Skip to content

Commit 0278ee2

Browse files
author
Peter Higgins
committed
updates
2 parents 9e1941b + a68693d commit 0278ee2

10 files changed

Lines changed: 1735 additions & 0 deletions

‎drafts/codawork-2026/EITT_CoDa_Cheatsheet.pdf‎

-2 Bytes
Binary file not shown.

‎tools/pipeline/adaptive_tap_test_2026april10.txt‎

Lines changed: 267 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,267 @@
1+
================================================================================
2+
ADAPTIVE TAP DECIMATION TEST — 2026-04-10
3+
d(CoDa)/dt as the TAP CONTROLLER: compression adapts to rate of change.
4+
Fast-moving balances → less compression. Slow balances → more compression.
5+
================================================================================
6+
7+
================================================================================
8+
TEST 1: ENERGY WORLD — CAN ADAPTIVE TAPS RESCUE Solar|Rest?
9+
N=25, 2000→2024
10+
================================================================================
11+
12+
─── d(CoDa)/dt Profile (Solar|Rest tree) ───
13+
14+
Year Max |dB/dt| B1 Solar|Rest Suggested tap
15+
──────────────────────────────────────────────
16+
2000 0.1853 0.1853 3:1
17+
2001 0.1719 0.1719 3:1
18+
2002 0.1726 0.1726 3:1
19+
2003 0.1849 0.1849 3:1
20+
2004 0.2056 0.2056 3:1
21+
2005 0.2231 0.2231 2:1
22+
2006 0.2786 0.2786 2:1
23+
2007 0.3219 0.3219 2:1 (fast)
24+
2008 0.3589 0.3589 2:1 (fast)
25+
2009 0.4517 0.4517 2:1 (fast)
26+
2010 0.4415 0.4415 2:1 (fast)
27+
2011 0.3914 0.3914 2:1 (fast)
28+
2012 0.3831 0.3831 2:1 (fast)
29+
2013 0.2885 0.2885 2:1
30+
2014 0.2467 0.2467 2:1
31+
2015 0.2468 0.2468 2:1
32+
2016 0.2107 0.2107 3:1
33+
2017 0.1971 0.1971 3:1
34+
2018 0.1942 0.1942 3:1
35+
2019 0.1708 0.1708 3:1
36+
2020 0.1707 0.1707 3:1
37+
2021 0.1790 0.1790 3:1
38+
2022 0.1857 0.1857 3:1
39+
2023 0.1971 0.1971 3:1
40+
2024 0.1952 0.1952 3:1
41+
42+
─── Fixed 5:1 Decimation ───
43+
44+
EITT residual: -0.0870%
45+
B1: Solar|Rest: mean -1.2640 → -1.2640 (shift=+0.0000) [PASS]
46+
B2: Fossil|Clean: mean +5.7478 → +5.7478 (shift=-0.0000) [PASS]
47+
B3: Coal|Gas: mean +0.4000 → +0.4000 (shift=+0.0000) [PASS]
48+
B4: Nuclear|RenewExSol: mean +4.7325 → +4.7325 (shift=-0.0000) [PASS]
49+
50+
─── Fixed 2:1 Decimation ───
51+
52+
EITT residual: -0.5891%
53+
B1: Solar|Rest: mean -1.2640 → -1.3853 (shift=-0.1213) [FAIL]
54+
B2: Fossil|Clean: mean +5.7478 → +5.7594 (shift=+0.0115) [PASS]
55+
B3: Coal|Gas: mean +0.4000 → +0.4041 (shift=+0.0041) [PASS]
56+
B4: Nuclear|RenewExSol: mean +4.7325 → +4.7557 (shift=+0.0231) [PASS]
57+
58+
─── Adaptive Decimation (rate-controlled taps) ───
59+
60+
Config: Sensitive (low=0.02, high=0.15)
61+
Blocks: 12, sizes: [2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2]
62+
Mean block size: 2.0
63+
Effective compression: 2.1:1
64+
EITT residual: -0.5891%
65+
B1: Solar|Rest: shift=-0.1213 [FAIL]
66+
B2: Fossil|Clean: shift=+0.0115 [PASS]
67+
B3: Coal|Gas: shift=+0.0041 [PASS]
68+
B4: Nuclear|RenewExSol: shift=+0.0231 [PASS]
69+
ALL TAPS PRESERVED: NO
70+
71+
Config: Standard (low=0.05, high=0.30)
72+
Blocks: 9, sizes: [3, 3, 2, 2, 2, 2, 3, 3, 4]
73+
Mean block size: 2.7
74+
Effective compression: 2.8:1
75+
EITT residual: -1.5577%
76+
B1: Solar|Rest: shift=-0.2986 [FAIL]
77+
B2: Fossil|Clean: shift=+0.0286 [PASS]
78+
