Useful work.
Accountable Points.

Compute for useful outcomes.

The aim is medicine-first research support, followed by reviewed science and original game assets. Tasks have bounded inputs and explicit validators. Hashes identify evidence and protect integrity; guessing hashes is not the scientific workload.

Reproducing a descriptor calculation, passing a numerical reference or rendering a texture establishes only that specific result. It does not establish a new medicine, clinical safety, physical feasibility or a finished game. Work without a reliable validator remains pending evidence or qualified review.

Live accounting. Explicit limits.

Account sign-in, opt-in device Nodes, fixed research and procedural game calculations, server result verification, account balances, contribution history and the annual reward schedule operate in the existing service. The Windows client can also use installed local AI and Blender for supported jobs after permission is given.

With the automatic computation reward policy enabled, complete allowlisted game calculations, chemistry descriptors and passing numerical AI references qualify for the ordinary reward after full server recalculation. These Points pay for verified computation, not a new medicine, AI execution attestation or finished production art. Local-only model training, historical rechecks, unaccepted renders and unverified material proposals earn zero. A finite original sound queue uses automatic WAV content and amplitude checks. Unique accepted sound files qualify when sound rewards are enabled. Unaccepted 3D assets still need executed production checks. Some research models, production game checks and future services remain unfinished. No public PC or console AAA game release is claimed.

Points purchase, automated redemption, user-to-user transfer, DEX trading, cash-out and crypto conversion are unavailable. Digidash demonstration accounting is separate from live Points.

0.1 → 0.05 → 0.025 Points

Reading the live yearly schedule…

The approved ordinary base is 0.1 Points per qualifying complete block from 11 October 2026 at 00:00 UTC. It halves on each anniversary: 0.05 from 11 October 2027, then 0.025 from 11 October 2028. Reward calculations use integer micropoints, rounded down; rewards below one micropoint become zero under the current precision.

Annual halving changes future issuance, never existing balances. Failed, incomplete, duplicate or unaccepted submissions earn zero. Eligible pool contributions use the disclosed pool share. The separate one-time 50-Point whole Saucer 01 asset reward remains pending independent acceptance; it is not a first-user bonus.

When the fixed work-weight policy is enabled, one ordinary reward block requires 1,048,576 accounting-weight units. Smaller successful results combine across the same account’s Nodes and sectors, and excess weight carries forward. Candidate or data-sample evaluations add one unit; a passing bounded science-reference answer adds 65,536; each checked sound PCM sample at 48 kHz adds one; a complete production-accepted 8K asset adds one full block. These coefficients are accounting rules, not equal scientific value, FLOPS or file size. The live /api/points response discloses activation, weights and saved account progress. Only new successful results qualify; previous balances and reward history are unchanged. The current annual rate applies when a full block completes.

Evidence before rewards.

1. A supported job is assigned to an authenticated, consenting Node.
2. The client performs the allowlisted calculation or approved asset recipe.
3. A signed submission identifies the job, result and evidence. Leases, sequence checks and duplicate protection bind it to the assignment.
4. The server reproduces the deterministic calculation or applies the configured review process. Scientific and production acceptance are separate from numerical correctness.
5. The reward gate checks a qualifying successful outcome, then records any eligible credit once in the persistent account ledger.

A miner's own statement that a job passed is insufficient. Automatic numerical checks require no owner approval. Production asset checks, including material, render, provenance and budget checks, must finish before asset credit. Simulation output is research evidence, not medical approval. No amount of simulation covers every possible real-world condition.

Issued from recorded contributions.

There is no fixed total supply stated by this policy. Issuance depends on accepted qualifying blocks and the active reward schedule. Total issued Points and circulation are read from the service; neither is an estimate of AI tokens, FLOPS or device utilisation.

Account balances and transaction history are persistent server records. Evidence references are hash-linked where supported. This is a centrally operated hosted system, not an independently decentralised blockchain. Backups, replay protection, access controls and audit records support recovery and accountability.

A future blockchain requires implemented consensus, independent operation and reviewed migration rules. No 1:1 on-chain conversion or tradable-token launch is active.

Services must exist before spending.

Planned uses include clearly priced Etheraio digital goods and services. Each available item must have defined delivery, a published Points cost and verified debit accounting before spending is enabled. There is no current entitlement to redeem a Point for cash or a share of research revenue.

The proposed initial purchase price is £1 per Point. The existing proposed post-first-sale quote formula is 99% of recorded net funding divided by total issued earned and purchased Points, with the purchase total rounded up to whole pence. Sales are disabled; this proposed quote policy does not create an exchange rate, guaranteed value or cash redemption right.

Planned funding sources are settled receipts from paid research or licensing, after recorded deductions and fees. A simulation, candidate molecule or medical research activity does not itself create money or liquidity. Any future activation requires deliverable goods, correct accounting and applicable payment-provider approval.

Keep conclusions in their tested context.

Public chemistry reference analysis uses pinned public descriptors. Positive or negative reviewed benchmark outcomes may inform consented search priorities only within the tested subject, model and conditions. Negative evidence can be retained for separately reviewed uses; a failed medicine candidate is not automatically a useful material or game asset.

AI-proposed metallurgy and material properties need provenance, units, reproducible simulation inputs, versioned tools and appropriate independent checks. Known materials must be identified as known. An attractive PBR texture depicts a surface; it does not prove the depicted material can be manufactured or meets its claimed properties.

Current chemistry outcome learning adjusts a scoped selector. It does not establish a general scientific intelligence or automatically fine-tune language-model weights. Weight training needs a reviewed model, licensed and consented examples, evaluation and versioned approval.

Your device. Your decision.

Mining starts only after a user action and follows selected CPU/GPU effort, power and activity settings. The Stop control cancels work. Compatible acceleration depends on hardware, drivers, available jobs and memory; no fixed utilisation or earnings rate is guaranteed. Browsers may expose less hardware capacity and stop work in hidden tabs.

Private research details and owner documents belong in authenticated owner-only storage and encrypted vaults. Public reports should expose only approved summaries. Shared learning and reuse require the relevant consent. Workers necessarily receive the inputs required for their assigned job; access restrictions do not make those inputs invisible to the executing device.

No arbitrary remote shell commands, automatic model downloads or security bypasses are part of the compute workflow.

Publish what has actually passed.

Unreal game production, larger useful workloads, expanded validators, materials research and historical knowledge auditing are development priorities. Original missions, free roam, spacecraft, multiplayer and optional sector downloads are goals; they are not a promise that a production AAA title already exists.

The owner controls approved workloads and production changes. A reward-policy change must be versioned and disclosed. Unsupported evidence should be marked Needs Evidence or Human Review Required instead of being promoted to trusted knowledge. This whitepaper does not replace the Points terms or change existing balances.

Read the operating rules.

Live Points and annual schedule · Public Points status · Shared computation learning · Work and acceptance status · Points terms · Privacy

Version 1.2 describes automatic computation and optional fixed work-weight accounting; its live activation is disclosed by /api/points. Later features require executed verification and updated documentation. Contact: etheraio@proton.me.