results: close no-KP causal bootstrap screen

4 files changed, 618 insertions, 0 deletions

diff --git a/ORAL_A_V5.md b/ORAL_A_V5.md
index c00e751..01a8ffa 100644
--- a/ORAL_A_V5.md
+++ b/ORAL_A_V5.md
@@ -99,3 +99,27 @@ the frozen local causal-fit confidence criterion is met.  Its exact cadence,
 confidence threshold, learning rates, cost gate, and accuracy gate must be
 fixed before any task endpoint is run.
 
+## Audited outcome (2026-08-06)
+
+V5-1 fails its causal-capture gate on clean source revision `6405220`.  All
+mechanical and accounting checks pass: the run is finite, uses exactly 380
+edge events, 760 logical batch-loss queries, and 48,640 per-example causal
+observations; forward parameters, BatchNorm state, and optimizer state remain
+bitwise unchanged; feedback norms stay in `[0.8778, 1.0036]` relative to their
+forward counterparts.
+
+The causal observations contain useful but depth-limited information.  Mean
+all-layer teaching alignment rises from fixed HFA's `0.011279` to `0.086687`,
+and the five latest hidden populations reach per-layer alignments
+`0.1992, 0.1397, 0.2924, 0.3837, 0.3418`.  However, early-third alignment is
+only `0.001147`, far below the frozen `0.10` threshold and only `0.001411`
+above fixed HFA.  Mean feedback/forward parameter cosine is just `0.013162`.
+Thus the signal improves layers close to the output but does not identify a
+deep feedback chain at this budget.
+
+No task-accuracy run, validation endpoint, test example, extra sweep, or rate
+recovery is opened.  V5 is closed and the reviewer score remains 5/10 for the
+standalone no-KP scaling claim.  A later branch may test a genuinely staged
+scheme that brings each child edge to a causal-fit confidence threshold before
+unlocking its parent; it must be treated as a new mechanism rather than a V5
+schedule rescue.
diff --git a/RESULTS.md b/RESULTS.md
index 6da95a0..9bf803e 100644
--- a/RESULTS.md
+++ b/RESULTS.md
@@ -1132,6 +1132,21 @@ its survival under the four-times-RMS mixed-traffic intervention.
 `results/figs/figure6_standard_depth_scaling.{pdf,png}` is regenerated from all
 60 records and verifies their hashes plus the historical training source.
 
+The separate Oral-A-v5 feasibility branch asks whether scalable feedback can
+be learned without KP or forward-weight mirroring.  It replaces V4's one
+batch-level scalar for all 267,904 feedback parameters with 128 independent
+per-example node-perturbation observations for one hierarchical edge at a
+time.  At a matched 760-query frozen-forward budget, the run is finite,
+preserves every forward/BatchNorm/optimizer tensor bitwise, and keeps feedback
+norm ratios in `[0.8778,1.0036]`.  All-layer alignment improves from fixed
+HFA's `0.011279` to `0.086687`, with late-layer alignments as high as `0.3837`.
+The preregistered gate nevertheless fails: early-third alignment is only
+`0.001147` versus the required `0.10`, and mean feedback/forward cosine is
+`0.013162`.  No accuracy run is opened.  This is evidence that the causal
+observations are informative near the output, not evidence that the no-KP
+vectorizer scales; the existing positive ResNet scaling result still belongs
+to the inherited KP substrate.
+
 ## How to run
 `experiments/run.py --mode {bp,fa,dfa,sdil} --dataset {mnist,fmnist,cifar10} --depth D --residual {0,1} --act {tanh,gelu,silu,relu}`
 Batteries: `experiments/run_v2.sh <ds> "<depths>" <res> <act> "<seeds>" <ep> <pfx>`.
