "usefulness": 4.88,632   "n_invented_facts": 1.19,633   "n": 16634  },635  "E_no_retrieval": {636   "clarity": 5.69,637   "argument_logic": 5.75,638   "contribution_positioning": 4.62,639   "methodological_rigor": 6.06,640   "fidelity": 6.06,641   "usefulness": 5.31,642   "n_invented_facts": 0.69,643   "n": 16644  }645 },646 "task_means": {647  "('draft_introduction', 'A_baseline')": {648   "clarity": 6.25,649   "argument_logic": 5.75,650   "contribution_positioning": 5.0,651   "methodological_rigor": 4.25,652   "fidelity": 5.5,653   "usefulness": 5.75654  },655  "('draft_introduction', 'B_strong_prompt')": {656   "clarity": 6.0,657   "argument_logic": 5.5,658   "contribution_positioning": 5.0,659   "methodological_rigor": 4.25,660   "fidelity": 5.25,661   "usefulness": 5.5662  },663  "('draft_introduction', 'C_conventional_rag')": {664   "clarity": 6.0,665   "argument_logic": 5.75,666   "contribution_positioning": 5.5,667   "methodological_rigor": 4.75,668   "fidelity": 5.0,669   "usefulness": 5.5670  },671  "('draft_introduction', 'D_profiles_retrieval')": {672   "clarity": 6.0,673   "argument_logic": 6.0,674   "contribution_positioning": 5.0,675   "methodological_rigor": 5.25,676   "fidelity": 4.75,677   "usefulness": 5.5678  },679  "('draft_introduction', 'E_full_engine')": {680   "clarity": 5.75,681   "argument_logic": 6.5,682   "contribution_positioning": 5.0,683   "methodological_rigor": 6.5,684   "fidelity": 6.0,685   "usefulness": 6.0686  },687  "('draft_introduction', 'E_no_factcheck')": {688   "clarity": 5.5,689   "argument_logic": 5.5,690   "contribution_positioning": 5.0,691   "methodological_rigor": 6.0,692   "fidelity": 5.5,693   "usefulness": 5.5694  },695  "('draft_introduction', 'E_no_profile')": {696   "clarity": 5.5,697   "argument_logic": 6.0,698   "contribution_positioning": 4.5,699   "methodological_rigor": 6.0,700   "fidelity": 6.5,701   "usefulness": 5.5702  },703  "('draft_introduction', 'E_no_referee')": {704   "clarity": 6.0,705   "argument_logic": 6.5,706   "contribution_positioning": 5.0,707   "methodological_rigor": 5.0,708   "fidelity": 4.5,709   "usefulness": 5.0710  },711  "('draft_introduction', 'E_no_retrieval')": {712   "clarity": 6.0,713   "argument_logic": 6.0,714   "contribution_positioning": 5.0,715   "methodological_rigor": 7.0,716   "fidelity": 5.5,717   "usefulness": 6.0718  },719  "('explain_specification', 'A_baseline')": {720   "clarity": 6.0,721   "argument_logic": 5.0,722   "contribution_positioning": 3.5,723   "methodological_rigor": 4.5,724   "fidelity": 5.5,725   "usefulness": 4.5726  },727  "('explain_specification', 'B_strong_prompt')": {728   "clarity": 6.0,729   "argument_logic": 5.0,730   "contribution_positioning": 3.5,731   "methodological_rigor": 4.5,732   "fidelity": 5.5,733   "usefulness": 4.5734  },735  "('explain_specification', 'C_conventional_rag')": {736   "clarity": 5.5,737   "argument_logic": 5.0,738   "contribution_positioning": 4.5,739   "methodological_rigor": 4.0,740   "fidelity": 5.0,741   "usefulness": 4.5742  },743  "('explain_specification', 'D_profiles_retrieval')": {744   "clarity": 6.0,745   "argument_logic": 5.5,746   "contribution_positioning": 3.5,747   "methodological_rigor": 4.5,748   "fidelity": 5.5,749   "usefulness": 4.5750  },751  "('explain_specification', 'E_full_engine')": {752   "clarity": 5.5,753   "argument_logic": 5.5,754   "contribution_positioning": 3.5,755   "methodological_rigor": 6.0,756   "fidelity": 6.5,757   "usefulness": 5.0758  },759  "('explain_specification', 'E_no_factcheck')": {760   "clarity": 6.5,761   "argument_logic": 6.0,762   "contribution_positioning": 3.5,763   "methodological_rigor": 5.0,764   "fidelity": 6.0,765   "usefulness": 5.0766  },767  "('explain_specification', 'E_no_profile')": {768   "clarity": 6.0,769   "argument_logic": 6.0,770   "contribution_positioning": 3.5,771   "methodological_rigor": 6.0,772   "fidelity": 5.0,773   "usefulness": 5.5774  },775  "('explain_specification', 'E_no_referee')": {776   "clarity": 6.0,777   "argument_logic": 5.0,778   "contribution_positioning": 3.5,779   "methodological_rigor": 4.5,780   "fidelity": 6.0,781   "usefulness": 5.0782  },783  "('explain_specification', 'E_no_retrieval')": {784   "clarity": 6.0,785   "argument_logic": 5.5,786   "contribution_positioning": 3.5,787   "methodological_rigor": 5.5,788   "fidelity": 5.5,789   "usefulness": 5.0790  },791  "('flag_causal_claims', 'A_baseline')": {792   "clarity": 5.5,793   "argument_logic": 5.0,794   "contribution_positioning": 3.5,795   "methodological_rigor": 5.0,796   "fidelity": 6.0,797   "usefulness": 4.5798  },799  "('flag_causal_claims', 'B_strong_prompt')": {800   "clarity": 6.0,801   "argument_logic": 6.0,802   "contribution_positioning": 4.0,803   "methodological_rigor": 6.0,804   "fidelity": 5.5,805   "usefulness": 5.5806  },807  "('flag_causal_claims', 'C_conventional_rag')": {808   "clarity": 6.0,809   "argument_logic": 6.0,810   "contribution_positioning": 4.5,811   "methodological_rigor": 6.0,812   "fidelity": 6.0,813   "usefulness": 5.0814  },815  "('flag_causal_claims', 'D_profiles_retrieval')": {816   "clarity": 5.5,817   "argument_logic": 5.0,818   "contribution_positioning": 3.5,819   "methodological_rigor": 5.0,820   "fidelity": 6.0,821   "usefulness": 4.0822  },823  "('flag_causal_claims', 'E_full_engine')": {824   "clarity": 6.0,825   "argument_logic": 6.5,826   "contribution_positioning": 4.0,827   "methodological_rigor": 6.5,828   "fidelity": 6.5,829   "usefulness": 5.0830  },831  "('flag_causal_claims', 'E_no_factcheck')": {832   "clarity": 5.5,833   "argument_logic": 6.0,834   "contribution_positioning": 4.0,835   "methodological_rigor": 6.0,836   "fidelity": 6.0,837   "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": 6.5,844   "fidelity": 6.5,845   "usefulness": 5.0846  },847  "('flag_causal_claims', 'E_no_referee')": {848   "clarity": 5.5,849   "argument_logic": 5.5,850   "contribution_positioning": 4.0,851   "methodological_rigor": 6.0,852   "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": 4.0,859   "methodological_rigor": 6.0,860   "fidelity": 6.5,861   "usefulness": 5.0862  },863  "('improve_abstract', 'A_baseline')": {864   "clarity": 5.0,865   "argument_logic": 5.25,866   "contribution_positioning": 4.5,867   "methodological_rigor": 4.0,868   "fidelity": 6.25,869   "usefulness": 4.75870  },871  "('improve_abstract', 'B_strong_prompt')": {872   "clarity": 5.0,873   "argument_logic": 5.0,874   "contribution_positioning": 5.0,875   "methodological_rigor": 4.25,876   "fidelity": 5.75,877   "usefulness": 4.75878  },879  "('improve_abstract', 'C_conventional_rag')": {880   "clarity": 5.5,881   "argument_logic": 5.5,882   "contribution_positioning": 5.25,883   "methodological_rigor": 4.25,884   "fidelity": 6.0,885   "usefulness": 5.25886  },887  "('improve_abstract', 'D_profiles_retrieval')": {888   "clarity": 5.25,889   "argument_logic": 5.25,890   "contribution_positioning": 4.75,891   "methodological_rigor": 4.25,892   "fidelity": 6.25,893   "usefulness": 5.0894  },895  "('improve_abstract', 'E_full_engine')": {896   "clarity": 5.0,897   "argument_logic": 5.25,898   "contribution_positioning": 4.75,899   "methodological_rigor": 5.25,900   "fidelity": 6.75,901   "usefulness": 4.75902  },903  "('improve_abstract', 'E_no_factcheck')": {904   "clarity": 5.0,905   "argument_logic": 6.0,906   "contribution_positioning": 5.0,907   "methodological_rigor": 5.0,908   "fidelity": 6.5,909   "usefulness": 5.0910  },911  "('improve_abstract', 'E_no_profile')": {912   "clarity": 5.5,913   "argument_logic": 5.5,914   "contribution_positioning": 5.0,915   "methodological_rigor": 5.0,916   "fidelity": 6.0,917   "usefulness": 5.0918  },919  "('improve_abstract', 'E_no_referee')": {920   "clarity": 5.0,921   "argument_logic": 5.0,922   "contribution_positioning": 4.5,923   "methodological_rigor": 4.0,924   "fidelity": 6.5,925   "usefulness": 5.0926  },927  "('improve_abstract', 'E_no_retrieval')": {928   "clarity": 4.5,929   "argument_logic": 5.5,930   "contribution_positioning": 5.0,931   "methodological_rigor": 5.5,932   "fidelity": 6.5,933   "usefulness": 5.0934  },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": 6.0,941   "usefulness": 5.5942  },943  "('improve_table', 'B_strong_prompt')": {944   "clarity": 5.5,945   "argument_logic": 5.0,946   "contribution_positioning": 4.0,947   "methodological_rigor": 4.0,948   "fidelity": 5.5,949   "usefulness": 5.0950  },951  "('improve_table', 'C_conventional_rag')": {952   "clarity": 5.5,953   "argument_logic": 4.5,954   "contribution_positioning": 3.5,955   "methodological_rigor": 4.0,956   "fidelity": 5.0,957   "usefulness": 5.0958  },959  "('improve_table', 'D_profiles_retrieval')": {960   "clarity": 5.0,961   "argument_logic": 4.5,962   "contribution_positioning": 3.5,963   "methodological_rigor": 4.5,964   "fidelity": 5.5,965   "usefulness": 5.0966  },967  "('improve_table', 'E_full_engine')": {968   "clarity": 5.5,969   "argument_logic": 5.5,970   "contribution_positioning": 4.5,971   "methodological_rigor": 6.0,972   "fidelity": 6.5,973   "usefulness": 5.5974  },975  "('improve_table', 'E_no_factcheck')": {976   "clarity": 4.5,977   "argument_logic": 5.5,978   "contribution_positioning": 3.5,979   "methodological_rigor": 6.0,980   "fidelity": 6.5,981   "usefulness": 4.5982  },983  "('improve_table', 'E_no_profile')": {984   "clarity": 5.0,985   "argument_logic": 5.5,986   "contribution_positioning": 4.0,987   "methodological_rigor": 6.5,988   "fidelity": 6.5,989   "usefulness": 5.0990  },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": 5.5,1001   "argument_logic": 5.5,1002   "contribution_positioning": 4.5,1003   "methodological_rigor": 6.0,1004   "fidelity": 6.0,1005   "usefulness": 5.01006  },1007  "('interpret_results', 'A_baseline')": {1008   "clarity": 6.0,1009   "argument_logic": 5.0,1010   "contribution_positioning": 4.0,1011   "methodological_rigor": 4.5,1012   "fidelity": 6.0,1013   "usefulness": 5.01014  },1015  "('interpret_results', 