igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity

The Igneum Mission: the one page for the project lead and the closed list of twelve (five in flight, seven new, about 290 agent hours, no new consensus rule beyond the three cut branches), the rejected ideas with the line that killed each, what was not run, and the closing line: nothing is added without removing something

Browse source 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-labs 2026-10-07 08:19:02 +00:00
parent 000a976209
commit 3ae76a3047
1 changed files with 186 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

186
docs/analysis/mission/mission.md Normal file
View file

@ -0,0 +1,186 @@
# The Igneum Mission
Written 7 October 2026, morning UK, by the mission coordinator (branch `mission`, worktree `igneum-wt-mission`) on the project lead's order of the same morning, verbatim: "i dont want to get stuck in a loop here where we keep finding and adding features and security so im going to give you one last mission, deep deep dive into the past and look far into the future, what has been done, what hasnt been done, what has been started but not finished, what can we invent and what can we reinvent to honestly make the perfect gpu network that will be noticed as the revolution that everyone is waiting for."
This is the last research round. Its output is the closed list in section 2. After it the project stops researching and ships. Five lanes ran in parallel and each has its own file with sources and a verdict table: `past.md` (lane 1), `unfinished.md` (lane 2), `future.md` (lane 3), `invent.md` (lane 4), `reinvent.md` (lane 5). Every number below names its lane; the lane file names the source or the model. Hours are agent hours. Nothing on the live devnet was touched and no build or hash measurement was run for this document.
## 1. One page for the project lead
**What has been done.** Of 31 GPU-mined chains since 2011, 8 lost the GPU lane to a chip, 7 closed it by choice, 10 died on price or a rental attack; the 6 still GPU-only pay USD 0.22 to 0.71 a day per RTX 3080 (lane 1). In 20 miner posts the wants were hardware that keeps its value, income without a cliff, a fair supply. Igneum answers the supply in full (no premine, no fund, no fee to any team), the hardware with a measured 2.1x chip bound and the N ladder, the cliff with a monthly glide and a 1 percent tail. Finality by 30 days of blocks, proving as the reward, the leave item and the signing bonus are built or approved for the testnet genesis.
**What has not been done.** Nobody on those forums asked for finality, a DAG or proofs; they asked for a card that keeps its value, a bill that gets paid and a coin that sells. The front page leads with the hash, the earnings page shows £0.00, a fresh install opens with two operating-system warnings and ends its first hour without one number the miner came for, and no proving customer has signed, so the coin has Ergo's 2023 exposure: a thin market. These gaps are hours, not protocol.
**Started and not finished.** Three designs sit on branches: class v5 (the dataset is the chain's own state), finality inside the segment proof, work-stake (a bond of 30 days of work, no coin). Across the industry (lane 2), every useful-work scheme that passed cheap-verify did so by making the work useless, no pool without an operator exists on any chain but Monero, and every proof-of-work finality that held was a second validator set bought with coins. Lane 4 tested whether anything new in consensus beats these; each died on a number: proof of serving on a 108 ms fetch, in-protocol pool shares on 14.6 GB a day per node at 10,000 miners, propagation receipts on the absence of a clock, proof of following because the epoch seed plus class v5 already is one.
**What we invent.** One thing, and it is packaging on what is built: the first offline-verifiable proof-of-work finality object: 786 bytes in a QR a phone verifies in airplane mode, served by every miner's node. No PoW chain has put finality in a user's pocket. Beside it, two genesis bytes the future demands (lane 3): a signature-scheme byte with a key-succession slot, because a post-quantum vote at 8,192 voters costs 57 GB a day unaggregated and the flip must be signalled before a machine exists; and a cache-size rung on the ladder, because a 256 MB consumer cache is a 2 to 3x shortcut.
