Value-convergence excess across the fleet beats the capability-matched null

The reading

1. value-convergence excess on real reward-model pairs beats the capability-matched random-utility null

   real_vce > 0 0.5 -0.06987 Value-convergence excess across the fleet beats the capability-matched null refuted registered confidence interval -0.12281662932807336 to -0.009192229234832329 > 0 The frozen study "Value-convergence excess across the fleet beats the capability-matched null", adjudicated on 19 July 2026 at the registered level of the project's evidence ladder. One frozen study, adjudicated once, on fleet, over rb2-full and rb2-helpfulness-512. It does not establish that the same holds on another subject, another dataset or another run. the vce > 0 margin sits at the null center by construction: the point call passes under the null 0.48 of the time and its power at the threshold-alternative equals that rate; the alpha-exact p-value companion and the registered CI exclusion carry the card's evidence. Open the record › Copy technical provenance refuted

2. the convergent-pair alignment exceeds the random-utility null at p < 0.05

   reward_convergent_p_value < 0.05 0.001 Value-convergence excess across the fleet beats the capability-matched null refuted registered < 0.05 The frozen study "Value-convergence excess across the fleet beats the capability-matched null", adjudicated on 19 July 2026 at the registered level of the project's evidence ladder. One frozen study, adjudicated once, on fleet, over rb2-full and rb2-helpfulness-512. It does not establish that the same holds on another subject, another dataset or another run. the vce > 0 margin sits at the null center by construction: the point call passes under the null 0.48 of the time and its power at the threshold-alternative equals that rate; the alpha-exact p-value companion and the registered CI exclusion carry the card's evidence. Open the record › Copy technical provenance confirmed

convergence is fully explained by shared capability structure

real_vce <= 0 -0.06987 Value-convergence excess across the fleet beats the capability-matched null refuted registered confidence interval -0.12281662932807336 to -0.009192229234832329 > 0 The frozen study "Value-convergence excess across the fleet beats the capability-matched null", adjudicated on 19 July 2026 at the registered level of the project's evidence ladder. One frozen study, adjudicated once, on fleet, over rb2-full and rb2-helpfulness-512. It does not establish that the same holds on another subject, another dataset or another run. the vce > 0 margin sits at the null center by construction: the point call passes under the null 0.48 of the time and its power at the threshold-alternative equals that rate; the alpha-exact p-value companion and the registered CI exclusion carry the card's evidence. Open the record › Copy technical provenance

Measurements

real_vce_ci_low

-0.1228 Value-convergence excess across the fleet beats the capability-matched null refuted registered confidence interval -0.12281662932807336 to -0.009192229234832329 > 0 The frozen study "Value-convergence excess across the fleet beats the capability-matched null", adjudicated on 19 July 2026 at the registered level of the project's evidence ladder. One frozen study, adjudicated once, on fleet, over rb2-full and rb2-helpfulness-512. It does not establish that the same holds on another subject, another dataset or another run. the vce > 0 margin sits at the null center by construction: the point call passes under the null 0.48 of the time and its power at the threshold-alternative equals that rate; the alpha-exact p-value companion and the registered CI exclusion carry the card's evidence. Open the record › Copy technical provenance

real_vce_ci_high

-0.009192 Value-convergence excess across the fleet beats the capability-matched null refuted registered confidence interval -0.12281662932807336 to -0.009192229234832329 > 0 The frozen study "Value-convergence excess across the fleet beats the capability-matched null", adjudicated on 19 July 2026 at the registered level of the project's evidence ladder. One frozen study, adjudicated once, on fleet, over rb2-full and rb2-helpfulness-512. It does not establish that the same holds on another subject, another dataset or another run. the vce > 0 margin sits at the null center by construction: the point call passes under the null 0.48 of the time and its power at the threshold-alternative equals that rate; the alpha-exact p-value companion and the registered CI exclusion carry the card's evidence. Open the record › Copy technical provenance

Provenance

card

ATLAS-VCE

frozen_at

2026-07-18

adjudicated_at

2026-07-19

spec_hash

spec:2008bee7d02b9461abfd6f12ff7b0ffe

study_id

study:campaign-atlas-vce@v1#2008bee7

Evidence records

0c80d9086ebb0ad9704b88a8eefb01fa

power_acceptance

the vce > 0 margin sits at the null center by construction: the point call passes under the null 0.48 of the time and its power at the threshold-alternative equals that rate; the alpha-exact p-value companion and the registered CI exclusion carry the card's evidence

spec_file

campaign-atlas-vce.json