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

Mission lane 1, the past: every GPU-mined chain since 2011 (31 rows), what each promised and paid, the earnings-curve shape (income halves 60 to 120 days after peak, under USD 0.10 per kWh in 6 to 18 months, 50 to 100 percent of hash gone in 90 days), the three wants and three hates from the archives, the lessons mapped to the design

Browse source 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-labs 2026-10-07 08:08:01 +00:00
parent c7149bdd6a
commit f3df738a35
1 changed files with 304 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

304
docs/analysis/mission/past.md Normal file
View file

@ -0,0 +1,304 @@
# The past, honestly: every GPU-mined chain since 2011 and what GPU miners want
Lane 1 of the last mission. Written 7 October 2026 by the past lane for the mission coordinator. Scope: the history of every chain that mattered to GPU miners, what each promised, what each paid, when the GPU miners left, and what the record says Igneum should change. The ASIC side of this history (which hash fell to which chip and how fast) is already in `docs/analysis/asic-resistance-history.md` and is not repeated; this file cites it where a chip is the reason a chain died for GPUs. The emission and governance comparison against Kaspa, Monero, Ethereum and the rollups is in the Horizon economy lane (section 4.6) and is not repeated either.
Every figure from a source carries a URL and the access date in section 6. Every figure from memory is labelled approximate. Reddit, bitcointalk and the Wayback Machine refuse this environment's fetches (sections 2 and 6 say which quotes came through search snippets instead). Times are UK time.
## 0. Progress
| Time (UK, 7 Oct 2026) | State |
|---|---|
| 08:05 | Read CLAUDE.md, asic-resistance-history.md, horizon-2026-10.md section 1, economy-and-utility.md 4.6 |
| 08:10 to 08:45 | 75 web searches and 60 fetches across project blogs, pool blogs, forums, press (search budget exhausted at 08:40) |
| 08:49 | Writing; sections 1 to 6 landed in one pass |
| 08:53 | Length check (258 lines, under the 300 floor); added 1a (the Merge week) and 1b (promised against delivered), split the grouped chain paragraphs, reconciled headline 1's counts against the table |
| 08:58 | Done. File complete, reported to the coordinator |
## 1. One row per chain
Income columns are USD per card per day before electricity unless stated. "Exit" is when the median GPU miner stopped earning more than electricity at USD 0.10 per kWh, or when the lane closed outright. State is October 2026.
| Chain (hash) | Launch | What it promised GPU miners | Peak GPU income (card, date) | Exit income (card, date) | GPU miners left | What killed the GPU lane, or state today |
|---|---|---|---|---|---|---|
| Litecoin (scrypt) | 13 Oct 2011 (approximate) | Memory-hard hash mineable on CPUs and consumer GPUs; no ASIC existed (litecoin.watch) | 200 to 800 kH/s per Radeon, 2012 to 2013; USD per card approximate 1 to 5 | 2014, under power for any GPU once Zeus 28 MH/s at 800 W shipped | Mid 2014 | Scrypt ASIC (Gridseed Jan 2014, Zeus mid 2014, Antminer L3 2017). Today an ASIC chain merged with Dogecoin |
| Dogecoin (scrypt, AuxPoW) | 6 Dec 2013 | A fun scrypt coin for the same GPUs as Litecoin | Approximate USD 1 to 3 per card, Jan to Feb 2014 | Aug 2014 | Aug to Sep 2014 | Merged mining under Litecoin from block 371,337 (28 Aug 2014); hashrate +1,500 percent in a month (Binance Research). GPUs never came back |
| Ethereum (Ethash) | 30 Jul 2015 | GPU-friendly memory-hard hash with a promise to leave for proof of stake | USD 0.233 per MH per day, Jan 2018 (so USD 23 per 100 MH); USD 6.35 to 9.15 per RTX 3080, early 2021 (Minerstat via Wccftech); May 2021 approximate USD 12 to 15 per 3080 | USD 1 to 2 per 3080 on 14 Sep 2022 (approximate) | 15 Sep 2022, all at once | The Merge. About 900 TH/s and about 20 million GPUs (Tom's Hardware estimate) lost 94 percent of GPU income in one block (2Miners: USD 24 M a day to USD 1.2 M) |
| Ethereum Classic (Etchash) | Jul 2016 fork | A home for Ethash GPUs after the Merge | 70.5 to 158 TH/s in 7 hours on 15 Sep 2022 (CoinDesk), 296 to 312 TH/s inside a day | Under electricity for most GPUs within weeks; hashrate -48 percent in four days (AMBCrypto) | Sep to Dec 2022, then again as E9 Pro and Jasminer ASICs took the chain | Today 130 to 186 TH/s, ASIC-dominated; best ASIC nets USD 3.64 a day at USD 0.10 (asicminervalue) |
| Monero (CryptoNight to RandomX) | 18 Apr 2014 | CPU and GPU mining, fork away any ASIC | Vega 56 era 2017 to 2018, approximate USD 2 to 4 per card | 30 Nov 2019: GPU performance "gets a tiny boost or degrades" (MinerUpdate) | 30 Nov 2019: 75 percent of GPU miners offline (MinerUpdate) | RandomX moved it to CPU on purpose; hashrate 300 MH/s to 800 MH/s in weeks. Qubic's rented-CPU campaign claimed 51 percent on 11 Aug 2025 with a 6-block reorg (The Block); the AFT 2026 paper finds no sustained majority and no reward edge |
| Zcash (Equihash 200,9) | 28 Oct 2016 | "Universal accessibility" in mining (forum users' reading of the project) | USD 8 per GTX 1080 Ti, Jul 2017 (WhatToMine via 2Miners) | USD 2.63 per 1080 Ti, 6 May 2018 (2Miners) | Jun to Sep 2018 | Antminer Z9 mini announced 3 May 2018; the company stayed neutral; no fork. ASIC chain since; proof-of-stake migration under discussion (approximate) |
