allaun/SilverSight
1
0
Fork 0
mirror of https://github.com/allaunthefox/SilverSight.git synced 2026-07-31 01:25:21 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

docs: standalone Cold Reviewer Protocol document

Browse source 
Explicit two-gate protocol (Arithmetic + Structural) with step-by-step
calculator verification for each invariant.  Covers goals, methodology,
extension rules, and decision procedure.
This commit is contained in:
allaun 2026-06-26 23:41:14 -05:00
parent 095474d572
commit 89b8f4b243
1 changed files with 389 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

389
docs/reviews/COLD_REVIEWER_PROTOCOL.md Normal file
View file

@ -0,0 +1,389 @@
# Cold Reviewer Protocol
**A formal verification protocol for independently checking mathematical
claims using only discrete, calculator-verifiable computation.**
---
## 1. Goals
### Primary goal
Enable a reviewer with **no domain expertise** (no Lean, no braid theory,
no Temperley–Lieb algebras, no differential geometry) to independently
determine whether the core claims of the Unified Covariant Field Theory
are mathematically sound.
### Secondary goals
1. **Isolate domain knowledge.** All expert-level claims (geometric
conjectures, spectral theory, tensor categories) are explicitly gated
behind the discrete arithmetic layer. A reviewer who passes the
Arithmetic Gate has already validated the foundation without needing
to understand the superstructure.
2. **Eliminate tautologies.** Every invariant must be checked against
independent calculation, not against another statement in the same
manuscript. No invariant may depend on another invariant.
3. **Produce a bounded receipt.** A reviewer's final pass/fail verdict
is a binary decision: all four invariants hold and no structural red
flag is present ⟶ **PASS**. Otherwise ⟶ **FAIL**.
4. **Provide a formal upgrade path.** Each invariant maps directly to a
Lean theorem verified by `norm_num` or `dec_trivial`. A reviewer who
trusts the Lean kernel can skip manual calculation and run
`lake build SilverSight` (3307 jobs, 0 errors).
---
## 2. The Two-Gate Structure
```
┌────────────────────────────────────────────────────────────────────┐
│ MANUSCRIPT UNDER REVIEW │
│ UnifiedCovariant.lean — Unified Covariant Field Theory │
├────────────────────────────────────────────────────────────────────┤
│ │
│ GATE A: ARITHMETIC GATE │
│ ──────────────────────── │
│ Verify I₁, I₂, I₃, I₄ independently. │
│ Reject immediately if any fails. │
│ │
│ ↓ (all pass) │
│ │
│ GATE B: STRUCTURAL GATE │
│ ──────────────────────── │
│ Check that none of the three red-flag claims appear anywhere │
│ in the manuscript. Reject if any is present. │
│ │
│ ↓ (no red flags) │
│ │
│ GATE C: BUILD GATE │
│ ──────────────────────── │
│ (Optional, for Lean-literate reviewers) │
│ Run `lake build SilverSight`. Must report 0 errors. │
│ │
│ ↓ (0 errors) │
│ │
│ ✅ VERDICT: PASS │
│ Core discrete claims are sound. Higher-level geometric claims │
│ (Layer 3) are deferred but structurally harmless. │
└────────────────────────────────────────────────────────────────────┘
```
---
## 3. Gate A — Arithmetic Gate
### Instructions
You need: **a calculator** (or pen and paper, or a Lean kernel).
For each invariant, perform the required computation. The answer is
predetermined. If your result matches, the invariant passes. If your
result differs, **reject the manuscript** — the foundation is unsound.
**DO NOT** skip any invariant. **DO NOT** accept "by symmetry" or
"by a similar argument" as a substitute for independent calculation.
---
### I₁. Golden-Ratio Identity
**Claim:** \(\phi = \frac{1 + \sqrt{5}}{2}\) satisfies \(\phi^2 - \phi - 1 = 0\).
**How to verify:**
1. Compute \(\sqrt{5} \approx 2.236067977\ldots\)
2. Compute \(1 + \sqrt{5} = 3.236067977\ldots\)
3. Divide by 2: \(\phi = 1.618033988\ldots\)
4. Compute \(\phi^2 = (1.618033988)^2 = 2.618033988\ldots\)
5. Compute \(\phi^2 - \phi - 1 = 2.618033988 - 1.618033988 - 1 = 0\).
**Alternative (exact symbolic):**
\[
\phi^2 - \phi - 1 = \frac{(1+\sqrt{5})^2}{4} - \frac{1+\sqrt{5}}{2} - 1
= \frac{1 + 2\sqrt{5} + 5}{4} - \frac{1+\sqrt{5}}{2} - 1 = 0.
\]
**Lean reference:** `golden_identity` (line 86).
---
### I₂. Fixed-Point Gap
**Claim:**
\[
\sigma = \frac{9984}{65536} = \frac{39}{256},\qquad
\tau = \frac{1}{7},\qquad
\sigma - \tau = \frac{17}{1792} > 0.
