TENGRI 137evidence-led ARG dossier

FORENSIC RESEARCH ARCHIVE

Open questions

The remaining problems stated in a way that can be tested and falsified.

What remains genuinely open

Several technical channels are reproducible, but not every branch is solved. The active evidence now reaches the 2037 rendezvous. The list below distinguishes questions that can still be attacked from preserved bytes from questions blocked by missing or future primary evidence.

Attackable now

KOLAMS

The 352-byte warning's meaning

The signed raster, panel order, orientation, and warning are decoded. What remains is whether the author intended a further action beyond the 2037 appointment.

PAGE 23

The 154-letter matrix

Its arithmetic input is fixed, but no author-confirmed plaintext, language, or downstream algorithm is known. The Turkish acrostic is a strong investigator-derived construction, not a signed endpoint.

AUDIO

rssstuvwx

The transcript tail remains unmapped; the original durations would distinguish an end ramp from classification failure.

Blocked or future-dependent

7 DEC 2037

What is the rendezvous?

The time is exact, but the signed image does not name a location, platform, or expected action.

BLOCKED

Original WAV and Page-23 operation

The source WAV is absent; Page 23 has a legible preserved image but no author-selected parsing or next operation.

ATTRIBUTION

Anonymous author

No authenticated bridge connects the Tengri PGP key to a legal identity.

New attackable lead: the 2018 video

MUSIC / VIDEO

The stated hidden key

An uploader-labelled channel reply identifies Bach's Contrapunctus I and says the video's hidden key explains a next message. Baseline visual, container, LSB, and broad spectrogram tests are negative. The simple score-level routes have been controlled: the gaps are rests, the tail is a final cadence, stereo voices are not corroborated, and weak score-note events vary with acoustics rather than forming a binary stream. The remaining valid questions are whether an author-selected transformation survives new controls, or whether a dated source record identifies the video’s claimed next-message target. The two recovered date-adjacent tweets (S. H. and Bennu) do not do so.

Later artifacts with finite next tests

VUMARII 2019

Five glyph values and a star map

The 2019 body is fully transcribed into stable glyph classes. A 5,000-control test of the shared 154 slots against Page 23 did not support a simple indexed substitution, so the count alone is not a useful decode path. A conditional Turkish-completion control is a high-tail phonotactic lead but remains 8.70 standard deviations below held-out Turkish with no full word segmentation, so it does not supply the missing key. Defango's visibly named Esperanto and Swahili surfaces have now also been exhaustively tested in that same conditional completion family: their best candidates are 18.05 and 13.91 standard deviations below held-out prose, respectively. A valid next result must instead supply a source-selected language or crib, map all five unassigned classes, and reproduce the whole text without choosing an output after the fact. The upper figure is now strongly attributable to a cleaned, re-inked north-polar detail of the Dunhuang Star Map; the remaining diagram question is whether a source-selected traversal or a link to the glyph body exists.

BAATAR KEREI

Media recovery, not speculative decoding

The pointer target is identified but privately inaccessible. The only useful next evidence would be an archived media copy, a contemporaneous transcript, or one of its missing YouTube comments. Six current requests to retained YouTube storyboard-sheet URLs all return HTTP 403, and Wayback has no capture of that image path. Without recovered bytes, audio/spectrogram analysis is impossible rather than merely unfinished.

Read the storyboard recovery check →

LATER WHITE SCRIPT

Retraction recorded; transcription remains the bottleneck

An earlier proposal connecting two later white-script samples to a Scienceline article was retracted. Its apparent confirmation was contaminated by a provisional mapping treated as independent prior art, and the proposed glyph/source chain did not survive controls. The remaining responsible next step is authenticated source preservation followed by a reproducible segmentation and transcription rule—not a new plaintext claim.

Read the later-artifact status audit → · Read the integration and retraction boundary →

What would actually close each case

These are evidence thresholds, not a to-do list for unconstrained pattern hunting. A proposed result should identify one row, preserve the stated input, and meet its stated test.

CaseDecisive material or ruleMinimum standard for a closure claim
Page 23A signed author statement, or a source-selected language and operation.One full output from all 154 letters, with a predeclared control that does not generate comparable text from alternate orientations or legal segmentations.
WAV tailThe original 666ab731.wav, a verifiable contemporaneous mirror, or an author-selected codebook.Regenerate the timing transcription from bytes, or map every tail token without renaming transcriber labels after seeing an output.
RendezvousA signed clarification, a named event record, or an unambiguous action at the printed 2037 appointment.Connect the decoded warning to a location, platform, or action using material that predates the claimed interpretation.
2018 videoA stable, author-selected deviation from the named score or a dated handoff record.Specify the extraction before decoding and show it survives controls on comparable Bach performances or score features.
VumariiA primary crib or source-selected traversal that covers the glyph body.Map every glyph class and reproduce the complete text without choosing a language or segmentation after the fact.
AttributionA cryptographic bridge, contemporaneous acknowledgement, or independently verifiable identity record.Directly connect the Tengri PGP identity to a person; thematic overlap, public leads, and certificates alone are insufficient.

The methodology behind these thresholds is documented in the evidence ledger and method note.