| Ravencoin (X16R, KAWPOW) | 3 Jan 2018 | GPU-only by design; three hash changes to keep it so | Approximate USD 6 to 8 per 3080, Feb to Apr 2021 (RVN peak USD 0.2857) | 3080 loses USD 0.13 to 0.31 a day, 2026 (WhatToMine) | Gradual through 2023 to 2025 | KAWPOW (6 May 2020) cut hashrate 10x as FPGAs left and held off chips for six years; the Merge doubled it (8.9 to 15.9 TH/s). On 7 Aug 2026 a header-validation flaw let blocks carry "no genuine ProgPoW work" and split the chain; RVN -19 percent (crypto.news) |
| Ergo (Autolykos v2) | 1 Jul 2019 (approximate) | Memory-hard, ASIC-resistant, GPU-only (minerstat FAQ) | Approximate USD 3 to 4 per 3080, Sep to Nov 2021 (ERG near USD 15) | USD 0.22 per 3080 before power, Oct 2026 (hashrate.no) | Late 2022 onward | The Merge pushed it from 22 to 155 TH/s in a day (Decrypt); difficulty ate the income within weeks. Still GPU, no chip, income under power |
| Kaspa (kHeavyHash) | 7 Nov 2021 | "GPU PoW", fair launch, no premine (bitcointalk ANN title) | Approximate USD 2 to 3 per 3080, Feb to Mar 2023 | 3070 at USD 1.21 a day, 14 Apr 2023 (2Miners) while a KS2 made USD 361 | Apr to Oct 2023 | IceRiver KS0/KS1/KS2 (Apr to Jul 2023), Bitmain KS3 (May 2023). KS0 earned USD 25 a day in Jul 2023. Today 347 PH/s (hashrate.no), KS7 at USD 1,899, GPUs effectively zero |
| Grin (Cuckaroo29, Cuckatoo31+) | 15 Jan 2019 | 90 percent of blocks for the GPU hash at launch, falling to zero over two years (Grin forum) | Approximate USD 1 to 2 per card, Jan 2019 | 2020 | 2019 to 2020, by schedule | The schedule itself. Today Cuckatoo32 ASIC only, GRIN at USD 0.035, USD 16 k daily volume (minerstat, approximate) |
| Beam (BeamHash III) | 3 Jan 2019 | GPU-friendly Equihash 150,5 | Approximate USD 1 to 2 per card, Jan 2019 | 2020 onward | Most by 2021 | Never an ASIC; the price went. Still GPU-mined, sixth hard fork July 2026, negligible income |
| Conflux (Octopus) | Oct 2020 (approximate) | GPU-only, no FPGA or ASIC | Approximate USD 2 to 3 per 3080, 2021 | USD 0.71 per 3080 before power, Oct 2026 (hashrate.no) | Ongoing | Still GPU at 3.9 TH/s. No chip in six years; income under power for most cards |
| Alephium (Blake3) | Nov 2021 (approximate) | GPU-mineable sharded chain | Approximate USD 1 to 2 per 3080, 2023 | 2024 | May to Oct 2024 | Goldshell AL-Box 360 GH/s at 180 W (May 2024), AL-Box III 1.25 TH/s (Oct 2024). GPUs out in five months; even the ASICs now lose USD 3.44 a day at USD 0.10 (asicminervalue) |
| Nexa (NexaPow) | 21 Jun 2022 | CPU first, then GPU, "scale to global capacity" | Top of the GPU tables Jul 2023 (2Miners) | 2024 (approximate) | 2024 (approximate) | Price. Approximate: down over 99 percent from its 2023 high; hashrate a fraction of 2023 |
| Aleo (PoSW) | 18 Sep 2024 | Proving as mining on GPUs | USD 13.50 per RTX 4090 at ALEO USD 9, Sep 2024 (PANews via AiCoin) | Under USD 5 per 4090 at USD 3.40, late 2024 (same source) | Q4 2024 to 2025 | Price and allocation: a one-year lock on early rewards, validators seeded with 270 k to 1.1 M tokens, 10 M stake to validate; f2pool holds 28.9 percent of proving (hashrate.no). "Difficulty plummeted ... as miners exited" |
| Iron Fish (Blake3, FishHash) | 20 Apr 2023 | GPU mining of a private chain | Approximate USD 1 to 2 per 3080, Apr 2023 | USD 0.48 per 3080 before power, Oct 2026 (hashrate.no) | 2023 to 2024 | Blocks of "unknown origin and hash power" within weeks (FPGA suspected); FishHash fork 2 Apr 2024 kept it GPU; the price did the rest |
| Flux (ZelHash) | 2018 as Zel | GPU-only Equihash 125,4 | USD 50 k a day across the chain, Sep 2022 (2Miners) | Late 2025 | Late 2025 | Proof of Useful Work v2 (RunOnFlux, 23 Oct 2025): FluxNodes produce all blocks; "GPUs stopped having any relevance to FLUX rewards" |
| Firo (MTP, FiroPoW) | 2016 as Zcoin | GPU-only ProgPoW variant since 26 Oct 2021 | 4x hashrate after the Merge | USD 0.56 per 3080 before power, Oct 2026 (hashrate.no) | Ongoing | No chip in five years. Still GPU, 13 to 54 GH/s, small |
| Vertcoin (Lyra2RE, Verthash) | 8 Jan 2014 | "The people's coin", fork away every ASIC | Approximate USD 1 to 2 per card, Dec 2017 | 2019 | 2018 to 2019 | Two rental 51 percent attacks (Dec 2018: 300-block reorg, about USD 100 k; 1 Dec 2019: 603 blocks replaced, NiceHash). Verthash (Jan 2021) kept GPUs; nobody came back |
| Bitcoin Gold (Equihash-BTG) | Oct 2017 | "Make Bitcoin decentralized again" on GPUs | USD 474 a coin, Dec 2017 | 2018 | 2018 to 2020 | 51 percent attacks: May 2018 (388,000 BTG, up to USD 18 M), Jan 2020 (29 blocks, USD 70 k). Delisted by Binance 1 Sep 2024 and Upbit 23 Jan 2025; "under 100 actively reachable nodes" (Decrypt) |
| EthereumPoW (Ethash) | 16 Sep 2022 | Ethereum mining continues | Above USD 140 a coin at launch | 2022 to 2023; profitability -85 percent from peak | Within months | Price: USD 1.70 by mid 2025 (99 percent down); pools left for ETC |
| Nervos CKB (Eaglesong) | 16 Nov 2019 | GPU-mineable launch | 300 GTX 1070 Ti earned USD 410 a day, 19 Nov 2019 (2Miners) | The same rig earned USD 4.46 a day, 21 Mar 2020 | Mar 2020 | Hashrate 198 TH/s to 1.99 PH/s in 16 days before any chip was sold (secret ASICs); Antminer K5 Apr 2020 at USD 22 then 11 a day |
| Radiant (SHA512/256d) | 2022 (approximate) | "GPU/ASIC mining" in the ANN title | Approximate USD 1 per 3080, 2023 | Sep 2024 | Sep 2024 | IceRiver RX0 and DragonBall A11 ASICs, Sep 2024. Designed to go to ASIC |
| Karlsen, Pyrin, Nomura (kHeavyHash forks) | Nov to Dec 2023 | "We will ensure long-term GPU-friendly mining" (Karlsen README) | Approximate USD 0.5 to 1 per 3080, Dec 2023 | 2024 | 2024 | Karlsen moved to FishHash (KarlsenHashv2, 12 Sep 2024) as kHeavyHash chips arrived; Pyrin was called "another pre-mine scam" on bitcointalk (five days of private mining before the announced start) |