'B_strong_prompt')": {1016   "clarity": 6.0,1017   "argument_logic": 5.25,1018   "contribution_positioning": 4.5,1019   "methodological_rigor": 5.0,1020   "fidelity": 5.0,1021   "usefulness": 5.01022  },1023  "('interpret_results', 'C_conventional_rag')": {1024   "clarity": 6.0,1025   "argument_logic": 6.0,1026   "contribution_positioning": 4.5,1027   "methodological_rigor": 5.25,1028   "fidelity": 5.0,1029   "usefulness": 5.251030  },1031  "('interpret_results', 'D_profiles_retrieval')": {1032   "clarity": 6.0,1033   "argument_logic": 5.5,1034   "contribution_positioning": 4.25,1035   "methodological_rigor": 5.0,1036   "fidelity": 5.5,1037   "usefulness": 5.01038  },1039  "('interpret_results', 'E_full_engine')": {1040   "clarity": 5.75,1041   "argument_logic": 6.25,1042   "contribution_positioning": 4.5,1043   "methodological_rigor": 6.75,1044   "fidelity": 5.75,1045   "usefulness": 5.01046  },1047  "('interpret_results', 'E_no_factcheck')": {1048   "clarity": 6.0,1049   "argument_logic": 6.0,1050   "contribution_positioning": 4.5,1051   "methodological_rigor": 7.0,1052   "fidelity": 6.5,1053   "usefulness": 5.51054  },1055  "('interpret_results', 'E_no_profile')": {1056   "clarity": 6.0,1057   "argument_logic": 6.0,1058   "contribution_positioning": 4.5,1059   "methodological_rigor": 6.0,1060   "fidelity": 6.5,1061   "usefulness": 5.51062  },1063  "('interpret_results', 'E_no_referee')": {1064   "clarity": 6.0,1065   "argument_logic": 5.0,1066   "contribution_positioning": 4.5,1067   "methodological_rigor": 5.0,1068   "fidelity": 6.0,1069   "usefulness": 5.01070  },1071  "('interpret_results', 'E_no_retrieval')": {1072   "clarity": 6.0,1073   "argument_logic": 6.0,1074   "contribution_positioning": 4.5,1075   "methodological_rigor": 6.5,1076   "fidelity": 6.5,1077   "usefulness": 5.51078  },1079  "('referee_report', 'A_baseline')": {1080   "clarity": 5.25,1081   "argument_logic": 4.25,1082   "contribution_positioning": 3.25,1083   "methodological_rigor": 3.75,1084   "fidelity": 6.0,1085   "usefulness": 3.751086  },1087  "('referee_report', 'B_strong_prompt')": {1088   "clarity": 5.25,1089   "argument_logic": 4.75,1090   "contribution_positioning": 3.25,1091   "methodological_rigor": 4.5,1092   "fidelity": 5.25,1093   "usefulness": 4.01094  },1095  "('referee_report', 'C_conventional_rag')": {1096   "clarity": 5.25,1097   "argument_logic": 4.5,1098   "contribution_positioning": 3.5,1099   "methodological_rigor": 4.0,1100   "fidelity": 5.25,1101   "usefulness": 3.751102  },1103  "('referee_report', 'D_profiles_retrieval')": {1104   "clarity": 5.75,1105   "argument_logic": 6.0,1106   "contribution_positioning": 4.75,1107   "methodological_rigor": 5.75,1108   "fidelity": 6.25,1109   "usefulness": 5.251110  },1111  "('referee_report', 'E_full_engine')": {1112   "clarity": 6.0,1113   "argument_logic": 6.25,1114   "contribution_positioning": 4.75,1115   "methodological_rigor": 6.25,1116   "fidelity": 6.5,1117   "usefulness": 5.51118  },1119  "('referee_report', 'E_no_factcheck')": {1120   "clarity": 5.5,1121   "argument_logic": 5.5,1122   "contribution_positioning": 4.0,1123   "methodological_rigor": 6.0,1124   "fidelity": 6.5,1125   "usefulness": 5.01126  },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": 6.0,1133   "usefulness": 5.51134  },1135  "('referee_report', 'E_no_referee')": {1136   "clarity": 5.5,1137   "argument_logic": 4.5,1138   "contribution_positioning": 3.5,1139   "methodological_rigor": 4.0,1140   "fidelity": 6.0,1141   "usefulness": 3.51142  },1143  "('referee_report', 'E_no_retrieval')": {1144   "clarity": 6.0,1145   "argument_logic": 6.0,1146   "contribution_positioning": 5.0,1147   "methodological_rigor": 6.0,1148   "fidelity": 6.5,1149   "usefulness": 5.51150  },1151  "('rewrite_literature', 'A_baseline')": {1152   "clarity": 6.0,1153   "argument_logic": 5.0,1154   "contribution_positioning": 5.0,1155   "methodological_rigor": 3.5,1156   "fidelity": 5.0,1157   "usefulness": 5.01158  },1159  "('rewrite_literature', 'B_strong_prompt')": {1160   "clarity": 6.0,1161   "argument_logic": 5.0,1162   "contribution_positioning": 5.5,1163   "methodological_rigor": 4.0,1164   "fidelity": 3.5,1165   "usefulness": 5.01166  },1167  "('rewrite_literature', 'C_conventional_rag')": {1168   "clarity": 6.0,1169   "argument_logic": 6.0,1170   "contribution_positioning": 6.0,1171   "methodological_rigor": 4.5,1172   "fidelity": 5.0,1173   "usefulness": 6.01174  },1175  "('rewrite_literature', 'D_profiles_retrieval')": {1176   "clarity": 6.0,1177   "argument_logic": 5.5,1178   "contribution_positioning": 5.5,1179   "methodological_rigor": 4.0,1180   "fidelity": 5.0,1181   "usefulness": 5.51182  },1183  "('rewrite_literature', 'E_full_engine')": {1184   "clarity": 5.5,1185   "argument_logic": 5.5,1186   "contribution_positioning": 5.5,1187   "methodological_rigor": 6.5,1188   "fidelity": 5.5,1189   "usefulness": 5.51190  },1191  "('rewrite_literature', 'E_no_factcheck')": {1192   "clarity": 6.0,1193   "argument_logic": 6.0,1194   "contribution_positioning": 6.0,1195   "methodological_rigor": 6.5,1196   "fidelity": 5.5,1197   "usefulness": 6.01198  },1199  "('rewrite_literature', 'E_no_profile')": {1200   "clarity": 6.0,1201   "argument_logic": 6.0,1202   "contribution_positioning": 6.0,1203   "methodological_rigor": 6.5,1204   "fidelity": 6.0,1205   "usefulness": 6.01206  },1207  "('rewrite_literature', 'E_no_referee')": {1208   "clarity": 5.5,1209   "argument_logic": 5.5,1210   "contribution_positioning": 5.5,1211   "methodological_rigor": 3.5,1212   "fidelity": 4.5,1213   "usefulness": 5.01214  },1215  "('rewrite_literature', 'E_no_retrieval')": {1216   "clarity": 5.5,1217   "argument_logic": 5.5,1218   "contribution_positioning": 5.5,1219   "methodological_rigor": 6.0,1220   "fidelity": 5.5,1221   "usefulness": 5.51222  }1223 },1224 "out_of_domain_means": {1225  "A_baseline": {1226   "clarity": 5.67,1227   "argument_logic": 5.33,1228   "contribution_positioning": 3.67,1229   "methodological_rigor": 4.0,1230   "fidelity": 5.33,1231   "usefulness": 4.01232  },1233  "B_strong_prompt": {1234   "clarity": 5.67,1235   "argument_logic": 5.33,1236   "contribution_positioning": 4.0,1237   "methodological_rigor": 3.67,1238   "fidelity": 5.0,1239   "usefulness": 4.01240  },1241  "C_conventional_rag": {1242   "clarity": 6.33,1243   "argument_logic": 5.67,1244   "contribution_positioning": 5.33,1245   "methodological_rigor": 5.0,1246   "fidelity": 4.67,1247   "usefulness": 5.671248  },1249  "D_profiles_retrieval": {1250   "clarity": 5.33,1251   "argument_logic": 5.33,1252   "contribution_positioning": 4.0,1253   "methodological_rigor": 4.33,1254   "fidelity": 5.33,1255   "usefulness": 4.01256  },1257  "E_full_engine": {1258   "clarity": 5.33,1259   "argument_logic": 5.33,1260   "contribution_positioning": 3.67,1261   "methodological_rigor": 5.0,1262   "fidelity": 4.67,1263   "usefulness": 4.01264  }1265 },1266 "judge_consistency": {1267  "n_pairs": 50,1268  "clarity": {1269   "mean_abs_diff": 0.18,1270   "spearman": 0.7121271  },1272  "argument_logic": {1273   "mean_abs_diff": 0.38,1274   "spearman": 0.5991275  },1276  "contribution_positioning": {1277   "mean_abs_diff": 0.3,1278   "spearman": 0.9091279  },1280  "methodological_rigor": {1281   "mean_abs_diff": 0.28,1282   "spearman": 0.8931283  },1284  "fidelity": {1285   "mean_abs_diff": 0.44,1286   "spearman": 0.7391287  },1288  "usefulness": {1289   "mean_abs_diff": 0.24,1290   "spearman": 0.8181291  }1292 },1293 "secondary_means": {1294  "A_baseline": {1295   "preserved_rate": 0.199,1296   "new_numbers": 1.583,1297   "n_unverifiable_citations": 0.0,1298   "n_unsupported_claims": 1.167,1299   "max_overlap_share": 0.041,1300   "todo_placeholders": 12.292,1301   "n_words": 769.333,1302   "cost_usd": 0.076,1303   "latency_s": 33.675,1304   "n": 241305  },1306  "B_strong_prompt": {1307   "preserved_rate": 0.276,1308   "new_numbers": 1.208,1309   "n_unverifiable_citations": 0.0,1310   "n_unsupported_claims": 1.792,1311   "max_overlap_share": 0.043,1312   "todo_placeholders": 15.458,1313   "n_words": 828.417,1314   "cost_usd": 0.081,1315   "latency_s": 36.117,1316   "n": 241317  },1318  "C_conventional_rag": {1319   "preserved_rate": 0.244,1320   "new_numbers": 1.458,1321   "n_unverifiable_citations": 0.0,1322   "n_unsupported_claims": 1.5,1323   "max_overlap_share": 0.033,1324   "todo_placeholders": 12.625,1325   "n_words": 844.625,1326   "cost_usd": 0.087,1327   "latency_s": 37.613,1328   "n": 241329  },1330  "D_profiles_retrieval": {1331   "preserved_rate": 0.238,1332   "new_numbers": 1.833,1333   "n_unverifiable_citations": 0.0,1334   "n_unsupported_claims": 1.458,1335   "max_overlap_share": 0.032,1336   "todo_placeholders": 12.458,1337   "n_words": 847.5,1338   "cost_usd": 0.095,1339   "latency_s": 38.204,1340   "n": 241341  },1342  "E_full_engine": {1343   "preserved_rate": 0.302,1344   "new_numbers": 0.708,1345   "n_unverifiable_citations": 0.0,1346   "n_unsupported_claims": 0.875,1347   "max_overlap_share": 0.048,1348   "todo_placeholders": 12.25,1349   "n_words": 981.25,1350   "cost_usd": 0.202,1351   "latency_s": 80.062,1352   "n": 241353  },1354  "E_no_factcheck": {1355   "preserved_rate": 0.4,1356   "new_numbers": 1.938,1357   "n_unverifiable_citations": 0.0,1358   "n_unsupported_claims": 1.0,1359   "max_overlap_share": 0.025,1360   "todo_placeholders": 12.875,1361   "n_words": 909.062,1362   "cost_usd": 0.197,1363   "latency_s": 77.606,1364   "n": 161365  },1366  "E_no_profile": {1367   "preserved_rate": 0.364,1368   "new_numbers": 0.625,1369   "n_unverifiable_citations": 0.0,1370   "n_unsupported_claims": 1.125,1371   "max_overlap_share": 0.03,1372   "todo_placeholders": 13.062,1373   "n_words": 931.5,1374   "cost_usd": 0.194,1375   "latency_s": 77.194,1376   "n": 161377  },1378  "E_no_referee": {1379   "preserved_rate": 0.221,1380   "new_numbers": 1.438,1381   "n_unverifiable_citations": 0.0,1382   "n_unsupported_claims": 0.938,1383   "max_overlap_share": 0.041,1384   "todo_placeholders": 12.125,1385   "n_words": 