01
Claim
What the architecture is supposed to accomplish and the condition under which the claim would be false.
Architecture Decision Evidence
Evidence before commitment.
Architecture claims become decision-grade only when the evidence can change a commitment. Living Cipher separates public proof from protected implementation machinery so that technical diligence can be serious without giving away the computer.
Measured Evidence
Implementation evidence that constrains the architecture.
The public layer demonstrates that architecture decisions are grounded in actual implementation rather than diagrams alone.
53.4 MHz
Post-route ML-DSA datapath maximum frequency at a slow, low-voltage, high-temperature corner on PolarFire SoC.
Zero
Vendor cryptographic cores in the custom trust path; security-critical behavior remains visible for review at RTL.
End to end
Requirements through RTL, simulation, synthesis, place-and-route, timing closure, target integration, and pre-fabrication evidence.
Evidence Ladder
Not every artifact answers the same question.
02
Measured proof
Timing, throughput, error behavior, resource use, latency, energy, adversarial results, or other observable evidence.
03
Decision consequence
What the evidence changes: partition, interface, security boundary, deployment path, node, supplier, rights, or qualification.
04
Capital consequence
What can now be released, resized, held, retired, or deliberately funded to generate the next evidence.
Public Layer / Private Layer
Show enough to support the decision. Keep the machinery protected.
Living Cipher publishes architecture concepts, measured outcomes, limitations, and decision implications where useful. Quantitative scoring methods, private thresholds, protected equations, implementation mechanisms, and client-confidential architecture remain proprietary.
Decision Use
The evidence exists to move a decision.
A benchmark without a commitment consequence is interesting. A benchmark that establishes what should remain flexible, what should be built, and what should not yet be funded is architecture evidence.