B3: Coal|Gas: shift=+0.0118 [PASS]
79+
B4: Nuclear|RenewExSol: shift=+0.0521 [PASS]
80+
ALL TAPS PRESERVED: NO
81+
82+
Config: Tolerant (low=0.10, high=0.50)
83+
Blocks: 7, sizes: [4, 4, 3, 3, 4, 4, 3]
84+
Mean block size: 3.6
85+
Effective compression: 3.6:1
86+
EITT residual: +0.1480%
87+
B1: Solar|Rest: shift=+0.0765 [PASS]
88+
B2: Fossil|Clean: shift=-0.0072 [PASS]
89+
B3: Coal|Gas: shift=-0.0020 [PASS]
90+
B4: Nuclear|RenewExSol: shift=-0.0196 [PASS]
91+
ALL TAPS PRESERVED: YES
92+
93+
94+
─── Adaptive on Standard SBP (Fossil|Renew) for comparison ───
95+
96+
Blocks: 6, sizes: [4, 4, 4, 3, 4, 5]
97+
EITT residual: -1.1120%
98+
B1: Fossil|Renew: shift=+0.0861 [PASS]
99+
B2: Coal|Gas: shift=+0.0091 [PASS]
100+
B3: Wind|Solar: shift=+0.0805 [PASS]
101+
B4: Hydro|VarRenew: shift=+0.0901 [PASS]
102+
103+
104+
================================================================================
105+
TEST 2: GERMANY — CAN ADAPTIVE TAPS HANDLE THE NUCLEAR PHASE-OUT?
106+
N=36, 1990→2025
107+
================================================================================
108+
109+
─── d(CoDa)/dt Profile ───
110+
111+
Year Max |dB/dt| B1 Foss|Ren B2 Coal|Gas B3 Win|Sol
112+
──────────────────────────────────────────────────────
113+
1990 0.5025 0.2429 0.0130 0.5025
114+
1991 0.5324 0.2479 0.0082 0.5324
115+
1992 0.5333 0.2361 0.0126 0.5333
116+
1993 0.3625 0.1665 0.0329 0.3625
117+
1994 0.2785 0.2785 0.0568 0.0588
118+
1995 0.2815 0.2221 0.0731 0.1403
119+
1996 0.2820 0.1897 0.0590 0.2009
120+
1997 0.3232 0.2641 0.0458 0.3221
121+
1998 0.6727 0.0699 0.0132 0.6727
122+
1999 0.2497 0.2497 0.0195 0.1880
123+
2000 0.3104 0.3084 0.0140 0.3104
124+
2001 0.3024 0.2433 0.0193 0.2655
125+
2002 0.9377 0.4913 0.0415 0.9377
126+
2003 0.2878 0.2104 0.0499 0.2822
127+
2004 0.3091 0.1784 0.0565 0.3091
128+
2005 0.2763 0.2047 0.0421 0.2763
129+
2006 0.4035 0.3612 0.0755 0.2868
130+
2007 0.3663 0.3604 0.0414 0.2271
131+
2008 0.3724 0.3724 0.0471 0.2570
132+
2009 0.4039 0.3551 0.0397 0.2905
133+
2010 0.2753 0.1997 0.0276 0.2753
134+
2011 0.2517 0.2517 0.0547 0.2166
135+
2012 0.2394 0.2394 0.0750 0.1183
136+
2013 0.2641 0.2641 0.0651 0.0290
137+
2014 0.1935 0.1762 0.0201 0.0170
138+
2015 0.1458 0.0972 0.0756 0.0812
139+
2016 0.1262 0.0805 0.0851 0.0721
140+
2017 0.1491 0.0901 0.1491 0.0484
141+
2018 0.1466 0.1436 0.1466 0.0401
142+
2019 0.0986 0.0986 0.0761 0.0290
143+
2020 0.0810 0.0810 0.0365 0.0320
144+
2021 0.1035 0.1035 0.0278 0.0416
145+
2022 0.1101 0.1101 0.0077 0.0591
146+
2023 0.3387 0.2599 0.0670 0.0745
147+
2024 0.4983 0.3596 0.1417 0.0731
148+
2025 0.6919 0.6919 0.0940 0.1313
149+
150+
─── Fixed 5:1 Decimation ───
151+
152+
EITT residual: -1.2609%
153+
B1: Fossil|Renew: shift=+0.0475 [PASS]
154+
B2: Coal|Gas: shift=+0.0231 [PASS]
155+
B3: Wind|Solar: shift=+0.0535 [PASS]
156+
B4: Hydro|VarRenew: shift=+0.0380 [PASS]
157+
158+
─── Adaptive Decimation ───
159+
160+
Config: Sensitive (low=0.02, high=0.15)
161+
Blocks: 16, sizes: [2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3]
162+
Effective compression: 2.2:1
163+
EITT residual: -2.6822%
164+
B1: Fossil|Renew: shift=+0.3034 [FAIL]
165+
B2: Coal|Gas: shift=+0.0681 [PASS]
166+
B3: Wind|Solar: shift=+0.2051 [FAIL]
167+
B4: Hydro|VarRenew: shift=+0.2888 [FAIL]
168+
ALL TAPS PRESERVED: NO
169+
170+
Config: Standard (low=0.05, high=0.30)
171+
