The alignment explainer established that best-fit pairwise offsets between same-prefix EAM messages cluster at multiples of 44. But why multiples of 44? Two hypotheses:
The discriminating test: when two messages align at offset ±44, do all block pairs match at similar rates, or does one block pair dominate?
Each example shows a message pair whose best sliding alignment is at a multiple of 44. The overlapping region is broken into 44-character blocks and compared block-by-block. Green highlighting marks shared blocks (high match rate); dimmed blocks are unrelated content that happens to be adjacent.
In every case, the match rate is concentrated in one or two block pairs; the rest fall to near-random levels (~3% expected for a 32-character alphabet). A periodic cipher would produce roughly uniform match rates across all block pairs.
| Prefix | Offset | Block pair | Match rate | Shared? |
|---|
The evidence shows that EAM messages are composed of sequences of 44-character blocks. Messages with the same prefix share some blocks in common, but these shared blocks can appear at different positions within each message. A shift of ±44 brings a shared block into alignment, producing a spike in the match count — but only for that specific block pair.
This rules out a simple periodic cipher with period 44: such a cipher would re-align its key at every ±44 shift, making all overlapping positions equally likely to match, not just one block-sized region.
The picture that emerges: messages are built from a vocabulary of 44-character blocks, with block insertion or deletion shifting shared blocks to different positions across messages.