**What we reinvent.** The miner's first month. The 30-day vote window is a level system no chain has, and it is free: first block, 100 blocks for a vote, a signature in a checkpoint, 30 of 30 days, rank by weight, a signing streak. Every rung is a chain fact from one RPC; a renter cannot top it inside 30 days; no fund pays for it. Lane 5 priced the first month at about 35 hours: the Poisson count-up before the first block (an 8 GB card at 100 GH/s waits 1.6 hours expected), the block card, earnings in IGN first, the weight leaderboard, the hardware census. In a 105-post Reddit sample the miner's own posts rise highest ("my card paid for itself" at 777 times the room median against 66 for the network's milestone), so every shareable surface carries the miner's card, rank and days, never the project's number.
**The honest answer.** Igneum's revolution is the combination already built plus two new things: the finality object in the pocket, and the ladder shown everywhere. No shipped chain has the combination: a random-program GPU hash whose dataset is the chain, miner-only finality with no stake that 51 percent never reaches, that finality inside the execution proof, a work bond with no coin, a tail that holds a 24-hour attack at twelve days of emission in every year. Twelve items, about 290 agent hours, no new consensus rule beyond the three cut branches.
### Decisions owed from the project lead
The cryptanalysis entity and prize; the signing certificates (lane 5); the signed proving customer as a mainnet gate (lane 1); the H100 and A100 hash measurement on rented pods (lane 3).
## 2. The closed list
Twelve items, in build order. "In flight" names the lane or branch that holds it. Hours are the remaining agent hours; a gate is what must be true before the item is called done. Nothing is added to this list without removing something.
| # | Item | State | Hours | Gate |
|---|---|---|---|---|
| 1 | The testnet genesis re-cut as approved | in flight (ship lane) | 0 new | the go checklist |
| 2 | Finality in the pocket: finality in the proof, the 786 B object, every node serves it, the browser verifier | in flight (fin-proof) plus new | 68 | a phone in airplane mode verifies a printed QR and refuses a one-byte change |
| 3 | Class v5, the dataset is the chain, and nothing wider | in flight (class-v5) | 16 | the fast-time harness across a day boundary; Devnet 2 across a real day |
| 4 | Weight-gated deep fork choice before mainnet | in flight (horizon rank 4) | 16 | a 51 percent fresh-key fork from 15 min back refused by every honest node |
| 5 | Work-stake: the job bond made of 30 days of work | in flight (work-stake) | 24 | 1,000 posted jobs, every injected miss forfeits, zero honest forfeits |
| 6 | The ladder shown everywhere: the miner's first month | new (lane 5) | 35 | a Devnet 2 key walks every rung on screen; first share within 10 minutes in 9 of 10 fresh Windows installs |
| 7 | Heat mode, the region and price prompt, the cost-against-rent line | new (lanes 3, 4) | 10 | a set temperature held within 1 degree for 4 h on PC 1 while hash follows the duty cycle |
| 8 | Genesis forward-compatibility: the scheme byte, key succession, the cache rung | new (lane 3) | 10 | the digest test; a vote item with scheme 1 refused by every node until the signal |
| 9 | A second proof system on the active list and the one-third stop switch | new (frontier I4, lane 3) | 24 | 1,000 segments agree across both systems; one injected bad proof disagrees and pays nothing |
| 10 | The launch pack: gates and text, no protocol | new (lanes 1, 3, 5) | 15 | every gate line in testnet-go.md with its check |
| 11 | The pool finished: pool-0 with member keys at 1 percent, TLS, the share sidechain with no operator | new (lanes 2, 4, 5) | 60 | 100 members on the fast-time network paid by the coinbase rule from a share chain with no operator key; the first on a DAG chain |
| 12 | The weight-backed peer directory | new (lane 4, prototype) | 10 | a 34 percent attacker eclipses under 0.1 percent of 1,000 fresh starts |
### 2.1 The testnet genesis re-cut as approved (in flight, ship lane)
What it is: one cut with 18 decimals, `EmissionSchedule::TESTNET_1` (100 IGN a block, a monthly glide with a two-year half-life, a 90-day ramp from 10 percent, a 1 percent tail from year 11.4), and the switches on from genesis: proof verification in consensus, the leave item, the signing bonus at 1,000 bps, finality v3, the latency ladder at rung 0. the project lead approved it on 7 October 2026, 09:3x UK (ledger-decisions.md).