Blocks: 15, sizes: [2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 4, 4, 2]
172+
Effective compression: 2.4:1
173+
EITT residual: -3.7493%
174+
B1: Fossil|Renew: shift=+0.4523 [FAIL]
175+
B2: Coal|Gas: shift=+0.0699 [PASS]
176+
B3: Wind|Solar: shift=+0.3277 [FAIL]
177+
B4: Hydro|VarRenew: shift=+0.4523 [FAIL]
178+
ALL TAPS PRESERVED: NO
179+
180+
Config: Tolerant (low=0.10, high=0.50)
181+
Blocks: 11, sizes: [2, 2, 4, 2, 3, 4, 3, 4, 4, 5, 3]
182+
Effective compression: 3.3:1
183+
EITT residual: -3.4929%
184+
B1: Fossil|Renew: shift=+0.4551 [FAIL]
185+
B2: Coal|Gas: shift=+0.0608 [PASS]
186+
B3: Wind|Solar: shift=+0.2735 [FAIL]
187+
B4: Hydro|VarRenew: shift=+0.4665 [FAIL]
188+
ALL TAPS PRESERVED: NO
189+
190+
191+
192+
================================================================================
193+
TEST 3: GOLD/SILVER — ADAPTIVE ON 338 YEARS
194+
N=339, 1688→2026
195+
================================================================================
196+
197+
Rate profile: min=0.0000, median=0.0029, max=0.0790
198+
199+
Fastest-changing decades:
200+
2020s: mean |dB/dt| = 0.0519
201+
1980s: mean |dB/dt| = 0.0500
202+
1920s: mean |dB/dt| = 0.0446
203+
1940s: mean |dB/dt| = 0.0434
204+
1930s: mean |dB/dt| = 0.0367
205+
1890s: mean |dB/dt| = 0.0300
206+
1960s: mean |dB/dt| = 0.0293
207+
2010s: mean |dB/dt| = 0.0232
208+
1970s: mean |dB/dt| = 0.0214
209+
1910s: mean |dB/dt| = 0.0209
210+
211+
Fixed 10:1 (failed earlier at +1.39%):
212+
EITT residual: +1.3880%
213+
214+
Adaptive (base_ratio=10):
215+
Sensitive (low=0.02, high=0.15):
216+
Blocks: 35, mean size: 9.7
217+
Effective compression: 9.7:1
218+
EITT residual: -1.4829%
219+
vs fixed 10:1: WORSE or SAME
220+
221+
Standard (low=0.05, high=0.30):
222+
Blocks: 34, mean size: 10.0
223+
Effective compression: 10.0:1
224+
EITT residual: -0.4319%
225+
vs fixed 10:1: IMPROVED
226+
227+
Tolerant (low=0.10, high=0.50):
228+
Blocks: 34, mean size: 10.0
229+
Effective compression: 10.0:1
230+
EITT residual: -0.4319%
231+
vs fixed 10:1: IMPROVED
232+
233+
234+
235+
================================================================================
236+
TEST 4: FINANCIAL K=9 — ADAPTIVE ON ALREADY-PRISTINE DATA
237+
N=74, 2020-01→2026-02
238+
================================================================================
239+
240+
Rate profile: min=0.0003, median=0.0146, max=0.0609
241+
242+
Fixed 5:1:
243+
EITT residual: -0.0077%
244+
245+
Adaptive (base_ratio=5):
246+
Blocks: 15, mean size: 4.9
247+
EITT residual: +0.0284%
248+
249+
250+
================================================================================
251+
SUMMARY — ADAPTIVE TAP DECIMATION
252+
================================================================================
253+
254+
Core finding: d(CoDa)/dt serves as the TAP CONTROLLER.
255+
It determines the maximum safe compression ratio at each time step.
256+
257+
Mechanism:
258+
1. Compute |dB/dt| for each ILR balance at each time step
259+
2. Take max across balances → local rate of change
260+
3. Map rate to block size: slow → large blocks, fast → small blocks
261+
4. Apply variable-width geometric-mean decimation
262+
263+
This gives d(CoDa)/dt a CONCRETE OPERATIONAL ROLE:
264+
Not just 'exploratory temporal derivative'
265+
But the CONTROL SIGNAL that sets the balance transformer's taps.
266+
267+
Date: 2026-04-10