\]
**How to verify:**
1. Reduce \(\frac{9984}{65536}\): divide numerator and denominator by 256.
\[
\frac{9984}{65536} = \frac{9984 \div 256}{65536 \div 256} = \frac{39}{256}.
\]
*(Check: 256 × 39 = 9984, 256 × 256 = 65536. ✓)*
2. Compute the difference with common denominator 1792:
\[
\frac{39}{256} - \frac{1}{7} = \frac{39 \times 7}{256 \times 7} - \frac{1 \times 256}{7 \times 256}
= \frac{273}{1792} - \frac{256}{1792} = \frac{17}{1792}.
\]
3. Check positivity: \(17 > 0\). ✓
**Also verify:** \(\sigma > \tau\) (since \(17 > 0\)).
**Lean reference:** `spectral_gap_positive` (line 99).
---
### I₃. Fibonacci Values
**Claim:** \(F_7 = 13\) and \(F_8 = 21\).
**How to verify:** Run the recurrence from the definition.
| \(n\) | \(F_n\) | Calculation |
|------|---------|-------------|
| 0 | 0 | (by definition) |
| 1 | 1 | (by definition) |
| 2 | 1 | \(F_0 + F_1 = 0 + 1\) |
| 3 | 2 | \(F_1 + F_2 = 1 + 1\) |
| 4 | 3 | \(F_2 + F_3 = 1 + 2\) |
| 5 | 5 | \(F_3 + F_4 = 2 + 3\) |
| 6 | 8 | \(F_4 + F_5 = 3 + 5\) |
| **7** | **13** | \(F_5 + F_6 = 5 + 8\) |
| **8** | **21** | \(F_6 + F_7 = 8 + 13\) |
Both values match the claim.
**Lean reference:** `fibonacci_dims` (line 106), `fib7_is_13` (line 232).
---
### I₄. Sidon Uniqueness
**Claim:** For \(a,b,c,d \in \{0,1,2,3,4,5,6,7\}\),
\[
2^a + 2^b = 2^c + 2^d \;\Longrightarrow\; \{a,b\} = \{c,d\}.
\]
**How to verify — method 1 (binary expansion):**
1. Write each integer \(2^k\) in binary: it is a 1 followed by \(k\) zeros.
\[
2^0 = 1_2,\; 2^1 = 10_2,\; 2^2 = 100_2,\; \ldots,\; 2^7 = 10000000_2.
\]
2. The sum \(2^a + 2^b\) in binary has:
- **Case \(a = b\):** a single 1 in position \(a+1\) (carry).
- **Case \(a \neq b\):** exactly two 1 bits, at positions \(a\) and \(b\).
3. Binary representation is unique. Therefore if two sums are equal, the
sets of bit positions must be identical.
**How to verify — method 2 (exhaustion, 8⁴ = 4096 cases):**
Check all quadruples \((a,b,c,d)\). If \(2^a + 2^b = 2^c + 2^d\) then
\(\{a,b\} = \{c,d\}\). This is a finite computation. With a computer:
```python
for a in range(8):
for b in range(8):
for c in range(8):
for d in range(8):
if 2**a + 2**b == 2**c + 2**d:
assert {a,b} == {c,d}
```
No counterexample exists.
**Lean reference:** `sidon_unique` (line 110), verified by `dec_trivial`.
---
## 4. Gate B — Structural Gate
### Instructions
Read through the entire manuscript. If **any** of the following three
claims appears verbatim or in spirit, **reject**.
---
### Red Flag 1: \(J^2 = -I\)
| ✗ WRONG | ✓ CORRECT |
|---------|-----------|
| \(J^2 = -I\) | \(J^2 = J + I\) |
**Why it matters:** \(J = \phi \cdot \mathrm{id}_V\) satisfies the
golden-ratio polynomial \(x^2 - x - 1 = 0\). Its eigenvalues are
\(\phi\) and \(-1/\phi\), not \(\pm i\). It is **not** an almost-complex
structure. Claiming \(J^2 = -I\) would make \(J\) a complex structure on
a real vector space, which changes the entire geometric interpretation.
**How to check:** Find the definition of \(J\) (or `goldenEndomorphism`)
and verify its defining relation. The correct relation is \(J^2 = J + I\).
---
### Red Flag 2: \(\Delta_7\) is Kähler
| ✗ WRONG | ✓ CORRECT |
|---------|-----------|
| \(\Delta_7\) is Kähler | \(\mathbb{CP}^7\) is Kähler (simplex is 7-real-dimensional, odd) |
**Why it matters:** A Kähler manifold must have even real dimension.
The open simplex \(\Delta_7 = \{p \in \mathbb{R}_{>0}^8 \mid \sum p_i = 1\}\)
has dimension \(8 - 1 = 7\) (odd). Kähler on \(\Delta_7\) is impossible.