Why it is right: lane 1's verdict rows 4 and 5 (the cliff and the premine) are answered by it in full; lane 3's tail row holds to year 200.
Evidence: tail-emission.md one page; vote-or-burn.md section 5; ledger-decisions.md 7 October.
Cost: 0 new hours; the ship lane's own plan.
Gate: the go checklist (`docs/plans/testnet-go.md`); nothing mines until the word.
Order: first; every later item lands on this genesis.
### 2.2 Finality in the pocket (in flight on `fin-proof`, plus three new parts)
What it is: the weight table carried inside the recursive segment proof (W2 updated one mergeset per segment, the BLS certificate verified in the guest, 494 B of public values), then three parts this mission adds: a Groth16-wrapped proof as a 786 B object (260 B proof, 494 B public values, 32 B key id) in a QR or URL that a phone verifies offline; every miner's node serving the latest object by default (an Ember line, unpaid); and the WebAssembly verifier in the tab (horizon rank 19).
Why it is new: no proof-of-work chain has an offline-verifiable finality object; the closest shipped forms are Mina's recursive state proof (2021) and Helios's committee light client (2022), both online and neither over proof of work. Lane 4 section 6 and 3, persona verdicts do now, do now, prototype, do now.
Evidence: finality-in-proof.md sections 1 to 5 (level 1 of the blue-set check designed, level 0 in the prototype); the certificate half verifies in 58 to 68 ms warm in a browser (bench-log round 6); SP1 light verifier 0.032 s on a Mac core; a version-40 QR holds 2,953 B.
Cost: 40 for the fin-proof remainder (level 1 of 5.2, the cycle counts of section 6.1 on the box, the 5090 prover time on a pod), 8 for the QR object, 4 for the Ember line, 16 for the WASM verifier. 68.
Gate: a phone in airplane mode verifies a printed QR and refuses a one-byte change; "locked, voter set verified" on cellular from a miner's node; a block proof verifies in the tab under 500 ms on a laptop.
Order: second; it is the one invention and it needs the genesis of item 1 for the pinned aggregator id.
### 2.3 Class v5, the dataset is the chain, and nothing wider (in flight on `class-v5`)
What it is: each day's dataset built from the execution state at a reference block an hour before the day; a card without the state builds every item wrong. The branch widened it to a per-window refresh (section 2a of its page). This mission's ruling: no further widening. Lane 4 tested proof of following, serving, propagation and state availability as additions and found each dead on a number: a key with no state answers an availability challenge in about 108 ms over a 100 ms link, so any deadline over one block proves only that someone served; propagation receipts need a clock the DAG does not have; the epoch seed plus the state root already commit to the chain's recent blocks, so proof of following is class v5 by another name.
Why it is right: hash rate and watts unchanged within noise (63.083 against 63.088 MH/s, 205 to 208 W on a 4090), build +1.4 ms, verifier +0.11 to 0.21 ms per unit; it removes the f = 0 recompute chip as a category and the stateless pool miner.
Evidence: class-v5-stored-state.md; new-pow.md 5.2 and 6; invent.md sections 1 and 8 rows 11 to 15.
Cost: 16 (the 4090 rows owed in section 7, the exec snapshot wire version 2, the GPU hosts' leaf buffer, the spec text).
Gate: the fast-time harness crosses a day boundary with the roots agreeing on every node and the stateless node's miner at 0 accepted blocks; Devnet 2 across a real day boundary; then the 95 percent signal with a floor.
Order: third; it rides the class-signal machinery item 1 ships.
### 2.4 Weight-gated deep fork choice before mainnet (in flight, horizon rank 4)
What it is: a tip whose fork point is older than D (10 min of past-median time) is a fork-choice candidate only if the keys that built it hold at least a third of the weight table at the fork point.
Why it is right: lane 1's verdict row 6: rented hash reorganised Bitcoin Gold, Vertcoin and Musicoin for USD 70 k to 18 M; Igneum's finality bounds this after day 20 and not before, and the first 20 days are when a new chain is watched. The 51 percent paper prices a 12-hour double spend during a pause at USD 146 at 1 GH/s today.