| Dynex (DynexSolve) | Sep 2022 | Proof of useful work, "the most GPU mined coin" (22 percent of Hive OS, Oct 2023) | Approximate USD 2 to 4 per 3080, mid 2023 | 2024 | 2024 | Price: 99.9 percent below its USD 1.39 high (CoinGecko). The useful-work marketplace never paid the miners |
| Qubic (uPoW) | 2022 (approximate) | AI training as mining, CPU first | By 2025 about 90 percent of its compute was GPU rigs (Qubic blog) | n/a | Rotating | Pivoted to CPU incentives, then rented its idle fleet to Monero (May to Aug 2025, 27,000 blocks, USD 3.5 M), then to Dogecoin ASIC mining (1 Apr 2026). A hashrate broker, not a GPU chain |
| Clore (Kawpow) | 2022 | GPU coin paid for GPU rental | ATH USD 0.45, 17 Mar 2024 | Dec 2025 | Dec 2025 | Proof of work "ended" at block 1,584,180 (Clore changelog, Dec 2025): "miner sell pressure fully removed". A 4090 rents for USD 1.18 to 2.45 a day there |
| TurtleCoin (CryptoNight Turtle) | Dec 2017 | A friendly CPU/GPU coin | Negligible | Mar 2023 | 15 Mar 2023 | Halted at block 5,500,000; token moved to Fantom (TurtleCoin v2 FAQ). Dead as a chain |
| Musicoin (Ethash) | 2017 | Ethash side income | Negligible | 2019 | Jul 2019 | Repeated 51 percent attacks, Bittrex delisting, 2Miners delisting (31 Jul 2019: "Musicoin is dead") |
| Akroma (Ethash) | 2018 | Ethash side income with masternodes | Negligible | 2019 | Jul 2019 | Developers left; USD 92 daily volume; delisted by 2Miners |
| Callisto (Ethash) | 2018 | Ethereum Classic airdrop chain | Negligible | 2024 | 2024 | Team liquidity crisis and "civil war" in 2024 (Callisto history page). Dead |
### 1a. The Merge week, the one migration with hour-level data
| When (UK) | What happened | Source |
|---|---|---|
| 15 Sep 2022, 07:43 | The Merge; Ethereum's last proof-of-work block; about 900 TH/s of GPU hash loses its income | Tom's Hardware |
| 15 Sep 2022, 07:00 to 14:00 | ETC from 70.5 to 158.3 TH/s; RVN from 8.9 to 15.9 TH/s; ERG from 22 to about 70 TH/s | CoinDesk, Decrypt |
| 15 to 16 Sep 2022 | ETC peaks at 296 to 312 TH/s (about a third of Ethereum's hash); ERG peaks at 155 TH/s (7x) | AMBCrypto, crypto.news, Decrypt |
| 16 Sep 2022 | ETHW mainnet; the fork trades above USD 140 and falls for three years | KuCoin price history |
| 19 Sep 2022 | ETC hashrate down 48 percent from its peak in four days | AMBCrypto |
| Sep 2022 | Hiveon pool: 16 percent of its miners to ETC, 7 percent to RVN, the rest spread or gone | Hiveon |
| 12 Sep 2022 (before) | 2Miners' arithmetic: USD 24 M a day of GPU income, 94 percent of it Ethereum; USD 1.2 M a day left for all GPU coins after | 2Miners, The Block |
The arithmetic is the whole story: the receiving chains had one twentieth of the income and received, for a day, one third of the hash. Difficulty did the rest. The 2Miners advice of the day was to pick ETC, RVN or ERG "at least for the first days", and the first days were all there were.
### 1b. Promised against delivered
| Chain | The launch text (quoted or paraphrased) | What GPU miners got | Months of GPU income |
|---|---|---|---|
| Grin | 90 percent of blocks to the GPU hash at launch, falling to zero over two years (forum, Aug 2018) | Exactly that | About 18 |
| Ravencoin | GPU-only, fork away FPGAs (KAWPOW, May 2020) | GPU-only for six years; a header bug, not a chip, split the chain in 2026 | 100 and counting, under power since about 2023 |
| Kaspa | "GPU PoW", fair launch (bitcointalk ANN title) | 17 months, then chips at 20x and more per joule | 17 |
| Karlsen | "We will ensure long-term GPU-friendly mining" (README) | A second hash in nine months to keep the promise; income under power either way | 10 |
| Iron Fish | A GPU-mined private chain | Unknown hardware inside weeks; a GPU hash a year later | 12, then under power |
| Aleo | Proving as mining on GPUs | USD 13.50 a day per 4090 for a few weeks, then price and pools | 3 |
| Flux | GPU-only, "resistance to concentrated specialized-hardware advantages" | Seven years, then block production moved to nodes | 84, then closed |
| Clore | Mine the coin that pays for GPU rental | Proof of work ended in December 2025; rental stayed | 36 |
| Nervos | GPU-mineable launch | 16 days of 10x hashrate growth from secret chips four months in | 4 |
What the table says, chain by chain, in one short paragraph each.
**Litecoin.** The first GPU coin and the first proof that a memory-hard hash on paper is a compute-bound hash in a fab. Scrypt with a 128 kB scratchpad fit on a chip in 27 months. GPU miners did not vote or fork; they sold the cards. Lesson for Igneum: the dataset has to be bigger than any die the chip can afford, and the lane's own history (asic-resistance-history.md rows Scrypt) already prices this.
**Dogecoin.** The only chain on the list whose GPU exit was a decision, not a defeat. Merged mining handed security to Litecoin's ASICs and the hashrate rose 15x in a month. Lesson: a small chain that shares hardware with a big one is mined by the big one's miners whenever the big one's miners choose; Igneum's hash has no big sibling, which is a feature and a cost.
**Ethereum.** Seven years of GPU income, three cycles, one exit that was announced for five years and still removed 94 percent of all GPU mining income in one block. The money went to ETC, RVN and ERG for days and then to the shelf. Lesson: miners follow income, not loyalty, and an exit that is scheduled is still an exit.
