Run · Voynich Manuscript
T41 diagnostic: the true key's basin has higher likelihood than any random EM restart (a search problem)
Gold text is used here only to build the true channel for the diagnostic; no key on the manuscript is chosen this way.
Result
| Run | Measured | Result | Against | Decision | Date |
|---|---|---|---|---|---|
| Run | Training log-likelihood and key accuracy: EM started from the true key vs 10 random restarts, on the planted control | one sign per letter, no nulls: EM from truth -40,505 (key 0.777), best random restart -41,315 (key 0.186). Full T39 setting: EM from truth -53,462 (key 0.851), best random restart -53,957 (key 0.250) | The true channel itself, before EM: -41,654 and -55,406 (key 0.777, 0.850 · the rest are nomenclator words of the control's plaintext that the Latin element model doesn't have) | The model is right and the search is wrong: a key close to the truth is a better optimum of the lattice likelihood than anything random restarts find, by 500-800 nats. This justifies spending the fresh-idea slot on annealing and population annealing, not on a new model. |
What it used and found
MTH
Latin genre texts
DATDataset
Parameters
pw = 0.25, alpha = 0.05, control = Pliny NH 16, period.generate seed 77, em_iterations = 60, restart_seeds = 1000-1009, random_restarts = 10
Cite it
Mysteries Department. (2026). T41 diagnostic: the true key's basin has higher likelihood than any random EM restart (a search problem) [Run RUN0DCG2Q30P808]. https://mysteriesdept.com/r/RUN0DCG2Q30P808
@misc{mysteriesdept_run0dcg2q30p808,
title = {T41 diagnostic: the true key's basin has higher likelihood than any random EM restart (a search problem)},
author = {{Mysteries Department}},
year = {2026},
note = {Run, catalog key RUN0DCG2Q30P808},
url = {https://mysteriesdept.com/r/RUN0DCG2Q30P808}
}