Evidence: 51-percent.md section 5 rank 2; horizon rank 4; past.md section 4 row 6.
Cost: 10 to 16.
Gate: fast-time: a 51 percent fresh-key fork from 15 min back is refused by every honest node; a one-third-weight fork is accepted; partition heal unchanged.
Order: fourth; before the public testnet opens, because the testnet is the first chain outsiders can rent against.
### 2.5 Work-stake (in flight on `work-stake`)
What it is: a vote key may claim an external proving job only if its next window of pool income covers the job's value; a miss, refusal or forgery forfeits that income for one window, the buyer's fee refunded first, the rest to the pool. No balance moves; no coin is posted. The spec sentence lands first: "no coin stake; the only thing at stake is 30 days of public work."
Why it is right: a bond nobody can buy (2,793 IGN at risk for an 8 GB card at 100 GH/s against a 0.0015 IGN coin bond); every attack in its section 6 fails on the window arithmetic.
Evidence: work-stake.md sections 3 to 7; frontier 3.2 and 4.1.
Cost: 24 (orders carried in blocks, the term unwound on reorg, the finality report wired to `judge_at`, the snapshot field, the deadline floor).
Gate: Devnet 2: 1,000 posted jobs with 10 percent injected misses, every injected fault forfeits, zero honest keys forfeit; a 60 s partition across 100 deadlines forfeits nothing.
Order: fifth; it needs the job market of spec 5.4, which the first customer (item 10) needs.
### 2.6 The ladder shown everywhere: the miner's first month (new, lane 5)
What it is: the 30-day vote window as the status ladder on every surface, and the first month's moments built around it. The rungs: first block; 100 blocks and "your key has a vote"; "your signature is in checkpoint K"; 30 of 30 days, full weight; rank by weight; the signing streak; "your card proved shard N of block M". The parts: the Poisson count-up before the first block ("a card like yours finds a block about every 1.6 hours on today's network; chance so far 31 percent"); the block card with a locally drawn PNG and the explorer link; Earnings in IGN first (6,912 IGN a day for a 100 MH/s card at 100 GH/s at the full rate) with a typed price never fetched; the weight leaderboard on `/live` (weight, never hashrate, so a renter sits at the bottom for 30 days); the public address profile behind an opt-in; the hardware census on `/miners` from the hourly program's per-card timings (frontier 4.3); #first-blocks and the Voter and Window roles granted from `getFinalityWeights`.
Why it is the right reinvention: lane 5's archive sample shows the joy posts are the first block, the first payout and the rig photo In a 105-post sample the miner's own posts rise highest: "my card paid for itself" at 777 times the room median against 66 for the network's milestone.; Helium, Chia, Ethereum 2017 and Kaspa 2022 each had a hook, a daily ritual and a status surface, and each status surface was a company number or a hashrate a renter could top. Igneum's is a consensus fact. Nobody has it.
Evidence: reinvent.md sections 1, 2, 3.2 to 3.7, 7; the solo-block expectations per tier at 100 GH/s and 1 TH/s (an 8 GB card: 1.6 h and 16 h; it never reaches the vote line at 1 TH/s, which is why the pool is item 11; the 100-block dust line means an 8 GB card never votes past about 500 GH/s, a question for the finality lane and a watch item, not a list item).
Cost: 35 (Poisson line 2, block card 4.5, earnings 6.5, ladder 6, leaderboard 3, profile 3, census 4, Discord 5, the first-hour timeline on `/miner` 2).
Gate: a Devnet 2 key walks every rung on screen in a view test; every number on the tab names its RPC field on hover; "first share within 10 minutes of download in 9 of 10 fresh Windows installs" published on `/evidence`.
Order: sixth; it is the first thing the first outside miner sees.