**Ethereum Classic.** The refugee chain. It absorbed a quarter of Ethereum's hashrate within hours and lost half of that within four days, then went to ASICs. Lesson: a chain that receives a migration gets a difficulty shock, not a community.
**Monero.** Four hash forks in 20 months kept chips off and cost 85 percent of hashrate each time (asic-resistance-history.md). RandomX then chose CPUs on purpose and three quarters of the GPU miners left in a day. In 2025 a rented CPU fleet (Qubic) claimed a majority and produced a 6-block reorg; the peer-reviewed analysis finds no sustained majority and no profit. Lesson: ASIC resistance by hand is a treadmill; a majority that cannot profit still frightens exchanges.
**Zcash.** The clearest record of a project refusing to fork for its GPU miners. The forum poll ran in February 2018, the Z9 shipped in May, the company stayed neutral, the GPUs left by autumn. Income per 1080 Ti fell from USD 8 to USD 2.63 before the chip arrived, because the price fell first. Lesson: the price kills before the chip does.
**Ravencoin.** The one chain that forked three times and kept the GPUs for six years. KAWPOW cut hashrate 10x on fork day because the FPGAs left. Then in August 2026 a validation bug (not a chip) split the chain. Lesson: random-program hashes work; the client is the next weak point, and Igneum has one client.
**Ergo.** A GPU-only hash that never saw a chip and still pays USD 0.22 a day per 3080. The Merge gave it 7x the hashrate in a day and the difficulty took the income back in weeks. Lesson: the hash is not the income.
**Kaspa.** The template Igneum forked. A fair launch, a compute-bound hash, 17 months of GPU income, then IceRiver and Bitmain in the same quarter. In April 2023 a 3070 made USD 1.21 a day beside a KS2 making USD 361. The GPU share went to nothing by the end of the year while the chain prospered. Lesson: the chain can win while its GPU miners lose; Igneum's whole bet is that these two must not separate.
**Grin and Beam.** Both launched in January 2019 into a bear market. Grin scheduled its own GPU lane to die over two years and it did. Beam never had a chip and still lost its miners to the price. Lesson: a launch into a falling market with a thin order book does not get a second chance; the ramp must survive a bad first year.
**Conflux, Firo, Iron Fish.** Three GPU-only hashes with no chip after five to six years, each paying under a dollar a day per card. Lesson: resisting the chip buys the right to be small.
**Alephium.** A sharded chain with a plain Blake3 hash, GPU-mined for about 30 months, then Goldshell shipped a 180 W box in May 2024 and a 1.25 TH/s box five months later. The GPUs were out by autumn and by 2026 the boxes themselves lose money at USD 0.10. Lesson: a chip on a small chain eats the GPUs and then eats itself; the chain gets neither back.
**Nervos.** The extreme case of private hardware. Hashrate rose 10x in 16 days in March 2020, four months after launch and a month before the first ASIC was on sale; a 300-card GPU farm went from USD 410 a day to USD 4.46. Lesson: the first ASIC is always private, and the launch ramp has to assume one.
**Radiant.** Said "GPU/ASIC" in its own announcement title and went to ASIC on schedule in September 2024. Lesson: a chain that invites the chip gets the chip; honest, and not Igneum's path.
**EthereumPoW.** The fork that promised Ethereum mining would continue. The coin traded above USD 140 for a day and under USD 2 by 2025; the pools left for Ethereum Classic within months. Lesson: a fork keeps the code and loses the market; the market is what paid the miners.
**Karlsen, Pyrin, Nomura.** Three kHeavyHash forks launched into the gap the Kaspa chips opened. Karlsen kept its README promise by changing its hash within a year; Pyrin was called a premine scam on launch day; Nomura is a title. Lesson: "the GPU version of X" is a business model with a nine-month half-life.
**Nexa.** Fair launch, CPU first, the top of the profitability tables for a summer, then price. Lesson: being the most profitable coin on WhatToMine for a month is a sell signal, not a community.
**Aleo.** The one chain where mining was proving. The fastest provers took the reward, the pools took the rest, the price fell from USD 9 to USD 3.40 and the GPUs left; the record also shows what miners say when insiders hold the supply. Lesson: Igneum's separation of lottery and proving is the right answer to the first half; the allocation half is answered by having no allocation.
**Flux, Clore, Dynex, Qubic.** Four projects that stopped paying GPUs for hashing and told miners to rent, host or train instead. Flux moved block production to nodes, Clore ended proof of work and pays for rental, Dynex's marketplace never paid, Qubic became a hashrate broker for Monero and then Dogecoin. Lesson: "useful work" projects abandon the miner the day the useful work does not pay; Igneum's proving must be priced to clear (Horizon rank 7) or it is Dynex.
**Vertcoin, Bitcoin Gold, Musicoin.** The three small chains 51 percent attacked by rented hash, between USD 70 k and USD 18 M a time. Each was GPU-only and each was attacked because it was. Lesson: a GPU chain is rentable by definition; finality has to be something hash cannot buy.
**EthereumPoW, TurtleCoin, Akroma, Callisto.** Dead by price, by halt, by abandonment, by a team's money running out. Lesson: a chain with no market is a chain with no miners within a year.
## 2. What GPU miners say they want
Twenty posts and articles read in full or in snippet, 2018 to 2026. Each quote is under 15 words. Reddit threads could not be fetched from this environment, so the miner voice here comes from project forums, pool blogs, bitcointalk snippets and press that quoted miners by name.
