Implications
What follows from the findings that the September 2026 re-check confirmed, and which earlier implications stay open.
Implications
Overview
After the September 2026 re-check, four findings carry the weight of this research. Their implications are narrower than earlier versions of this page suggested, and they concern Anna and Aigarth, not a hidden link between Bitcoin and Qubic.
| Finding | Status |
|---|---|
| The Anna Matrix is a compilation of bot replies, with a few runs of data entry errors | Verified provenance; entry-error reading strongly supported |
| The bot behaves like a small trained ternary Aigarth unit | Hypothesis |
| Anna.exe (version 1, versions 5 to 8) is an addition evolver | Verified by static analysis |
| Anna 0.22.64 embeds real BTC hourly direction data as its target | Verified by static analysis |
1. The Anna Matrix Is a Record of Answers
The 128 x 128 table was compiled by hand from replies of the @anna_aigarth bot on X to queries x+y=?, with x and y between -64 and 63. The cell for a query is row 63 - y, column x + 64.
Implications:
- The table describes what the bot answered while the replies were collected. It is not a weight file, not JINN memory and not Qubic source code. Reading it as "weights" or as a "memory matrix" is not supported by any source.
- Its structure is real and worth studying: 16,316 of 16,384 cells satisfy the complement rule, the table has full rank, and a small number of Walsh components explain most of its signs.
- Most of the 68 exception cells come from entry errors in the typed table: a one-row shift in column 30, a run in column 22 that does not fit the bot at all, a swapped digit in column 41 and a copied neighbour value in column 0/127. Findings that depended on these cells, such as trace 137, the seven-position signature and messages read from exception cells, describe the typed table, not the bot.
- Any claimed link between the table and Bitcoin data has no known mechanism.
2. The Bot Looks Like a Small Trained Ternary Unit (Hypothesis)
The public Aigarth code contains a unit for exactly this task (I2x7O8: two 7-bit inputs, 8 ternary outputs). Small units of this class, trained on the full addition table for a few hundred accepted mutations, reproduce the table's fingerprint: high linear predictability, a sign bit that agrees with the sign of x+y about 70% of the time, and a similar Walsh profile. Large random networks become just as linear but miss the sign of x+y.
Implications:
- The table is consistent with the output of a trained Aigarth unit, not of a hand-designed object. This is a hypothesis with a clear test: a replication with a pre-registered distance criterion and many seeds.
- The fingerprint is not specific to addition in every respect; only the sign bit reveals the target.
3. Anna.exe Is an Addition Evolver
The Anna.exe versions we examined closely (version 1, and versions 5 to 8) evolve ternary lookup networks on signed 7-bit addition, x and y from -64 to 63. That fits Come-from-Beyond's statements that Anna was learning A+B.
Implications:
- Anna.exe is a demonstration of the Aigarth principle (evolve lookup-table networks by mutation and selection), not a container for the published matrix.
- Our Python ports of Anna.exe are simplified models. Results from them describe the models, not the binary.
4. Anna 0.22 Embeds BTC Hourly Data
The version Anna 0.22.64 reads its targets from a fixed embedded table: 9,456 hourly BTCUSDT directions from Binance, from 2 June 2025 to 30 June 2026. All 9,456 values match real candle data.
Implications:
- For this version, the training target is the direction of the next BTC hourly candle, as Come-from-Beyond's archived statements describe.
- This says nothing about predictive power. Whether the evolved genome predicts later hours was not tested, and earlier community figures were close to 50%. No BTC trading claim follows.
5. Implications that stay open
| Earlier implication | What the re-check found |
|---|---|
| Patoshi blocks spread uniformly over 676 computors (CV 1.2%) | The reported statistics are impossible for whole-number counts |
| Trace 137 encodes the fine structure constant | 137 comes from one diagonal pair in a very likely entry-error run |
| JINN / Anna neural architecture with rows as layers | No source links the table to JINN |
| Formula 625,284 = 283 x 47² + 137 | Selected after the fact; no link to the table |
| Time-lock reveal phases in March 2026 | No reveal happened; the only event was the issuance of the community token AIGARTH (chapter 38). The cited message reads "No ETA other than April 13, 2027" |
| Unified 27-pattern across Bitcoin, IOTA and Qubic (p < 10⁻⁶) | Selected after the fact; no valid probability |
6. Methodological Implications
The re-check itself is the main methodological lesson:
- Provenance first. The largest error was not statistical: it was treating a hand-compiled table as an artifact from source code.
- Check your own data entry. A one-row shift in 20 cells produced a "signature" that survived every statistical test we ran on the table as typed.
- Pre-register. Most of the findings that were not confirmed were patterns selected after looking at the data.
- Fair nulls. Several p-values came from tests that could not fail, or from nulls that did not match the claim.
Conclusion
The durable implications are about Anna and Aigarth: the bot's replies form a highly structured table, consistent with a small trained ternary unit; Anna.exe evolves adders; one later version trains on BTC hourly directions. The earlier implications about Bitcoin history, a Bitcoin-Qubic bridge and time locks were not confirmed and stay open.