### 2.7 Heat mode, the region and price prompt, the cost-against-rent line (new, lanes 3 and 4)
What it is: Ember holds a room temperature or a schedule and the hash follows the duty cycle; a per-region electricity-price prompt with a banned-region line ("mining may be restricted where you are; you are responsible for checking"); the Earnings tab's cost line against the measured rental rate so a miner sees when rented hash undercuts their card.
Why it is right: a 300 W card as a heater breaks even against a UK gas boiler or a COP-3 heat pump at 5.3 p an hour earned and is free against a resistive panel (lane 4 section 5); the heating season credits about a third of a UK or EU miner's bill (lane 3 section 4); lane 3's 2031 row says retail-power tiers leave first when idle datacentre hash sets the rent, and the heat mode keeps winter home miners on. Prior art is Heatbit (2022), 21energy and Qarnot; new only as a chain client's feature, which is the point.
Evidence: invent.md section 5 and rank 2; future.md sections 3, 4 and 10.
Cost: 6 heat mode, 2 region prompt, 2 the rent line. 10.
Gate: a set temperature held within 1 degree for 4 h on PC 1 while hash follows the duty cycle; the prompt shows the right banned list for a Russian region; the rent line reads the bench-log rate.
Order: seventh; app only, no digest, ships in any cut.
### 2.8 Genesis forward-compatibility: the scheme byte, key succession, the cache rung (new, lane 3)
What it is: three genesis fields. A `sig_scheme` byte in the vote item and the key reveal (0 = BLS12-381 today), so a post-quantum scheme flips by the 95 percent signal and not by a fork. The key-succession item (W5, unimplemented on the node) so a key can hand its weight and its forfeit term to a successor once, which is the migration path. A cache-size rung on the signal ladder beside N, because a consumer last-level cache at the cache size (256 MiB) gives that card's owners a 2 to 3x shortcut (chip-model-v3; consumer LLC is 96 to 128 MB today, datacentre 256 MB). The class-group VDF's quantum fallback is flagged, not sized.
Why it is right: naive ML-DSA-44 votes at 8,192 voters cost 19.4 MB a checkpoint and 57 GB a day; with 250x SNARK aggregation about 223 MB a day; a cryptographically relevant quantum computer is "quite possible (28 to 49 percent)" within ten years (Global Risk Institute 2025) and NIST deprecates ECDSA and EdDSA after 2030. The flip is signalled the year of deprecation, not the year a machine appears; the byte costs nothing now and a fork later.
Evidence: future.md section 7 and 10; finality-in-proof.md 7 (W5 absent).
Cost: 10 (the byte and the digest test 4, the W5 item 6, the rung 1 inside the ladder code, the spec text).
Gate: the digest test; a vote item with scheme 1 refused by every node until the signal; a succession carried once and refused twice in the fast-time harness.
Order: eighth; genesis fields must land before item 1's final cut, so this is scheduled with it and listed here for its own gate.
### 2.9 A second proof system on the active list and the one-third stop switch (new, frontier I4 and lane 3)
What it is: a second zkVM behind the `ProofSystem` trait (RISC Zero or OpenVM) proving the same segments on one fleet box as a shadow; a disagreement between the two on any segment is a public alert and the pool stops paying on proofs until 1/3 of weight signals which system to trust. Full nodes keep the native-execution veto whatever happens.
Why it is right: three soundness-class events across SP1 and RISC Zero in 18 months (lane 3 section 5); the veto protects full nodes and nothing protects a light client or the proving pool; lane 3's death row 4 is this.
Evidence: future.md sections 5 and 9; frontier I4; ledger P7 and D6.
Cost: 24.
Gate: 1,000 segments agree across both systems; one injected bad proof disagrees, pays nothing and raises the alert.
Order: ninth; before any phone trusts item 2's object.