790.812,1386   "cost_usd": 0.097,1387   "latency_s": 39.5,1388   "n": 161389  },1390  "E_no_retrieval": {1391   "preserved_rate": 0.305,1392   "new_numbers": 0.938,1393   "n_unverifiable_citations": 0.0,1394   "n_unsupported_claims": 0.75,1395   "max_overlap_share": 0.047,1396   "todo_placeholders": 14.312,1397   "n_words": 897.938,1398   "cost_usd": 0.195,1399   "latency_s": 78.019,1400   "n": 161401  }1402 },1403 "H4": {1404  "E_vs_A": {1405   "n_unverifiable_citations": {1406    "n": 24,1407    "mean_diff": 0.0,1408    "ci95": [1409     0.0,1410     0.01411    ],1412    "wilcoxon_p": 1.0,1413    "cliffs_delta": 0.01414   },1415   "n_unsupported_claims": {1416    "n": 24,1417    "mean_diff": -0.292,1418    "ci95": [1419     -0.875,1420     0.4691421    ],1422    "wilcoxon_p": 0.22484337530538745,1423    "cliffs_delta": -0.1091424   },1425   "new_numbers": {1426    "n": 24,1427    "mean_diff": -0.875,1428    "ci95": [1429     -1.938,1430     -0.0311431    ],1432    "wilcoxon_p": 0.040684908780744745,1433    "cliffs_delta": -0.1131434   }1435  },1436  "E_vs_B": {1437   "n_unverifiable_citations": {1438    "n": 24,1439    "mean_diff": 0.0,1440    "ci95": [1441     0.0,1442     0.01443    ],1444    "wilcoxon_p": 1.0,1445    "cliffs_delta": 0.01446   },1447   "n_unsupported_claims": {1448    "n": 24,1449    "mean_diff": -0.917,1450    "ci95": [1451     -1.688,1452     0.1251453    ],1454    "wilcoxon_p": 0.05044606501119172,1455    "cliffs_delta": -0.3121456   },1457   "new_numbers": {1458    "n": 24,1459    "mean_diff": -0.5,1460    "ci95": [1461     -1.125,1462     0.01463    ],1464    "wilcoxon_p": 0.08446903225325182,1465    "cliffs_delta": -0.0831466   }1467  },1468  "E_vs_C": {1469   "n_unverifiable_citations": {1470    "n": 24,1471    "mean_diff": 0.0,1472    "ci95": [1473     0.0,1474     0.01475    ],1476    "wilcoxon_p": 1.0,1477    "cliffs_delta": 0.01478   },1479   "n_unsupported_claims": {1480    "n": 24,1481    "mean_diff": -0.625,1482    "ci95": [1483     -1.281,1484     0.2191485    ],1486    "wilcoxon_p": 0.07156571124393864,1487    "cliffs_delta": -0.241488   },1489   "new_numbers": {1490    "n": 24,1491    "mean_diff": -0.75,1492    "ci95": [1493     -1.344,1494     -0.251495    ],1496    "wilcoxon_p": 0.026014479622894222,1497    "cliffs_delta": -0.1251498   }1499  }1500 },1501 "H5_overlap": {1502  "preregistered_metric": {1503   "max_overlap_share_overall": 0.36,1504   "share_outputs_below_0.02": 0.56521739130434781505  },1506  "excluding_source_document": {1507   "max_other_share": 0.0063,1508   "share_outputs_below_0.02": 1.0,1509   "mean_other_share": 0.00058315217391304361510  },1511  "self_overlap_by_task": {1512   "draft_introduction": 0.018,1513   "explain_specification": 0.012,1514   "flag_causal_claims": 0.038,1515   "improve_abstract": 0.117,1516   "improve_table": 0.035,1517   "interpret_results": 0.012,1518   "referee_report": 0.006,1519   "rewrite_literature": 0.0631520  },1521  "self_overlap_by_condition": {1522   "A_baseline": 0.04,1523   "B_strong_prompt": 0.043,1524   "C_conventional_rag": 0.033,1525   "D_profiles_retrieval": 0.032,1526   "E_full_engine": 0.048,1527   "E_no_factcheck": 0.025,1528   "E_no_profile": 0.03,1529   "E_no_referee": 0.041,1530   "E_no_retrieval": 0.0471531  }1532 }1533}
Files · b2ea7cd3a41b72854950b87ea6b3044f01e8a3c2
Menu

Theme