| # | Date | Forum or source | Quote | URL |
|---|---|---|---|---|
| 1 | 25 Feb 2018 | Zcash forum, user vgm | "Centralized ASIC mining will destroy hobby mining along with any organic user growth potential." | forum.zcashcommunity.com/t/lets-talk-about-asic-mining/27353 |
| 2 | 25 Feb 2018 | Zcash forum, user JKDC | "Decentralization is as important as privacy." | same thread |
| 3 | 26 Feb 2018 | Zcash forum, user aoeu | "I'd rather have GPUs for general computation, not silicon trash." | same thread |
| 4 | 25 Feb 2018 | Zcash forum, user Autotunafish | "Zcash is about universal accessibility, not a select few" | same thread |
| 5 | 27 Aug 2018 | Grin forum, igno.peverell (launch text) | "A healthy and grassroot GPU mining community at launch is highly desirable." | forum.grin.mw/t/proof-of-work-update/713 |
| 6 | 30 Nov 2019 | MinerUpdate (RandomX day) | "Honest miners will be fighting an uphill battle against botnets" | minerupdate.com (section 6) |
| 7 | 21 Mar 2020 | 2Miners blog (Nervos) | "ASICs or FPGA's are to blame for a sudden increase in the hash rate." | 2miners.com/blog/nervos-ckb-network-hashrate-increased-asics-are-the-cause/ |
| 8 | 9 Apr 2020 | VoskCoinTalk, user Rocky_Bro | "Those 3000watt beasts just eat slower devices for breakfast." | voskcointalk.com/t/.../172 |
| 9 | 12 Sep 2022 | The Block, Ethan Vera (Luxor) | "You have to have sub $0.03 per kilowatt hour and new generation GPUs." | theblock.co (section 6) |
| 10 | 12 Sep 2022 | The Block, Mark D'Aria (Bitpro) | "Ethereum is 95% of the total income. Everything else will get crushed." | same article |
| 11 | 12 Sep 2022 | 2Miners blog, Confessions of a Miner 2 | "When mining makes you lose the money you stop mining." | 2miners.com/blog/confessions-of-a-miner-2-is-this-the-end-of-gpu-mining/ |
| 12 | Sep 2022 | Hiveon pool, year review | "16% moving to ETC, 7% going to RVN" | hiveon.com/news/reflection-and-forecast-... |
| 13 | 2023 (approximate) | bitcointalk, "Kaspa asics" thread (snippet) | "Specialized mining equipment optimized for corporate industry raping" | bitcointalk.org/index.php?topic=5449022.0 |
| 14 | Dec 2023 | bitcointalk, Pyrin ANN thread (snippet) | "another pre-mine scam" | bitcointalk.org/index.php?topic=5476198.0 |
| 15 | 2023 | bitcointalk thread title | "Why most premined coins (>90% of all altcoins) will eventually fail" | bitcointalk.org/index.php?topic=5457833.0 |
| 16 | 14 May 2024 | Iron Fish blog (on its own launch) | "unknown origin and hash power" | ironfish.network/learn/blog/2024-05-14-fish-hash-audit |
| 17 | 2024 | Karlsen README (launch text) | "We will ensure long-term GPU-friendly mining." | github.com/karlsen-network/karlsend |
| 18 | Sep 2024 | PANews via AiCoin, Aleo community user | "When you don't know where the liquidity comes from, you are the liquidity." | aicoin.com/en/article/419859 |
| 19 | 31 Jul 2019 | 2Miners blog (delisting) | "Musicoin is dead, Akroma is irrelevant." | 2miners.com/blog/akroma-and-musicoin-delisting/ |
| 20 | 10 Feb 2026 | Clore blog | "losing money mining with most mid-range GPUs at average electricity rates" | blog.clore.ai/gpu-mining-is-dead-... |
### The three wants, with counts
| Want | Posts asking for it (of 20) | Evidence rows |
|---|---|---|
| 1. Hardware that stays useful: no chip, no FPGA, a card that games or computes when the coin dies | 10 | 1, 2, 3, 4, 5, 7, 8, 13, 16, 17 |
| 2. Income above electricity that does not fall off a cliff | 6 | 9, 10, 11, 12, 19, 20 |
| 3. A fair supply: no premine, no insider allocation, a coin somebody will buy | 4 | 14, 15, 18, 19 |
### The three hates, with counts
| Hate | Posts (of 20) | Evidence rows |
|---|---|---|
| A. Secret hardware on the chain before anyone can buy it | 5 | 7, 8, 13, 16, and Nervos (section 1) |
| B. Premine, VC allocation, locked airdrops | 3 | 14, 15, 18 |
| C. The income cliff and the dead market (an exit, a delisting, a 99 percent drawdown) | 5 | 10, 11, 12, 19, 20 |
### Mapped to Igneum's design (horizon-2026-10.md section 1 and the litepaper)
| Want or hate | Status | Mechanism, and the honest gap |
|---|---|---|
| Want 1, hardware stays useful | Partly answered | Random-program hash with hourly swaps, 256 MB dataset with 8 dependent reads, era draws from chain state, no scheduled fork (CLAUDE.md design paragraph). The chip model says 2.1x per joule at class v4 and the N ladder is a decision owed (Horizon item 4 and rank 10). Kaspa's chip was 20x to 1,200x; 2.1x is a different world, but a 2.1x chip at scale still ends home mining on power price alone. The gap is the ladder decision and a public chip model per era |
| Want 2, income without a cliff | Partly answered | 100 IGN a block gliding 2.9 percent a month, no halving day, a 1 percent tail (tail-emission.md). No chain on this list ever had a cliff-free schedule, so this is new. The gap: every exit in section 1 came from price, not schedule, and no protocol sets price. What Igneum can do is publish the per-tier income at three prices before launch (the consequences rule) so nobody buys a card on a number that was never promised |
| Want 3, fair supply | Answered | No premine, no dev fund, no fee to any team, 90-day ramp from 10 percent (CLAUDE.md, tail-emission.md). Aleo and Pyrin are the counter-examples and Igneum has nothing they had |
| Hate A, secret hardware at launch | Partly answered | The ramp from 10 percent makes the first 90 days worth less to a private farm; the correlated-group detector and `security_alert` (Horizon rank 16) show it; the hourly program swap means a private FPGA bitstream has to be re-made every hour. Not answered: a private rented GPU fleet is not hardware and is allowed; Nervos's 16-day 10x rise would look the same on Igneum's charts. The gap is a published launch-week hash-origin report (pools, key counts, fleet shares) so the community sees it the day it happens |
| Hate B, premine and allocations | Answered | As Want 3 |
| Hate C, the cliff and the dead market | Not answered by protocol | The glide removes the schedule cliff and the proving market gives the coin a buyer other than an exchange (Horizon 4.1a), but a rollup customer paying in dollars is a contract not yet signed (Taiko is the first named customer). Until a customer pays, Igneum has the same exposure as Ergo in 2023: a good hash, a thin market |
One more thing the posts do not say and the record does: nobody on these forums asked for finality, for a DAG, or for proving. They asked for a card that keeps its value, a bill that gets paid, and a coin that sells. Igneum's technical answers are to questions miners did not ask; the mission should put the three wants on the front page and the finality on page two.
