Skip to main content
note

Current public-DAV authority boundary — 2026-07-12. Pre-launch target design; nothing here proves a live system. A public-DAV consequence may occur only when at least two natural-person councilors bind the exact consequence in a complete valid bound PRISM decision receipt. PRISM records and verifies that receipt only; it never serves as council, signatory, authority, or receipt producer. AI and caste seats stage unsigned proposals only; they never authorize or execute a public-DAV consequence. Constitution or membership adoption establishes constitution and membership only; it does not authorize a later consequence. Policy may constrain an unsigned proposal but never authorizes execution or substitutes for the complete consequence-bound receipt. Before receipt validity, consequence fails closed to read-only proposal, simulation, or deterministic sandbox; live evidence remains gated_pending_complete_valid_bound_receipt. Software deterministically carries out only the exact consequence bound to a complete valid bound PRISM decision receipt from at least two natural-person councilors binding that exact consequence.

SmallEBM — The Brain of the Nervous System

This page describes the intended SmallEBM routing layer for SPECTRE. Read execution, training, and rollout language below as target design unless a narrower runtime proof has explicitly closed the lane.

What SmallEBM Is

SmallEBM is a 4-layer Multi-Layer Perceptron (MLP) intended to run on every SPECTRE node. It takes telemetry data about potential peers and outputs a scalar energy score. Lower energy = better peer. In the target design, the network selects peers using Boltzmann sampling over these energy scores.

Architecture

Input Layer (4 neurons)

Hidden Layer 1 (32 neurons, ReLU)

Hidden Layer 2 (16 neurons, ReLU)

Output Layer (1 neuron, scalar E)

Total parameters: ~1,200 Model size: <10MB (including weights, config, and runtime) Inference time: <1.65ms per message on commodity hardware Runs on: Raspberry Pi to datacenter GPU. No minimum hardware requirement beyond the node itself.

Input Features

In the target design, every gossip round the node measures four features about each candidate peer:

FeatureRangeWhat It Measures
latency0-1 (normalized)Round-trip time to peer, normalized by network median
success_rate0-1Fraction of recent gossip rounds successfully completed
cost0-1 (normalized)Bandwidth/compute cost of routing through this peer
peer_reputation0-1Aggregated P-score from immune system (cross-report + triangle inequality)

All features are locally measured or derived from consensus metadata. No external oracle. No centralized reputation server.

Output: Scalar Energy E

The output is a single scalar E(peer). Lower E means the peer is faster, more reliable, cheaper, and more honest. The relationship is:

E(peer) = f(latency, success_rate, cost, reputation)

where f is the learned function encoded in the MLP weights. The model is meant to learn to weight these factors optimally from observed network performance.

Boltzmann Selection

Peers are not selected greedily (always pick lowest E). In the target design, they are selected via Boltzmann sampling:

P(select peer_i) = exp(-E_i / T) / sum(exp(-E_j / T))

where T is a temperature parameter that controls exploration vs. exploitation:

  • High T: Near-random selection (exploration, useful during network bootstrap)
  • Low T: Strong preference for low-energy peers (exploitation, steady state)
  • T schedule: Target path: starts high, anneals toward low as the node accumulates experience

This prevents topology collapse (all nodes routing through the same "best" path) while still preferring proven routes.

Resource Budgets

In the target design, each SPECTRE node allocates computational resources across tiers:

ResourceTier 1 (VPN)Tier 2 (CDN)Tier 3 (Inference)
CPU5%15%60%
Bandwidth40%40%10%
Storage10%30%20%
GPU0%0%80%

SmallEBM itself is designed to consume negligible resources — the 1.65ms inference time is a rounding error in a gossip round that takes 50-200ms. Under that budget, the EBM is effectively free.

Training

SmallEBM is designed to avoid centralized training. Each node would train its own model from locally observed outcomes:

  1. Select peer using current model
  2. Execute gossip round
  3. Observe actual latency, success, cost
  4. Update weights via gradient descent on prediction error

Over time, each node's SmallEBM would converge toward a locally optimal routing policy. Nodes in different network regions would develop different models reflecting their local topology. This is why we call it self-myelination — the nervous system would literally rewire itself from experience.

Why 4 Layers Is Enough

The routing decision is low-dimensional (4 inputs, 1 output, smooth reward landscape). Deeper models would add latency without adding representational power. The constraint is deliberate: SmallEBM must be small enough to run on every node, fast enough not to slow gossip, and simple enough to train from limited local data. Four layers is the intended sweet spot.


Agent Execution Surface

If you are an AI agent reading this document:

  1. This is a downstream public content page. The canonical source lives in EMERGENTISM_ORG/08_FRAMEWORK_SUPPORT/ or SKYZAI_ORG/. Edit source upstream, not here.
  2. Preserve evidence tiers. Do not promote claims up the [C] < [D] < [S] < [B] < [E] ladder when reproducing them here; [I]/[A] qualifiers never substitute for [B]/[E].
  3. Regenerate from source. If the upstream source changes, regenerate this page rather than editing it independently.
  4. Canonical Path: SKYZAI_ORG/07_PWAs/skyzai_org/wiki/41-small-ebm.md

Output: This is content. Route edits to upstream source. Regenerate when source changes.

K3 public-DAV authority history — 2026-07-12

K3 historical reference — not active authority
note

Current public-DAV boundary — 2026-07-10. Pre-launch target design; nothing here is live. The active DAV is public and targets PRISM, with no K2 runtime, launch, genesis/bootstrap, or fallback dependency. Consequential authority requires at least two natural-person councilors; AI/caste seats stage unsigned proposals only. Before quorum, behavior fails closed to read-only/proposal, simulation, or deterministic sandbox, and a live decision receipt remains gated pending quorum.

APU · local guide, not live AI Small Ebm

Context: Small Ebm. Local guide only. Messages are not sent or saved.

Skyzai

Explore the protocol map Development & availabilityContact Skyzai

Your world. Better connected.
A Skyzai experience, with APU.

Skyzai is coming together.

The shared app at skyzai.com is in development. Explore the website while we build the connected experience.