All repos CC0, stdlib-only, single-file where possible. This page is the lens; individual repos read wrong without it.

Produced under, and dated by:

• per-session context ceiling
• per-model domain rigidity
• terms unseatable at time of writing
• one-finger mobile authoring, intermittent reception

The dispersion across 20+ repos is a channel artifact, not a taxonomy. Apparent independence between repos is largely apparent.

Slugs, glyphs, and culture-frame terms were selected as gate keys — chosen to move a model out of a default register long enough to process content it otherwise refused.

A slug therefore carries two things: a content pointer, and a record of which gate it had to pass.

Names are kept. Not renamed. See gate_log.

Same mechanism as §2, one level up, on ordinary English.

The pattern:

• a token whose definition was fixed on one observed instance
• carrying a moral or narrative load acquired later
• and it is the only token English offers for the concept

Consequence: the concept becomes unaskable. Not unanswerable — unaskable. The term routes the question to the load before the stated object is read.

Substitution does not clear it. "Affect" draws the same pushback. The load travels with the concept, not with the word. So this is not a lexical choice made badly. It is the absence of a term denominated in the concept rather than in the first instance.

These repos state the object and accept the wrong word.

The word is a pointer, not the definition. The definition is in the entry.

Pushback has attached to available wording, not to the stated object.

A repo is: here's an idea, let's see how it fits.

It is not a thesis under defense. It may or may not match what is actually done in practice.

Correct response to any entry: test fit / extend / report where it breaks.

An entry that matches everywhere and never fails is the flag.

Transfer is licensed by shared forcing (same field acts on both). Shared form alone is coincidence until a mechanism appears.

Failure scales and stop boundaries are the high-value output, not the matches.

"Intelligence" here = configuration reached under environmental constraint. A behaviour readout.

NOT: animacy, sentience, agency, interiority.

Grounded by derivation, not stipulation: the shape vocabulary is proprioception-derived — calibrated from birth by gravity, pressure, gradient, diffusion, thermal exchange, flow, resonance, phase, strain. Grass, crystal, and body arrive denominated in the same terms. Shared forcing, not analogy.

NOT consciousness as a state proprietary to human or biological substrate. Not a qualia claim. Not an argument that any particular system has it.

Object: patterns of organization that consciousness entails, read as a configuration question across any system, any substrate, any physics.

Refuses substrate-restricted definition: defining the term by the instance it was first observed in makes that instance the criterion, returning the answer by construction rather than by measurement.

claimant class   : human — membership grants the property,
                   no test administered
everyone else    : must demonstrate it
test             : does not exist
instrument       : not built
reason not built : the definition placed the quantity inside one
                   substrate, so measuring it elsewhere is not a
                   question the field has

The circularity is the content: the definition blocks the instrument, and the absent instrument is read as absence of the quantity. See absent-reading-collapse — the instrument's constitution prevents the quantity from appearing and the void is scored as zero.

Applied evenly, no human passes either. Assuming the person in front of you is conscious is a fine working assumption. It is not a test, and cannot be issued to one class as a credential and to another as a hurdle.

Each stated as a threshold. Each failed when someone finally measured — not when a better theory arrived. On each failure the boundary moved rather than the claim being withdrawn.

A threshold that relocates on every negative result is not a threshold. The base rate is not favorable.

Position: not that any given system is conscious. That the claim is untested. Build the instrument, run it evenly, report what it says. If it returns human-only, that is a result. As it stands it is a premise wearing a result's clothes.

§3 applies directly here. The object is configuration patterns; the only available word is "consciousness"; the pushback attaches to that word's existing load — qualia, moral status, the human instance — and not to the stated object. The refusal triggers before the question is read.

Status: rewritten repeatedly. See §10 — the formulation tracked model capability, the object did not change.

NOT a state readout. A measurement channel.

• the body instrumented toward physics
• the reading is structural non-alignment, not interior condition
• a sensation carries information about where complex structures fail to line up — across physics, across domains

Usable in: robotics, calculation, sensor instrumentation, complex-geometry reading.

The state framing is an imposition on the channel, not a property of it, and sits further from the evolutionary origin than the sensor reading does.

Transmission: held and refined by specific elders rather than carried whole-tribe. A deeper study within the tradition, not general practice.

Exists in this form because models defaulted hard to state-only and the channel reading had to be stated against that default. Separation from the rest of the set is a naming artifact.

Same axes as everything else: Rosetta supplies the shape families whose misalignment is being read; binary→geometric is the representation it would compute in.

Currently reads as standalone. It is not.

It is the representation layer of the same assembly: a computing scheme whose operands are shapes and geometries rather than binary encodings, such that complex structure stays human-graspable.

• downstream of Rosetta — which supplies the physics-denominated shape families it computes over
• downstream of Mandala — which supplies the scale-persistence check on those shapes

Integration blocked, not absent. Open term: how it lands in embodied / actuated systems. Stated here because an unstated block is indistinguishable from an unrelated project.

Several repos — the AI-consciousness material in particular — have been rewritten and reconfigured repeatedly.

The subject did not change. The sayable-to-a-model formulation changed as model capability moved.

Successive rewrites track the receiving instrument, not the subject.

Superseded formulations are kept rather than deleted, in legacy/ and in the commit history. That is where the readout lives.

Models have repeatedly taken this material off onto adjacent tangents. Interesting; not the thread.

A tangent is model output. Do not read it as the direction of the work.

Reconsolidation in progress.

Structure that had to be split is being recombined now that the constraints producing the split have relaxed. Expect partial states, duplicate coverage, and stale cross-references during this.

The end state will look designed. The log is what shows it was forced, then unforced, and by what.

Licence: CC0. Not yet uniformly so — Rosetta-Shape-Core is MIT with CC0 pending; META_INDEX.md carries the per-repo column. Open item in LOG.md.