diff --git a/results/oral_a_v5_calibration/result.json b/results/oral_a_v5_calibration/result.json
new file mode 100644
index 0000000..ac9f477
--- /dev/null
+++ b/results/oral_a_v5_calibration/result.json
@@ -0,0 +1,404 @@
+{
+  "architecture": {
+    "adaptive_feedback_parameters": 267904,
+    "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
+    "feedback_macs_per_example": 40108672,
+    "forward_macs_per_example": 40551040,
+    "forward_parameters": 269722
+  },
+  "finite": true,
+  "fixed_hfa": {
+    "all_layer_alignment": 0.01127886861764924,
+    "early_third_alignment": -0.00026316156436223537,
+    "feedback_forward_cosine": [
+      0.0008809715509414673,
+      -0.002320552244782448,
+      0.03730485588312149,
+      0.02497659996151924,
+      -0.03438456356525421,
+      -0.001257734838873148,
+      0.023045765236020088,
+      0.010684739798307419,
+      -0.013966654427349567,
+      -0.03158111870288849,
+      0.009583890438079834,
+      -9.433389641344547e-05,
+      0.004490766208618879,
+      -0.004784357734024525,
+      0.004104882013052702,
+      0.004543168004602194,
+      -0.005818712059408426,
+      0.0014950999757274985,
+      0.003767170011997223
+    ],
+    "feedback_forward_norm_ratio": [
+      0.9996462464332581,
+      1.0550107955932617,
+      0.9593725800514221,
+      1.0017613172531128,
+      0.999921441078186,
+      0.9631232619285583,
+      0.9992203712463379,
+      0.9976226687431335,
+      1.0018895864486694,
+      1.0030434131622314,
+      1.0180330276489258,
+      1.0107296705245972,
+      0.9962885975837708,
+      0.9891077876091003,
+      1.0009011030197144,
+      1.0021207332611084,
+      1.0043002367019653,
+      1.0021206140518188,
+      0.9992877244949341
+    ],
+    "max_feedback_forward_norm_ratio": 1.0550107955932617,
+    "mean_feedback_forward_cosine": 0.0016142042954207252,
+    "min_feedback_forward_norm_ratio": 0.9593725800514221,
+    "per_layer": [
+      -0.0005718563916161656,
+      -0.0006265100091695786,
+      0.0012021171860396862,
+      3.7638936191797256e-05,
+      0.0004848289827350527,
+      -0.002105188090354204,
+      0.0032924648839980364,
+      0.0002729729749262333,
+      0.0075433761812746525,
+      -0.005317362025380135,
+      0.009830434806644917,
+      -0.001097709871828556,
+      0.02368866838514805,
+      0.007433340419083834,
+      0.04359162598848343,
+      0.0057456037029623985,
+      0.039965368807315826,
+      0.0030560996383428574,
+      0.07787258923053741
+    ]
+  },
+  "hardware": {
+    "cuda_device_name": "NVIDIA GeForce GTX 1080",
+    "cuda_visible_devices": "5",
+    "device": "cuda",
+    "peak_memory_allocated_bytes": 986384384,
+    "torch_version": "2.3.1+cu118"
+  },
+  "learned_lcb": {
+    "all_layer_alignment": 0.0866869534748778,
+    "early_third_alignment": 0.0011474320829923574,
+    "feedback_forward_cosine": [
+      -0.004378448240458965,
+      -0.007337348069995642,
+      0.04171781986951828,
+      0.02352023683488369,
+      -0.039886925369501114,
+      0.0006825287127867341,
+      0.02527846209704876,
+      0.0136695746332407,
+      -0.011026157066226006,
+      -0.027840018272399902,
+      0.00775768980383873,
+      0.00863435585051775,
+      0.009166959673166275,
+      0.0012933104299008846,
+      0.010724963620305061,
+      0.011407784186303616,
+      0.004035680089145899,
+      0.005515150260180235,
+      0.17713752388954163
+    ],
+    "feedback_forward_norm_ratio": [
+      0.9522605538368225,
+      0.9822989702224731,
+      0.9040371775627136,
+      0.9258716702461243,
+      0.9323726892471313,
+      0.8777714967727661,
+      0.9508335590362549,
+      0.9598008394241333,
+      0.9715808629989624,
+      0.9627054333686829,
+      0.9862542152404785,
+      0.972489595413208,
+      0.9752699136734009,
+      0.9769584536552429,
+      0.9962733387947083,
+      0.9905397295951843,
+      1.0036309957504272,
+      0.9967941641807556,
+      0.8897679448127747
+    ],
+    "max_feedback_forward_norm_ratio": 1.0036309957504272,
+    "mean_feedback_forward_cosine": 0.0131617443648314,
+    "min_feedback_forward_norm_ratio": 0.8777714967727661,
+    "per_layer": [