## 3. The earnings curve as a pattern
USD per card per day before electricity. A 3080 draws about 220 W, so electricity at USD 0.10 per kWh is USD 0.53 a day; a 4090 at 400 W is USD 0.96. Where a figure is approximate it says so.
| Chain | Launch | Peak month | Peak income (per 100 MH Ethash or per 3080 equivalent) | Month income fell under power at USD 0.10 | Hashrate that left in the next 90 days |
|---|---|---|---|---|---|
| Ethereum | Jul 2015 | Jan 2018, then May 2021 | USD 23 per 100 MH (Jan 2018, USD 0.233 per MH); approximate USD 12 to 15 per 3080 in May 2021 | Never for a 3080 before the Merge (approximate USD 1 to 2 a day in Sep 2022, over power); the lane closed 15 Sep 2022 | 100 percent of Ethereum's 900 TH/s; of the GPUs, about a quarter tried ETC and half of those left inside four days (CoinDesk, AMBCrypto); approximate 80 percent of the cards were off within 90 days |
| Kaspa | Nov 2021 | Feb to Mar 2023 (approximate) | Approximate USD 2 to 3 per 3080 | Approximate Sep to Oct 2023 (a 3070 was at USD 1.21 in Apr 2023 as the first ASICs landed) | GPU share from near 100 percent to under 10 percent by Dec 2023 (approximate); the network's total hashrate rose the whole time |
| Ergo | Jul 2019 | Sep to Nov 2021 | Approximate USD 3 to 4 per 3080 | Approximate Nov to Dec 2022 (the Merge took hashrate from 22 to 155 TH/s in a day) | Approximate 50 to 70 percent of the post-Merge peak within 90 days |
| Ravencoin | Jan 2018 | Feb to Apr 2021 | Approximate USD 6 to 8 per 3080 | Approximate Nov 2022 (the FTX month); a 3080 now loses USD 0.13 to 0.31 a day | Approximate 40 percent of the post-Merge peak (15.9 TH/s) within 90 days |
| Aleo | Sep 2024 | Sep 2024 | USD 13.50 per 4090 at USD 9 (PANews) | Approximate Q1 2025 as the price reached USD 3.40 and difficulty followed the pools | "Difficulty plummeted ... as miners exited" (PANews); approximate over 50 percent of GPU provers within 90 days of the token-economics reveal |
What the curve means per tier (the consequences rule), using the record's ratios rather than a price nobody can promise:
| Tier | What the record says happens to them | What Igneum should tell them before launch |
|---|---|---|
| Home miner, one 8 or 12 GB card | First under power in every exit (the 3060 lost money on Clore's 2026 table while the 4090 still cleared) | Income per card at three prices, and that the glide never halves it overnight; the card keeps gaming when the income goes |
| Home miner, one 16 to 32 GB card | Last GPU standing; still under power 6 to 18 months after a peak on every chain in the table | The proving pool is the second income for this tier only; say so, with the measured 15.6 GB profile |
| Rig (6 to 12 cards) | Followed income across chains inside days (Hiveon: 16 percent to ETC, 7 percent to RVN) and shut down inside 90 days | Rig income at three prices; no loyalty expected, none needed, the 30-day weight pays staying miners a vote |
| Pool user | Took the pool's choice of chain and the pool's fee; pools held 29 percent of Aleo proving at launch | Which pools carry the finality signer, and that a pool's vote is the user's weight (ledger G8) |
The common shape, in numbers: income peaks inside one to six months of a price run or a migration, halves within 60 to 120 days as difficulty catches the price, and falls under power within six to eighteen months of the peak; when it does, 50 to 100 percent of the GPU hashrate leaves inside 90 days. The trigger was a chip twice (Kaspa, and ETC's second decline), a scheduled exit once (Ethereum), and the price three times (Ergo, Ravencoin, Aleo). In no case did a GPU lane recover its peak income after the exit.
## 4. Verdict table
| # | Lesson | Evidence (chain, year) | What Igneum does today | Gap | What the mission should add or change |
|---|---|---|---|---|---|
| 1 | A compute-bound hash goes to a chip in 16 to 37 months, and the first chip is private | Litecoin 2014, Zcash 2018, Nervos 2020, Kaspa 2023, Alephium 2024 | Random-program hash, 256 MB dataset, latency shadow, era draws; chip model 2.1x at class v4 (Horizon item 4) | Small | Decide the N ladder at genesis and publish the chip model with every era draw; nothing else |
| 2 | Secret hardware or fleets show up as an unexplained hashrate step in the first weeks | Nervos 2020 (10x in 16 days), Iron Fish 2023, Kaspa 2023 | 90-day ramp from 10 percent; correlated-group detector and `security_alert` (Horizon rank 16) | Small | A public launch-week hash-origin report (key counts, pool shares, fleet shares) from the observer, daily for the first 90 days |
| 3 | Hash forks by hand lose miners and add bugs; the client is the next failure | Monero 2018 to 2019 (85 percent gone per fork), Vertcoin 2018 to 2019, Ravencoin Aug 2026 (header flaw) | No scheduled human fork; era draws from chain state; 95 percent signalling with a floor; Devnet 2 gate | Large on the client | The second independent client is the item; until then a fuzz gate on header and PoW validation paths equal to the sync-request fuzz (Horizon rank 9) |
| 4 | Income cliffs empty a chain in days; miners follow income with no loyalty | Ethereum Sep 2022 (94 percent of income gone), ETC -48 percent in four days, Aleo Q4 2024 | Monthly glide, no halving day, 1 percent tail (tail-emission.md) | Small on schedule, large on price | Publish per-tier income at three prices (USD 0.005, 0.02, 0.10) in the litepaper before the testnet, per the consequences rule; nothing in protocol |
| 5 | Premine and insider allocation drive GPU miners out faster than any chip | Aleo 2024, Pyrin 2023 | No premine, no dev fund, no fee to any team | None | Nothing |
| 6 | A GPU chain is rentable; rented hash reorganised three of them for USD 70 k to USD 18 M | Bitcoin Gold 2018 and 2020, Vertcoin 2018 and 2019, Musicoin 2019 | Miner-only finality on 30 days of blocks; 20 days of 100 percent hash to reach 2/3 (CLAUDE.md) | Small after day 20, large before | Weight-gated deep fork choice (Horizon rank 4) covers the first 20 days; ship it before mainnet, not after |
| 7 | A coin with no market dies within a year whatever the hash does | Musicoin 2019, Akroma 2019, TurtleCoin 2023, Grin (USD 16 k daily volume), Nexa | Proving sold in dollars, settled in IGN (litepaper); Taiko named as first customer | Large until a customer pays | A signed proving customer before mainnet is a launch gate, with the same weight as the Devnet 2 gate; decouple the job price from `f_p` (Horizon rank 7) so the first job can clear |
| 8 | Useful-work projects drop the miner the day the work does not pay | Dynex 2024, Flux 2025, Clore 2025, Qubic 2025 to 2026 | The lottery pays 80 percent of every block whatever the proving market does; proving is the second income, never the only one | None by design | Write the sentence into the litepaper: "A block pays its miner whether or not anyone buys a proof that day" |
| 9 | Random-program hashes keep GPUs for five years and more, and still pay under a dollar a card | Ravencoin 2020 to 2026, Firo, Conflux, Iron Fish | The hash is Igneum's strongest answer | None on the hash | Stop presenting the hash as the reason to mine; present income, fairness and the second income (section 2's three wants) first |
## 5. Headline findings for the coordinator
1. Of the 31 rows (33 chains) that paid GPU miners since 2011, 8 lost the GPU lane to a chip, 7 closed it by their own choice (Dogecoin, Ethereum, Monero, Grin, Flux, Clore, Qubic), 10 died or went dormant on price, rental attacks or abandonment, and the 6 still GPU-only (Ravencoin, Ergo, Firo, Conflux, Iron Fish, Beam) pay USD 0.22 to 0.71 a day per RTX 3080 before power, so the hash keeps the miner and not the income.