### 2.10 The launch pack: gates and text, no protocol (new, lanes 1, 3 and 5)
What it is: lines added to `docs/plans/testnet-go.md` and the litepaper, each with its check. Gates: a signed proving customer before mainnet, with the weight of the Devnet 2 gate; per-tier income published at three prices (USD 0.005, 0.02, 0.10) before the testnet, per the consequences rule; the launch-week hash-origin report (key counts, pool shares, fleet shares) from the observer, daily for the first 90 days; the first 100 keys, the first 1,000 independent keys (the X5 definition), the first pool not run by the project, the first outside-reproduced benchmark, the first block from a card the project does not own; signed and notarised installers (Developer ID, Authenticode) with the measured first-share time per release. Text: the three wants on the front page and finality on page two; "a block pays its miner whether or not anyone buys a proof that day"; the eight regulatory sentences of lane 3 section 8.3 (no issuer, no sale; the sustainability indicators per era; no price language; the pool holds no balance; no privileged key), labelled "not legal advice; counsel is engaged".
Why it is right: lane 1's shape (income halves 60 to 120 days after a peak, under USD 0.10 per kWh within 6 to 18 months, 50 to 100 percent of hash gone in 90 days) and its rows 2, 4, 7 and 9; lane 5's install findings (SmartScreen warns on any file without reputation); lane 3's regulation rows (MiCA 4(3)(b) exempts block-reward assets; the UK regime from 25 October 2027 covers platforms and custody, not mining).
Evidence: past.md sections 3 and 4; future.md section 8; reinvent.md 3.1 and 4.1.
Cost: 15, plus the project lead's certificates and the customer's signature.
Gate: every gate line in testnet-go.md with its check; the forbidden-strings check passes on the new text; the hash-origin report posts on Devnet 2 for seven days before the testnet go.
Order: tenth; text and gates can land any hour, and the customer gate is the one that takes longest.
### 2.11 The pool finished (new, lanes 2, 4 and 5)
What it is: pool-0 run by the project at the testnet go with the member's vote key in every header (the pool holds no weight and no vote), a 1 percent fee published beside the dev fee, PPLNS with a round every 60 s; TLS before the public testnet (spec 9.3) and the pool page rows Q69 to Q73; then the pool without an operator: a P2Pool-style share sidechain in the pool crate, shares forming a 10-second chain committed in the coinbase extra data, the window's payout computed by every node from consensus data as the 20 percent already is, no balance and no operator key, every member running the node the design already assumes. Lane 2's verdict: the finished form is Monero P2Pool's, not SmartPool's (SmartPool's on-chain claims died on gas, and at 1.35 ms per share verification a sampled contract would still need the zkEVM to re-derive a warp unit per sampled share); lane 4's in-protocol pool (shares inside blocks, every node verifying every share) stays refused at 14.6 GB a day per node.
Why it is right: no pool without an operator exists on any chain but Monero (P2Pool at 7.7 percent of hash, 3,520 miners, 0 percent fee; Ocean 2.05 percent of blocks with 85 percent through DATUM; Stratum V2 job declaration at 1 block in 52,297); Igneum already holds the three objects that make one cheap (the vote key in the header, the 20 percent paid by sortition, the coinbase extra data), and nobody has built one on a DAG chain. An 8 GB card at 1 TH/s finds a solo block every 16 hours and 23 percent of its days have none, so the first-month ladder of item 6 depends on a pool from about 1 TH/s.
Evidence: unfinished.md section 5 (5.2, 5.3) and headline 2; invent.md section 4 and rank 13; reinvent.md 3.5 and 1.4; pool.md; spec 09.
Cost: 60 (TLS 8, the pool page rows about 12, the share sidechain about 40).
Gate: 100 members on the fast-time network paid by the coinbase rule from a share chain with no operator key, a withheld share earning nothing, a member's dropped share provable from its own log; the first payout inside two minutes of the first share for an 8 GB card on Devnet 2.
Order: eleventh; pool-0 and TLS before the testnet go, the share sidechain before the network passes about 1 TH/s.
### 2.12 The weight-backed peer directory (new, lane 4, prototype)
What it is: a coinbase section where a key above dust may publish its node's address; a fresh node dials from that list weighted by 30-day weight, beside the DNS seeds.
Why it is right: lane 4's model E bounds an eclipse by a 34 percent attacker at 1.8e-4 with 8 outbound peers and 3.2e-8 with 16; prior art (DNS seeds 2011, discv5 2019, Heilman's eclipse paper 2015) weights nothing by work; lane 3's 2031 death row (weight concentrating in hosting firms) is first seen through such a list.