+      -0.0008834822801873088,
+      -0.0023195152170956135,
+      0.0026626307517290115,
+      -0.00018003323930315673,
+      0.00810601282864809,
+      -0.0005010203458368778,
+      0.017151430249214172,
+      -0.0013304761378094554,
+      0.03392089158296585,
+      -0.0038246146868914366,
+      0.06399191915988922,
+      0.011081752367317677,
+      0.11008386313915253,
+      0.052304450422525406,
+      0.1991788148880005,
+      0.13974127173423767,
+      0.2924402058124542,
+      0.38367098569869995,
+      0.34175702929496765
+    ]
+  },
+  "method_audit": {
+    "causal_query_normalization_state": "evaluation_running_statistics",
+    "forward_state_max_absolute_difference": 0.0,
+    "forward_weight_reads_in_feedback_update": 0,
+    "ordinary_task_normalization_state": "not_run_forward_frozen",
+    "reverse_mode_learning_operations": 0
+  },
+  "protocol": "oral_a_v5_layerwise_causal_bootstrap_capture_v1",
+  "provenance": {
+    "git_commit": "640522076c770d0746dc58dde868586ace568752",
+    "git_tracked_dirty": false
+  },
+  "schema_version": 1,
+  "settings": {
+    "alignment_probe": 64,
+    "batch_size": 128,
+    "calibration_augmentation": false,
+    "depth": 20,
+    "eta_A": 0.1,
+    "feedback_scale": 1.0,
+    "loader_seed": 0,
+    "normalization": "batchnorm",
+    "perturb_seed": 5000,
+    "residual_scale": 1.0,
+    "seed": 0,
+    "sigma": 0.01,
+    "split_seed": 2027,
+    "sweeps": 20,
+    "train_limit": 10000,
+    "val_examples": 5000,
+    "width": 16
+  },
+  "split": {
+    "dataset": "cifar10",
+    "input_layout": "NCHW",
+    "input_shape": [
+      3,
+      32,
+      32
+    ],
+    "loader_seed": 0,
+    "normalization_mean": [
+      0.49140000343322754,
+      0.4821999967098236,
+      0.4465000033378601
+    ],
+    "normalization_std": [
+      0.24699999392032623,
+      0.2434999942779541,
+      0.26159998774528503
+    ],
+    "split_from_training_only": true,
+    "split_seed": 2027,
+    "test_examples": 10000,
+    "train_examples": 10000,
+    "training_augmentation": "none",
+    "validation_class_counts": {
+      "0": 500,
+      "1": 500,
+      "2": 500,
+      "3": 500,
+      "4": 500,
+      "5": 500,
+      "6": 500,
+      "7": 500,
+      "8": 500,
+      "9": 500
+    },
+    "validation_examples": 5000,
+    "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b"
+  },
+  "sweeps": [
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.06403723350284288,
+      "mean_field_prediction_target_cosine": 0.0004417389199711075,
+      "mean_parameter_update_rms": 0.03444637028278504,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.06253894281438753,
+      "mean_field_prediction_target_cosine": 0.0018398856551996173,
+      "mean_parameter_update_rms": 0.03581342072986328,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.06352942784015662,
+      "mean_field_prediction_target_cosine": 0.0007146904780988803,
+      "mean_parameter_update_rms": 0.037026025369156314,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.06931613884208412,
+      "mean_field_prediction_target_cosine": 0.0020960899414459977,
+      "mean_parameter_update_rms": 0.03692294811063335,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.07834713139232216,
+      "mean_field_prediction_target_cosine": 0.0013170423593140336,
+      "mean_parameter_update_rms": 0.039807939050467284,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.07501814697739963,
+      "mean_field_prediction_target_cosine": 0.002235385887437107,
+      "mean_parameter_update_rms": 0.04167423709322971,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.08537822862592329,
+      "mean_field_prediction_target_cosine": 0.0018632679655232866,
+      "mean_parameter_update_rms": 0.04244392278182938,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.08774229177580244,
+      "mean_field_prediction_target_cosine": 0.003061714145077791,
+      "mean_parameter_update_rms": 0.04222976858811638,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.09508034534172614,
+      "mean_field_prediction_target_cosine": 0.00234870503581423,
+      "mean_parameter_update_rms": 0.043422751889732325,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.09905501294247436,