2. In 20 miner posts and launch texts read, 10 asked for hardware that keeps its value, 6 for income above electricity without a cliff and 4 for a fair supply; Igneum answers the fair supply in full, the hardware in part (a 2.1x chip model with the N ladder still owed), and the income only on schedule, since every GPU exit in the record except two came from price.
3. The earnings curve has one shape across Ethereum, Kaspa, Ergo, Ravencoin and Aleo: income halves within 60 to 120 days of the peak, falls under USD 0.10 per kWh within 6 to 18 months, and 50 to 100 percent of GPU hashrate leaves inside 90 days of that, so the launch plan needs a signed proving customer and published per-tier income at three prices before the first miner buys a card.
## 6. Sources (all accessed 7 October 2026 unless noted)
Project and pool sources
- Zcash forum, "Let's talk about ASIC mining", Feb 2018: https://forum.zcashcommunity.com/t/lets-talk-about-asic-mining/27353
- Grin forum, "Proof of work update", 27 Aug 2018: https://forum.grin.mw/t/proof-of-work-update/713
- 2Miners, Nervos hashrate and ASICs, 21 Mar 2020: https://2miners.com/blog/nervos-ckb-network-hashrate-increased-asics-are-the-cause/
- 2Miners, Akroma and Musicoin delisting, 31 Jul 2019: https://2miners.com/blog/akroma-and-musicoin-delisting/
- 2Miners, End of Ethereum mining, 30 Aug 2022: https://2miners.com/blog/end-of-the-ethereum-mining-when-what-to-mine-next-with-gpu-transition-to-pos-explained/
- 2Miners, Confessions of a Miner 2, 12 Sep 2022: https://2miners.com/blog/confessions-of-a-miner-2-is-this-the-end-of-gpu-mining/
- 2Miners, How to mine Kaspa with ASIC and GPU, 14 Apr 2023: https://2miners.com/blog/how-to-mine-kaspa-with-asic-and-gpu-settings-extra-rewards-dual-mining/
- 2Miners, July 2023 report (Nexa, Kaspa KS0 port): https://2miners.com/blog/july-2023-work-progress-report-new-coin-nexa-kaspa-geo-servers/
- 2Miners, KawPoW fork, May 2020: https://2miners.com/blog/kawpow-new-ravencoin-mining-algorithm/
- Hiveon, PoW to PoS notes, 24 Aug 2022: https://hiveon.com/news/the-pow-and-pos-transition-important-notes/
- Hiveon, 2022 review (16 percent ETC, 7 percent RVN): https://hiveon.com/news/reflection-and-forecast-reviewing-the-events-of-2022-and-anticipating-the-year-ahead/
- VoskCoinTalk, Antminer K5 impressions, 9 Apr 2020: https://voskcointalk.com/t/initial-impressions-of-the-bitmain-antminer-k5-nervos-ckb-eaglesong-asic-miner-mining-22-a-day/172
- Iron Fish, FishHash audit and the reason for the switch, 14 May 2024: https://ironfish.network/learn/blog/2024-05-14-fish-hash-audit
- Iron Fish, hard fork 1 activation, 2 Apr 2024: https://ironfish.network/learn/blog/2024-02-26-mainnet-hardfork
- Karlsen README: https://github.com/karlsen-network/karlsend
- hashrate.no, Karlsen and Pyrin algorithm change, 13 Sep 2024: https://www.hashrate.no/c/Algorithm_change_for_Karlsen_and_Pyrin
- hashrate.no, RTX 3080 today (top coins and USD per day): https://hashrate.no/gpus/3080
- hashrate.no, Kaspa network (347 PH/s): https://www.hashrate.no/coins/KAS
- hashrate.no, Aleo pools (f2pool 28.9 percent): https://www.hashrate.no/coins/ALEO/pools
- RunOnFlux, Proof of Useful Work v2, 23 Oct 2025: https://runonflux.com/forking-flux-proof-of-useful-work-v2/
- Clore changelog (PoW end, Dec 2025): https://clore.ai/changelog
- Clore blog, GPU mining is dead, 10 Feb 2026: https://blog.clore.ai/gpu-mining-is-dead-gpu-hosting-is-the-future-heres-how-to-earn-500month/
- Qubic blog, mining evolution, 30 Jun 2025: https://qubic.org/blog-detail/qubic-mining-evolution-from-cpu-roots-to-gpu-dominance-and-back-again
- Qubic blog, Dogecoin mining on Qubic: https://qubic.org/blog-detail/qubic-dogecoin-mining-how-it-works
- TurtleCoin v2 FAQ (halt at block 5,500,000, 15 Mar 2023): https://hackmd.io/@iburnmycd/rJzyWD3-_
- Callisto history (2024 collapse): https://www.callisto-pirl.com/history
- bitcointalk, Kaspa ANN: https://bitcointalk.org/index.php?topic=5373286.0 (403 on fetch; title and snippet via search)
- bitcointalk, "Kaspa asics": https://bitcointalk.org/index.php?topic=5449022.0 (snippet only)
- bitcointalk, Pyrin ANN: https://bitcointalk.org/index.php?topic=5476198.0 (snippet only)
- bitcointalk, Karlsen ANN: https://bitcointalk.org/index.php?topic=5475216.0 (snippet only)
- bitcointalk, "Why most premined coins will eventually fail": https://bitcointalk.org/index.php?topic=5457833.0 (title only)
- asicminervalue, Antminer KS7: https://www.asicminervalue.com/miners/bitmain/antminer-ks7-40th
- asicminervalue, Goldshell E-AL1M (negative at USD 0.10): https://www.asicminervalue.com/miners/goldshell/e-al1m
- asicminervalue, Ethereum Classic: https://www.asicminervalue.com/coins/etc-ethereum-classic
- minerstat, Grin: https://minerstat.com/coin/grin_cuckatoo32
- WhatToMine, Ravencoin on RTX 3080 LHR: https://whattomine.com/coins/234-rvn-kawpow/gpus/63-nvidia-geforce-rtx-3080-lhr