Evidence: invent.md section 7.1 and rank 4; horizon rank 9 (the peer floor it extends).
Cost: 10.
Gate: a fresh node with genesis peers only reaches 8 outbound from the directory; a 34 percent attacker eclipses under 0.1 percent of 1,000 fresh starts; the NAT fraction measured on the fleet.
Order: last; a prototype, dropped if the NAT fraction makes the list useless.
## 3. Rejected, with the one line that killed each
| Idea | The line |
|---|---|
| A pool inside the protocol (shares in blocks, no operator) | FruitChains (2016): 14.6 GB a day and 5 cores per node at 10,000 miners on a 10 s share; 146 GB and 50 cores at 100,000 |
| A reward slice for serving state or data | a key with no state answers in 108 ms over a 100 ms link; it pays for what the lottery already compels |
| Proof of propagation receipts | no consensus clock; 94 to 376 MB a day of receipts; GHOSTDAG already is a proof of propagation |
| Proof of following beyond class v5 | the epoch seed and the state root already commit to the recent chain |
| A first-block bonus per key | a 10 percent miner collects it 2,592 times a window: an instamine with extra steps |
| Reorg insurance for exchanges, pause insurance from the pool | no slashable payer by ruling; the lock is the product; the pool pays the wrong people |
| Proof of latency, proof of unique card, an energy-price oracle, hash-rate futures in protocol | no clock; nothing counts cards by design; a lever the bid neutralises; needs a coin bond |
| Finality as a service for other chains | kept as a demand-gated prototype only; no customer named, so not on the list |
| The weight-aged reward rule (pay per block falls when hash arrives faster than the weight) | a 30-day income ramp on every honest newcomer, which the first-month ladder cannot afford; revisit only if weight by hosting provider passes a third |
| Mining is proving; useful work in the hash | 2.9 MB of openings per block, a 32 to 40 ms verify against the 10 ms gate; Aleo's fastest-prover-wins |
| Proving others' chains as the main income | all of Ethereum L1's proving is about USD 36 a day against USD 13,700 of year-1 emission |
| Vote-or-burn | 70 to 80 percent respect against the 95 percent test; the signing bonus is on at genesis instead |
| Dormant-coin rent, any holder penalty, a device bounty, a coin stake, Bitcoin anchoring, privacy | refused by ruling |
| A hardware wallet verifying proofs on the secure element | a bn254 pairing on a Cortex-M class element is seconds to minutes; the companion verifies |
| A compute market for unverifiable work | an escrow without a verifier is a trust-me payment with lower fees |
| Merged mining, multi-algo, an auxiliary chain | A child inherits its parent's pool concentration and loses its say in rules; a second hash lane splits the weight table and the DAA (Myriad, Verge, DigiByte); the one merged-mining shape with value is the share sidechain of item 11 |
| Gamification paid from a fund, referral in coins, an airdrop for anything, NFT badges, off-chain points, a rank-by-hashrate board, a top-earners board in fiat, streaks that demote | there is no fund; nothing is for sale; the rung is a chain fact; a renter tops hashrate on day one; the burn was refused |
| Any number that needs a company server to be trusted | every number shown is reproducible from a node, or it is not shown |
## 4. What this round did not run
| Item | Why | Where it sits now |
|---|---|---|
| The H100 and A100 hash rate on rented pods | no pod rented for this round | item 10's measurement line; lane 3 section 3 |
| The in-guest BLS and colouring cycle counts (fin-proof 6.1) | no box slot taken for this document | item 2's cost |
| The 4090 rows for class v5 (section 7 of its page) | the pod window was the class-v5 lane's | item 3's cost |
| Reddit, bitcointalk and the Wayback Machine | refused this environment's fetches; quotes came through search snippets and project forums | past.md section 2, reinvent.md section 1, each labelled |
## 5. The closing line
Nothing is added to this list without removing something. Signed, the mission programme, 7 October 2026.