MEMORIA VIVUM — OFFICIAL TEST REPORT Test ID: MV-011 (Meaningfulness) Species: Memoria vivum (The Living Archive) Creator: Wagg · Owner: Family Tree Trust Date: 2026-09-25 System: Diamond Server (Python 3, runnable engine) Document Type: Full Technical Record PURPOSE OF TEST Test MV-010 reported a 69.2% verification rate on real Zenodo research papers. That number counted how many papers had enough connections to be verified. It did not ask whether those connections were real. Two papers that both contain the word "brain" will connect — even if one is about brain surgery and the other is about brain waves. They are not related. The connection exists, but it means nothing. This test asks the harder question: of the connections the species made, how many are real relationships, not coincidences? METHOD 1. Fetched the same kind of data from Zenodo (neuroscience search) 2. Fed 13 papers into the v1.8 engine 3. Ran 5 growth cycles to build connections 4. Collected every unique connection the species made 5. For each connection, counted how many meaningful words the two papers actually share SCORING RULES | Score | Meaning | | STRONG | 3 or more shared words | | MODERATE | 2 shared words | | WEAK | 0 or 1 shared word — coincidence | The hypothesis: if the species is meaningful, most of its connections will be STRONG or MODERATE. If it is just counting word overlaps, most connections will be WEAK. ACTUAL RESULTS | Connection Quality | Count | Percentage | | STRONG (3+ shared words) | 1 | 5.9% | | MODERATE (2 shared words) | 1 | 5.9% | | WEAK (0-1 shared words) | 15 | 88.2% | | Total connections | 17 | 100% | Real connections (moderate or better): 2 / 17 (11.8%) VERDICT COINCIDENTAL WHAT THIS SHOWS 1. The species made 17 connections between real papers. 2. Only 2 of those 17 connections represent a real relationship. 3. 88.2% of the connections are coincidences — two papers that happen to share one word without actually being related. 4. The 69.2% verification rate from MV-010 was counting — it verified papers because they had connections, not because those connections were meaningful. THE CORE FINDING The species was measuring quantity, not quality. It treated every shared word as equally meaningful. "Brain" and "brain" looked the same as "cognition" and "cognition." This is a fundamental limitation of the current Weaving Core. It does not distinguish between a word that appears in every paper (and therefore means nothing) and a word that only appears in two papers that are actually related. HONEST BOUNDARY This test proves that most of the species' current connections are coincidental, not meaningful. It does NOT prove the species is broken. It proves the species does not yet distinguish between common words and informative words. Two things are not measured here: - Whether a future fix would produce real connections - Whether the connection quality on other data types (not research papers) behaves differently LESSONS LEARNED 1. Counting connections is not the same as measuring meaning. 2. A word that appears everywhere connects everything — and connects nothing. 3. The species must learn to weight words by how informative they are, not by whether they simply match. 4. This is the most important failure in the project. It exposed a weakness that a simple verification count would have hidden. NEXT STEPS - Build v1.9: add word weighting to the Weaving Core so common words count less than rare, informative words. - Re-run MV-010 and MV-011 with v1.9 to see if meaningfulness rises above 70%. OUTPUT ARTIFACT Script: scripts/test_mv11_meaningfulness.py Text results: results/test_mv11_meaningfulness.txt Signed: Wagg · Owner: Family Tree Trust · Date: 2026-09-25 © 2026 Family Tree Trust. All rights reserved. Distributed by CTW, LLC.

MEMORIA VIVUM — OFFICIAL TEST REPORT Test ID: MV-012 (Meaningfulness Retest with v1.9) Species: Memoria vivum (The Living Archive) Creator: Wagg · Owner: Family Tree Trust Date: 2026-09-25 System: Diamond Server (Python 3, runnable engine) Document Type: Full Technical Record PURPOSE OF TEST Test MV-011 revealed that only 11.8% of the species' connections on real research data were meaningful. The rest were coincidences. Version 1.9 added Word Weighting to the Weaving Core — an attempt to fix this. The idea was that rare, informative words would count more than common words, so the species would stop connecting on meaningless overlaps. This test reruns MV-011's exact measurement on the fixed engine to see whether the fix worked. METHOD 1. Fetched the same kind of data from Zenodo (neuroscience) 2. Fed 13 papers into the v1.9 engine 3. Ran 5 growth cycles 4. Collected every unique connection 5. Measured connection quality: STRONG, MODERATE, WEAK 6. Compared to MV-011 THE v1.9 CHANGE Word Weighting: every word in the species now carries a weight based on how rare it is. The idea was that a connection built on rare words would score higher than one built on common words. The IDF formula was used: weight = log(total_items / (frequency + 1)) + 0.1 Rare words get higher weight. Common words get lower weight. A minimum connection score of 0.5 was set — anything below that would not count as a connection. ACTUAL RESULTS | Connection Quality | Count | Percentage | | STRONG (3+ shared words) | 1 | 5.9% | | MODERATE (2 shared words) | 1 | 5.9% | | WEAK (0-1 shared words) | 15 | 88.2% | | Total connections | 17 | 100% | Real connections: 2 / 17 (11.8%) VERDICT FAILED — the fix did not improve meaningfulness. COMPARISON TO MV-011 | Metric | MV-011 (before) | MV-012 (after v1.9) | | Total connections | 17 | 17 | | Strong | 1 (5.9%) | 1 (5.9%) | | Moderate | 1 (5.9%) | 1 (5.9%) | | Weak | 15 (88.2%) | 15 (88.2%) | | Real connections | 11.8% | 11.8% | Identical numbers. The Word Weighting change had zero effect on the outcome. WHY THE FIX FAILED The IDF formula requires a large collection of documents to work. It measures how rare a word is across the entire set. With only 13 papers, there is not enough data to distinguish a common word from a rare one. With 13 items, every word looks roughly equally rare. The weighting signal is drowned out by the small sample size. This is a real, honest finding. It is not a failure of the idea — it is a failure of applying that idea at the wrong scale. WHAT THIS SHOWS 1. The v1.9 Word Weighting change did not improve meaningfulness on this dataset. 2. The result was identical to MV-011 — 11.8% real connections. 3. The cause is a small sample size. IDF needs hundreds or thousands of documents to work as intended. HONEST BOUNDARY This test proved the fix failed. It did not prove the fix is impossible — only that this implementation, at this scale, does not work. With a much larger corpus, weighting might help. That remains to be tested. Two things were not measured here: - Whether a different weighting scheme would work at this scale - Whether a different fix (like a stricter connection rule) would succeed LESSONS LEARNED 1. Small datasets defeat statistical weighting methods. IDF needs scale to find signal. 2. Real fixes must match the data the species actually has — not the data it might have someday. 3. Failure reveals direction. The species is not broken. It is applying the wrong tool at the wrong scale. 4. The next fix should be simple and independent of sample size. NEXT STEPS - Build v1.10: require two or more shared words before counting a connection. Simple threshold, no statistics needed. - Re-run MV-012 as MV-013 with the threshold change. - Prediction: meaningfulness rises sharply because weak connections are eliminated at the source. OUTPUT ARTIFACT Script: scripts/test_mv12_meaningfulness_v2.py Text results: results/test_mv12_meaningfulness_v2.txt Signed: Wagg · Owner: Family Tree Trust · Date: 2026-09-25 © 2026 Family Tree Trust. All rights reserved. Distributed by CTW, LLC.