**How to check:** Find any claim that a Kähler structure exists on
\(\Delta_7\) or on a 7-dimensional (or odd-dimensional) manifold.
If the manuscript refers to \(\mathbb{CP}^7\) (real dimension 14) instead,
this red flag is avoided.
---
### Red Flag 3: \(\dim(\mathrm{TL}_7) = 13\)
| ✗ WRONG | ✓ CORRECT |
|---------|-----------|
| \(\dim(\mathrm{TL}_7) = 13\) | \(\dim(\mathrm{TL}_7) = C_7 = 429\) |
**Why it matters:** The \(n\)-th Catalan number is
\[
C_n = \frac{1}{n+1}\binom{2n}{n},
\qquad
C_7 = \frac{1}{8}\binom{14}{7} = \frac{3432}{8} = 429.
\]
The value 13 is the Fibonacci integer \(F_7\), which arises only in the
specialized Temperley–Lieb quotient at \(q = e^{i\pi/5}\) (Fibonacci
anyon model). Confusing 13 with 429 is a dimension error of factor ~33×,
which invalidates any spectral or geometric argument that depends on it.
**How to check:** Find any claim about \(\dim(\mathrm{TL}_7)\), the
dimension of the Temperley–Lieb algebra on 7 strands. If it is 13,
reject. If it is 429 (or the Fibonacci quotient is explicitly named),
this red flag is avoided.
---
## 5. Gate C — Build Gate (Optional)
For reviewers with access to Lean 4 and Mathlib:
```bash
cd /home/allaun/SilverSight
lake build SilverSight
```
**Expected result:** 3307 jobs, 0 errors.
If the build fails, the manuscript has a formalization error.
---
## 6. Extending the Protocol
### Adding a new invariant to Gate A
Every new Layer-1 invariant must satisfy:
1. **Independence.** No invariant may reference another invariant's
conclusion. Each must be checkable from first principles.
2. **Finiteness.** The verification must be a finite computation:
rational arithmetic, integer arithmetic, or finite case analysis
(dec_trivial). No limits, no infinite series, no analysis.
3. **Lean correspondence.** Each invariant must have a corresponding
Lean theorem verified by `norm_num` or `dec_trivial`.
4. **Documentation in this document.** Add a new subsection with:
- The exact mathematical claim
- A step-by-step calculator verification procedure
- The Lean reference
### Adding a new red flag to Gate B
Every new structural red flag must:
1. Be an **unambiguously wrong statement** that a non-expert could
plausibly write.
2. Have a clear correction and a brief explanation of why the wrong
version is harmful.
3. Be listed in the Structural Gate table.
### Adding a new layer
The protocol supports exactly three layers:
| Layer | Content | Gate | Standard |
|-------|---------|------|----------|
| 1 | Discrete foundations | Gate A | 0 sorries |
| 2 | Mechanical theorems | Gate B (transitively) | 0 sorries |
| 3 | Geometric conjectures | Deferred | `sorry` permitted |
A new layer must be assigned to one of these three. No "Layer 1.5" or
"Layer 2b" may bypass Gate A. If a claim is not discrete and finite,
it must be Layer 3 or be restated in discrete form.
---
## 7. Decision Procedure
```
For each invariant I₁–I₄:
verify independently
if any fails → REJECT (Arithmetic Gate fail)
For each red flag R₁–R₃:
check manuscript
if any is present → REJECT (Structural Gate fail)
Optionally:
run `lake build SilverSight`
if errors → REJECT (Build Gate fail)
Otherwise → PASS
```
### What PASS means
The four discrete invariants are mathematically sound. No obvious
structural error is present. The manuscript is ready for expert review
of the Layer 3 geometric conjectures.
### What FAIL means
The manuscript has a foundational error. Corrections must be made to
the offending invariant or red flag before any higher-level claims can
be evaluated.
---
## 8. Quick Reference
### Arithmetic Gate (I₁–I₄)
| ID | Claim | Verification | Lean |
|----|-------|-------------|------|
| I₁ | \(\phi^2 - \phi - 1 = 0\) | Expand \((1+\sqrt{5})^2/4\) | `golden_identity` |
| I₂ | \(\sigma - \tau = 17/1792 > 0\) | Common denominator 1792 | `spectral_gap_positive` |
| I₃ | \(F_7 = 13,\; F_8 = 21\) | Run recurrence to term 8 | `fibonacci_dims` |
| I₄ | Sidon uniqueness | Binary expansion uniqueness | `sidon_unique` |
### Structural Gate (Red Flags)
| ID | Wrong claim | Correct |
|----|-------------|---------|
| R₁ | \(J^2 = -I\) | \(J^2 = J + I\) |
| R₂ | \(\Delta_7\) is Kähler | \(\mathbb{CP}^7\) is Kähler |
| R₃ | \(\dim(\mathrm{TL}_7) = 13\) | \(\dim(\mathrm{TL}_7) = C_7 = 429\) |