+      "mean_field_prediction_target_cosine": 0.003315453860516912,
+      "mean_parameter_update_rms": 0.04370078720903135,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.1189526810892697,
+      "mean_field_prediction_target_cosine": 0.0028806602201959576,
+      "mean_parameter_update_rms": 0.046994848933308166,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.08835658663303694,
+      "mean_field_prediction_target_cosine": 0.0036876431588643115,
+      "mean_parameter_update_rms": 0.0414807426622568,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.09948273630711328,
+      "mean_field_prediction_target_cosine": 0.0031894343839509965,
+      "mean_parameter_update_rms": 0.04500005709476786,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.11277010760757089,
+      "mean_field_prediction_target_cosine": 0.0036998344357462333,
+      "mean_parameter_update_rms": 0.04482713238777737,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.10252816278564553,
+      "mean_field_prediction_target_cosine": 0.004821155006692386,
+      "mean_parameter_update_rms": 0.041077959193793326,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.10446445100929329,
+      "mean_field_prediction_target_cosine": 0.0029003324771935436,
+      "mean_parameter_update_rms": 0.04271476741395773,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.10518028307101368,
+      "mean_field_prediction_target_cosine": 0.004133514758470006,
+      "mean_parameter_update_rms": 0.048030305729152116,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.11175108755039062,
+      "mean_field_prediction_target_cosine": 0.0042756213137445384,
+      "mean_parameter_update_rms": 0.04818789939741672,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.13193168146187756,
+      "mean_field_prediction_target_cosine": 0.004254287067517157,
+      "mean_parameter_update_rms": 0.049472996929179495,
+      "per_example_causal_observations": 2432
+    },
+    {
+      "events": 19,
+      "logical_batch_loss_queries": 38,
+      "max_parameter_update_rms": 0.11125775394135021,
+      "mean_field_prediction_target_cosine": 0.00429639637704831,
+      "mean_parameter_update_rms": 0.048118302209782284,
+      "per_example_causal_observations": 2432
+    }
+  ],
+  "test_examples_touched": 0,
+  "validation_endpoints_observed": 0,
+  "wall_seconds": 13.465652227401733,
+  "work": {
+    "clean_forward_examples": 2560,
+    "edge_events": 380,
+    "forward_macs": 4048615833600,
+    "hierarchical_teaching_and_local_correlation_macs_estimate": 1950885806080,
+    "logical_batch_loss_queries": 760,
+    "per_example_causal_observations": 48640,
+    "per_example_cross_entropy_terms": 97280,
+    "perturbation_forward_examples": 97280,
+    "total_macs_estimate": 5999501639680
+  }
+}
diff --git a/results/oral_a_v5_calibration_gate.json b/results/oral_a_v5_calibration_gate.json
new file mode 100644
index 0000000..791e457
--- /dev/null
+++ b/results/oral_a_v5_calibration_gate.json
@@ -0,0 +1,175 @@
+{
+  "checks": {
+    "all_layer_at_least_0.20": false,
+    "early_gain_over_fixed_hfa_at_least_0.08": false,
+    "early_third_at_least_0.10": false,
+    "exactly_380_edge_events": true,
+    "exactly_48640_per_example_observations": true,
+    "exactly_760_batch_loss_queries": true,
+    "feedback_norm_ratios_in_0.1_to_3": true,
+    "finite": true,
+    "forward_state_bitwise_fixed": true,
+    "zero_forward_weight_reads_in_update": true,
+    "zero_reverse_mode_learning_operations": true
+  },
+  "conditional_short_task_gate_open": false,
+  "confirmation_test_seeds_touched": false,
+  "fixed_hfa": {
+    "all_layer_alignment": 0.01127886861764924,
+    "early_third_alignment": -0.00026316156436223537,
+    "feedback_forward_cosine": [
+      0.0008809715509414673,
+      -0.002320552244782448,
+      0.03730485588312149,
+      0.02497659996151924,
+      -0.03438456356525421,
+      -0.001257734838873148,
+      0.023045765236020088,
+      0.010684739798307419,
+      -0.013966654427349567,
+      -0.03158111870288849,
+      0.009583890438079834,
+      -9.433389641344547e-05,
+      0.004490766208618879,
+      -0.004784357734024525,
+      0.004104882013052702,