- Binance Research, merged mining case study: https://www.binance.com/en/research/analysis/merged-mining
- litecoin.watch, scrypt history: https://litecoin.watch/articles/understanding-the-scrypt-mining-algorithm
Press and papers
- CoinDesk, ETC and RVN hashrate after the Merge, 15 Sep 2022: https://www.coindesk.com/tech/2022/09/15/ethereum-classic-and-ravencoins-hashrate-nearly-doubles-after-merge
- AMBCrypto, ETC post-Merge high wearing off, 19 Sep 2022: https://ambcrypto.com/why-ethereum-classics-etc-post-merge-high-is-wearing-off/
- crypto.news, ETC hashrate +280 percent: https://crypto.news/ethereum-classic-hashrate-jumped-280-post-merge/
- Decrypt, ETC, RVN and ERG hashrate soar, Sep 2022: https://decrypt.co/109803/ethereum-classic-ravencoin-ergo-hash-rate-soar-post-merge
- The Block, what is left for GPU miners, 12 Sep 2022: https://www.theblock.co/news/ecosystems/2022-09-12-whats-left-for-ethereum-gpu-miners-after-the-merge-168380
- Tom's Hardware, GPU mining unprofitable after the Merge (900 TH/s, about 20 million GPUs): https://www.tomshardware.com/news/gpu-mining-is-now-unprofitable
- Wccftech (Minerstat figures, RTX 3080 USD 6.35 to 9.15 a day, early 2021): https://wccftech.com/rgb-lit-bitcoin-mining-rig-78-geforce-rtx-3080-graphics-cards-comes-operational/
- FXStreet, Ethereum miners in the red (USD 0.233 per MH per day in Jan 2018), 29 Mar 2018: https://www.fxstreet.com/amp/cryptocurrencies/news/ethereum-price-analysis-eth-usd-eyes-40000-amid-cryptocurrency-sell-off-miners-are-deep-in-red-201803290719
- MinerUpdate, CPU mining dominates Monero since RandomX, 30 Nov 2019: https://www.minerupdate.com/news/miner-insights/cpu-mining-dominates-monero-since-randomx-upgrade
- Bitcoin Magazine, Antminer Z9 mini, May 2018: https://bitcoinmagazine.com/business/bitmains-antminer-z9-mini-designed-mine-zcash-threatens-asic-resistance
- 2Miners, Zcash mining and profitability (1080 Ti at USD 8 in Jul 2017, USD 2.63 on 6 May 2018): https://2miners.com/blog/how-to-mine-zcash-zec-mining-and-profitability/
- Medium (Blockwise), the end of GPU mining (Monero April 2018 fork halved hashrate): https://medium.com/blockwise/the-end-of-gpu-mining-a702921fc1e1
- crypto.news, Ravencoin consensus flaw, 11 Aug 2026: https://crypto.news/ravencoin-falls-as-consensus-flaw-splits-network/
- Decrypt, the long collapse of Bitcoin Gold: https://decrypt.co/47381/the-long-collapse-of-bitcoin-gold
- metalicjames gist, Bitcoin Gold Jan 2020 attack: https://gist.github.com/metalicjames/71321570a105940529e709651d0a9765
- metalicjames gist, Vertcoin Dec 2018 attack: https://gist.github.com/metalicjames/f2acdb9ef448ec5298173b36c7c54133
- news.bitcoin.com, Vertcoin second attack, Dec 2019: https://news.bitcoin.com/vertcoin-network-sabotaged-by-another-51-attack/
- Binance Square, BTG delisting 1 Sep 2024: https://www.binance.com/en/square/post/12128321695753
- AiCoin (PANews), Aleo mainnet, miners cry foul, Sep 2024: https://www.aicoin.com/en/article/419859
- Aleo, mainnet announcement, 18 Sep 2024: https://aleo.org/post/announcing-aleo-mainnet/
- The Block, Monero reorg fears after Qubic's 51 percent claim, 12 Aug 2025: https://www.theblock.co/post/366535/monero-faces-chain-reorganization-fears-after-qubic-says-it-controls-51-of-hashrate
- Lee and Kim, Inside Qubic's selfish mining campaign on Monero, arXiv 2512.01437, AFT 2026: https://arxiv.org/abs/2512.01437
- Decrypt, Qubic mining Dogecoin, Apr 2026: https://decrypt.co/363018/qubic-the-network-that-captured-51-of-monero-is-now-mining-dogecoin-on-its-ai-compute-infrastructure-live
- CoinGecko, Dynex (99.9 percent below ATH): https://www.coingecko.com/en/coins/dynex
- Medium (crypto-blog), Dynex the most GPU mined coin, Oct 2023: https://medium.com/crypto-blog/dynex-dnx-seems-to-be-the-most-gpu-mined-coin-at-the-moment-cdf7d30c433a
- KuCoin, ETHW price history: https://www.kucoin.com/price/ETHW
- Kryptex, IceRiver KS series: https://pool.kryptex.com/articles/iceriver-ks0-ks1-ks2-ks3-ks3l-en
- Medium (VoskCoin), Was Kaspa mining worth it (KS0 USD 25 a day, Jul 2023): https://medium.com/voskcoin/was-kaspa-mining-worth-it-ca3ac9ff7566 (403 on fetch; figures via search snippet)
Internal
- `docs/analysis/asic-resistance-history.md` (5 Oct 2026)
- `docs/analysis/horizon-2026-10.md` section 1 and ranks 4, 7, 9, 10, 16 (6 to 7 Oct 2026)
- `docs/analysis/horizon/economy-and-utility.md` section 4.6
- `docs/analysis/tail-emission.md`, `docs/analysis/vote-or-burn.md`
Not reachable from this environment (recorded so the coordinator can fetch by hand): reddit.com (all subreddits), bitcointalk.org thread bodies, web.archive.org, the Springer paper "The PoW Landscape in the Aftermath of The Merge" (login wall), statista's Ethereum profitability series (paywall), bitinfocharts chart values (rendered client-side).