How to Use Compute Substrate

Compute Substrate is a shared system for recording proposals and attestations.

It does not execute actions, determine truth, or grant authority. It records structured signals into a verifiable timeline.

Outputs are always optional and external to decision-making.

1. Mental Model

You are not interacting with a system that decides anything. You are writing signals into a shared, persistent history.

A proposal is a claim or idea. An attestation is support for that proposal. The network aggregates these signals deterministically.

Nothing is enforced. Nothing is binding.

2. Propose

A proposal represents a structured claim within a domain.

csd wallet propose \
  --privkey <privkey> \
  --auto-input \
  --datadir /var/lib/csd/node \
  --fee 25000000 \
  --domain eth_price_weekly \
  --payload-hash 0x<32-byte-hash> \
  --uri "ETH > 4000 by Friday" \
  --expires-epoch 100 \
  --rpc-url http://127.0.0.1:8789

This builds, signs, and submits a proposal transaction to the network.

The proposal becomes part of the shared record and can be referenced by others.

3. Attest

An attestation expresses support for an existing proposal.

csd wallet attest \
  --privkey <privkey> \
  --auto-input \
  --datadir /var/lib/csd/node \
  --fee 25000000 \
  --proposal-id 0x<proposal-txid> \
  --score 100 \
  --confidence 90 \
  --rpc-url http://127.0.0.1:8789

Attestations increase the aggregate weight of a proposal in the system. They do not make the proposal true.

4. Read Outputs

The network aggregates proposals and attestations into ranked outputs.

curl http://127.0.0.1:8789/top/eth_price_weekly/0

These outputs are deterministic and reproducible, but not authoritative. They reflect expressed support, not correctness.

5. What This Is Not

Compute Substrate is not a prediction market, oracle, consensus truth system, or execution layer.

Systems built on top of it should not blindly act on its outputs.

6. What It Is For

Compute Substrate is useful wherever aggregation without authority is needed.

This includes cases where you want coordination without enforcement, or a shared memory of claims, beliefs, or signals.

Common examples include prediction aggregation, research ranking, model evaluation, and decentralized signal collection.

7. Invariant

A proposal existing, persisting, or ranking highly does not make it correct.

All outputs remain optional.

All interpretation happens outside the system.