+      0.004543168004602194,
+      -0.005818712059408426,
+      0.0014950999757274985,
+      0.003767170011997223
+    ],
+    "feedback_forward_norm_ratio": [
+      0.9996462464332581,
+      1.0550107955932617,
+      0.9593725800514221,
+      1.0017613172531128,
+      0.999921441078186,
+      0.9631232619285583,
+      0.9992203712463379,
+      0.9976226687431335,
+      1.0018895864486694,
+      1.0030434131622314,
+      1.0180330276489258,
+      1.0107296705245972,
+      0.9962885975837708,
+      0.9891077876091003,
+      1.0009011030197144,
+      1.0021207332611084,
+      1.0043002367019653,
+      1.0021206140518188,
+      0.9992877244949341
+    ],
+    "max_feedback_forward_norm_ratio": 1.0550107955932617,
+    "mean_feedback_forward_cosine": 0.0016142042954207252,
+    "min_feedback_forward_norm_ratio": 0.9593725800514221,
+    "per_layer": [
+      -0.0005718563916161656,
+      -0.0006265100091695786,
+      0.0012021171860396862,
+      3.7638936191797256e-05,
+      0.0004848289827350527,
+      -0.002105188090354204,
+      0.0032924648839980364,
+      0.0002729729749262333,
+      0.0075433761812746525,
+      -0.005317362025380135,
+      0.009830434806644917,
+      -0.001097709871828556,
+      0.02368866838514805,
+      0.007433340419083834,
+      0.04359162598848343,
+      0.0057456037029623985,
+      0.039965368807315826,
+      0.0030560996383428574,
+      0.07787258923053741
+    ]
+  },
+  "learned_lcb": {
+    "all_layer_alignment": 0.0866869534748778,
+    "early_third_alignment": 0.0011474320829923574,
+    "feedback_forward_cosine": [
+      -0.004378448240458965,
+      -0.007337348069995642,
+      0.04171781986951828,
+      0.02352023683488369,
+      -0.039886925369501114,
+      0.0006825287127867341,
+      0.02527846209704876,
+      0.0136695746332407,
+      -0.011026157066226006,
+      -0.027840018272399902,
+      0.00775768980383873,
+      0.00863435585051775,
+      0.009166959673166275,
+      0.0012933104299008846,
+      0.010724963620305061,
+      0.011407784186303616,
+      0.004035680089145899,
+      0.005515150260180235,
+      0.17713752388954163
+    ],
+    "feedback_forward_norm_ratio": [
+      0.9522605538368225,
+      0.9822989702224731,
+      0.9040371775627136,
+      0.9258716702461243,
+      0.9323726892471313,
+      0.8777714967727661,
+      0.9508335590362549,
+      0.9598008394241333,
+      0.9715808629989624,
+      0.9627054333686829,
+      0.9862542152404785,
+      0.972489595413208,
+      0.9752699136734009,
+      0.9769584536552429,
+      0.9962733387947083,
+      0.9905397295951843,
+      1.0036309957504272,
+      0.9967941641807556,
+      0.8897679448127747
+    ],
+    "max_feedback_forward_norm_ratio": 1.0036309957504272,
+    "mean_feedback_forward_cosine": 0.0131617443648314,
+    "min_feedback_forward_norm_ratio": 0.8777714967727661,
+    "per_layer": [
+      -0.0008834822801873088,
+      -0.0023195152170956135,
+      0.0026626307517290115,
+      -0.00018003323930315673,
+      0.00810601282864809,
+      -0.0005010203458368778,
+      0.017151430249214172,
+      -0.0013304761378094554,
+      0.03392089158296585,
+      -0.0038246146868914366,
+      0.06399191915988922,
+      0.011081752367317677,
+      0.11008386313915253,
+      0.052304450422525406,
+      0.1991788148880005,
+      0.13974127173423767,
+      0.2924402058124542,
+      0.38367098569869995,
+      0.34175702929496765
+    ]
+  },
+  "protocol": "oral_a_v5_layerwise_causal_bootstrap_gate_v1",
+  "review_score_after": 5,
+  "review_score_before": 5,
+  "score_change_rule": "causal capture alone cannot raise score",
+  "source_commit": "640522076c770d0746dc58dde868586ace568752",
+  "source_result": "results/oral_a_v5_calibration/result.json",
+  "status": "failed",
+  "work": {
+    "clean_forward_examples": 2560,
+    "edge_events": 380,
+    "forward_macs": 4048615833600,
+    "hierarchical_teaching_and_local_correlation_macs_estimate": 1950885806080,
+    "logical_batch_loss_queries": 760,
+    "per_example_causal_observations": 48640,
+    "per_example_cross_entropy_terms": 97280,
+    "perturbation_forward_examples": 97280,
+    "total_macs_estimate": 5999501639680
+  }
+}