‎tools/pipeline/backblaze_coda_results_2026april9.txt‎

Lines changed: 74 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,74 @@
1+
================================================================================
2+
HUF BACKBLAZE — FULL CoDa + EITT + d(CoDa)/dt ANALYSIS RESULTS
3+
Run: 2026-04-09
4+
24 Monthly Snapshots · Jan 2024 – Dec 2025 · K=4 Carriers
5+
Tool: analyze_backblaze_coda.py
6+
================================================================================
7+
8+
KEY FINDINGS
9+
============
10+
11+
1. EITT CONFIRMED ON NON-ENERGY DATA
12+
Monthly entropy mean: 1.221620
13+
Annual entropy mean: 1.221955
14+
EITT residual: 0.0275% at 12:1 compression
15+
Decimation ladder: 0.0011% (2:1) → 0.0021% (3:1) → 0.0038% (4:1)
16+
→ 0.0107% (6:1) → 0.0275% (12:1)
17+
VERDICT: EITT holds. First non-energy domain confirmation.
18+
Comparison: Tighter than some energy results. Monotonic ladder.
19+
20+
2. RELAY CHAIN — ELECTRONIC IS THE SOLE DONOR
21+
Electronic: net Δρ = -1.70 pp (↓)
22+
Media: net Δρ = +0.79 pp (↑)
23+
Offline: net Δρ = +0.59 pp (↑)
24+
Mechanical: net Δρ = +0.32 pp (↑)
25+
Closure: Σ = 0.000023 (machine zero)
26+
Physical meaning: Electronic failures replaced by Media+Offline as fleet
27+
ages and technology shifts. Drives that used to fail electronically now
28+
fail through media degradation or go offline.
29+
30+
3. B2 PERFECT MONOTONICITY — 100%
31+
Balance B2 (Electronic vs Media+Offline) has ZERO sign changes in dB/dt
32+
across 23 intervals. dB2/dt is negative in every single month.
33+
B2 trajectory: +0.44 → -0.04 (crossed zero ~Oct 2025)
34+
The October 2025 zero-crossing marks when Electronic failures became
35+
less common than Media+Offline combined — a structural phase transition.
36+
37+
B1 (Mech vs rest): 59% monotonicity (9 sign changes)
38+
B3 (Media vs Offline): 59% monotonicity (9 sign changes)
39+
40+
Implication: B2's perfect smoothness is a concrete example of the temporal
41+
persistence that EITT requires. Smoothest balance = strongest EITT.
42+
43+
4. VELOCITY DECELERATION
44+
Peak velocity: 0.0697 (Apr→May 2024)
45+
Late 2025 range: 0.006–0.019
46+
The fleet composition is settling. Structural transition decelerating.
47+
Cumulative 24-month drift: d_A = 0.4799
48+
49+
5. d²B/dt² ON B2 — CONSTANT RATE DECLINE
50+
B2 acceleration is small and mixed-sign. The Electronic decline proceeds
51+
at roughly constant rate — consistent with fleet procurement cycles,
52+
not sudden failure events.
53+
54+
6. DUAL-METRIC AGREEMENT — 100%
55+
All 23 consecutive intervals show TV and Aitchison agreeing on direction.
56+
No trace-only or dominant-only signals detected. Every shift is structural.
57+
58+
SIGNIFICANCE FOR CODAWORK 2026
59+
==============================
60+
61+
- Directly answers Attack 2 ("You only tested energy data")
62+
- B2's 100% monotonicity is a visual proof of temporal autocorrelation
63+
- The relay chain has clear physical interpretation (fleet aging)
64+
- EITT residual (0.0275%) is competitive with energy domain results
65+
- Zero-sum relay verification (Σ=0.000023) confirms CoDa closure holds
66+
67+
FULL OUTPUT
68+
===========
69+
70+
[See console output saved separately or re-run analyze_backblaze_coda.py]
71+
72+
Governance: CGS-2 (n=3), GDoF 264. No new constants.
73+
Data: Backblaze Drive Stats, CC BY 4.0 equivalent
74+
HUF:1.1.8 | OP_MIN=0.51 | TOOL_MAX=0.49

0 commit comments

Comments
 (0)