diff --git a/docs/analysis/chip-model-v3.md b/docs/analysis/chip-model-v3.md index 9eb536b17..b35e6bc10 100644 --- a/docs/analysis/chip-model-v3.md +++ b/docs/analysis/chip-model-v3.md @@ -181,7 +181,7 @@ speed, so HBM3E is HBM3 here, and 48 Gbps GDDR7 is 28 Gbps GDDR7. |---|---|---|---| | Channels, banks | 64 channels, 1,024 banks | 16 channels (32 pseudo-channels), up to 1,024 banks | 128 channels, 8,192 banks | | Bank-bound ceiling, banks / 45 ns (tRC, the HBM2 and GDDR5-class figure, approximate for both) | 22.8 G reads/s | 22.8 | 182 | -| Activate-bound ceiling (GDDR7: 4 per 12 ns per channel, approximate; HBM: 8 per 12 ns per channel, O'Connor Table 2) | 21.3 G reads/s | 10.7 | 85.3 | +| Activate-bound ceiling (GDDR7: 4 per 12 ns per channel, approximate; HBM: 8 per 12 ns per channel, O'Connor Table 2). UNMEASURED (6 October 2026, the Horizon lane analysis `docs/analysis/horizon/algorithm.md` section 5.1): the JEDEC HBM2 table gives tFAW 28 ns, 4 activates per channel per window (ICCAD 2021 Table I), which is 2.3 G reads/s for a 16-channel stack, and the one measured random-read rate of an HBM2 part (Shuhai, Alveo U280, FCCM 2020 Fig 7) is 2.4 G, equal to that tFAW ceiling; the 12 ns figure holds only if a bank-interleaved mapping lifts tFAW, which the die enforces per channel. The HBM columns below carry the 10.7 G row as the model's ceiling, unmeasured; an AWS F2 hour (Virtex UltraScale+ VU47P, the same HBM2 subsystem) is the measurement | 21.3 G reads/s | 10.7 (unmeasured; 2.3 at JEDEC tFAW) | 85.3 (unmeasured) | | The ceiling carried below | 21.3 G reads/s (the 5090 measures 17.5, 82 percent of it: the card is already near its memory's activate limit) | 10.7 | 85.3 | | Energy per random 32-byte read (approximate) | 2.0 nJ: 909 pJ activation (one atom per row opened, the HBM2 1 KB row taken for GDDR7's row) plus 4.5 pJ per bit x 256 bits of movement and I/O = 1,150 pJ | 1.2 nJ: the HBM2 sum (909 + (1.51 + 1.17 + 0.80) x 256 = 1,800 pJ) scaled by Samsung's 4.12 / 6.25 | 1.2 nJ | | Static power (refresh, standby, PLLs; approximate, from memory) | 20 W (about 1.25 W per device) | 4 W | 32 W | @@ -191,6 +191,8 @@ speed, so HBM3E is HBM3 here, and 48 Gbps GDDR7 is 28 Gbps GDDR7. | Reads per second per watt at the ceiling (memory, static and controller) | 0.27 G | 0.40 G | 0.49 G | | The 5090 for comparison | 17.5 G reads/s at 326 W = 0.054 G per W; 7,262 reads in flight (17.5 G x 415 ns), 22 per watt | | | +The FPGA line, public (6 October 2026, the Horizon lane analysis section 5.1): an HBM2 FPGA soft overlay (Alveo U280 or U55C class) carries only the measured row, 2.4 G reads/s per card (Shuhai, FCCM 2020 Fig 7, equal to the JEDEC tFAW ceiling of 2.3 G at 28 ns), which is 0.30x to 0.39x of the RTX 5090 per watt (U55C at 115 to 150 W; 0.20x on the U280). The 11.4 G bank-bound row and the 12.2 G ceiling quoted elsewhere rest on a 12 ns tFAW the JEDEC HBM2 table does not give and are unmeasured until an AWS F2 hour (f2.6xlarge, VU47P, 16 GB HBM2, USD 1.98 an hour on demand) runs the chase kernel at 1 GiB across all 32 pseudo-channels; the lane's pass line is 15 to 25 M reads/s/W, its alarm line 27 (0.5x of the 5090), and over 54 (1.0x) the FPGA lane becomes a Counter ASIC 4.0 item. Consequence per tier: none today (no FPGA mines); if the measured row holds, a soft-overlay FPGA at USD 4,000 to 5,000 a card (approximate) mines at an RX 9070 XT's rate per watt for 7x the price, so no home or rig tier is displaced by it. Ledger M33. + The GDDR7 system's own power at the 5090's 17.5 G reads/s is 17.5 x 2.0 nJ = 35 W plus 20 W static, 55 W: about 17 percent of the card's 326 W (approximate). The other 83 percent is the GPU: 21,760 ALUs spinning at 92.9 percent utilisation on 512 program ops per hash, their register files, schedulers, L1 and L2, and the clock trees, against a diff --git a/docs/fud-ledger.md b/docs/fud-ledger.md index 8e162a2c9..60fae6e91 100644 --- a/docs/fud-ledger.md +++ b/docs/fud-ledger.md @@ -164,6 +164,24 @@ Sweep (5 October 2026, evening): stated. `site/litepaper.html`, "For miners", Ha --- +### M32. "Automatic anti-ASIC escalators" overstates what the era draw and the instruction reserve do +"You sell the era draw and the reserve unlock as anti-ASIC escalators, as if not knowing next era's parameters stops a chip. A chip that stores the dataset reads every drawn parameter as firmware: an address permute, a rotator, an immediate table. The families, the reserve order, the mixer, the dataset schedule and the class v4 shadow are all public at genesis. So what does the draw actually defend against?" + +Status: Conceded, stated (6 October 2026, evening; the Horizon lane analysis `docs/analysis/horizon/algorithm.md` sections 5.4 and 8, lane 2): the era draw and the instruction reserve are automatic schedule changes against fixed datapaths and against human forks; against the stored-dataset chip every drawn parameter is firmware, and the defence against that chip is the latency-shadow work (class v4) and the price-per-joule model. Stated in `site/litepaper.html`, Mining section ("These are automatic schedule changes ... every drawn parameter is firmware") and the "A chip is impossible" item ("a chip wired for one program is a bad bet ... not the schedule"), the "Every six months" row of the comparison table, and `site/index.html`, the hourly-program note ("a chip wired for one program is useless"). The phrase "automatic anti-ASIC escalators" is withdrawn from public text; it stays in the internal design summary until that is next edited. + +Answer: Correct. The draw hides (M, R, pos, the op weights within +-2, the fold rotations) until 2 hours before each era, and none of those needs silicon. Biasing the draw is priced at 20 days of 100 percent of the network's hash for one more sample of the same space (lane section 5.4), so the draw is unbiasable at any price that matters and that is its whole job: it is a fairness device and a fork-free schedule, not a chip defence. What a chip wired for one program loses to is the hourly program itself; what the stored-dataset chip loses to is the latency shadow (class v4, 2.1x per joule at k = 1 against the 5090 bench row, 0.9x against the Apple M5 Max) and the price per joule, which is where the public claim now rests. + +Evidence: `docs/analysis/horizon/algorithm.md` sections 5.4 (the draw's randomness, the two routes priced) and 8 (the summary), 6 October 2026; the six-era hash-rate spread of 0.8 to 3.2 percent per card in bench-log "Counter ASIC 2.0, the numbers". + +### M33. The FPGA ceiling rests on a tFAW the JEDEC HBM2 table does not give +"Your chip model's HBM random-read ceiling takes 8 activates per 12 ns per channel from O'Connor and gets 10.7 G reads/s a stack; the epoch-length page's bank-bound row gets 11.4 and a 12.2 ceiling. JEDEC HBM2 tFAW is 28 ns with 4 activates per channel per window: 2.3 G. The one measured HBM2 FPGA random-read rate (Shuhai, FCCM 2020) is 2.4 G, right on the JEDEC ceiling. Your 1.9x FPGA ceiling is arithmetic on a timing the part does not have." + +Status: Conceded, stated (6 October 2026, evening; the Horizon lane analysis `docs/analysis/horizon/algorithm.md` section 5.1, the FPGA lane): the public FPGA line carries only the measured row, 2.4 G reads/s per card and 0.30x to 0.39x of the RTX 5090 per watt (Shuhai, FCCM 2020 Fig 7; the tFAW arithmetic from ICCAD 2021 Table I), and the 11.4 G bank-bound row and the 12.2 G ceiling are marked unmeasured until an AWS F2 hour measures them. Stated in `docs/analysis/chip-model-v3.md` section 5.3 (the activate-bound row marked UNMEASURED with the JEDEC figure beside it, and the FPGA paragraph after the table). The epoch-length analysis's 12.2 row is not on master yet and is corrected when it lands. + +Answer: Correct. The measured 2.4 G/s had been read on 6 October as a mapping artefact ("the paper's point is that this mapping is the wrong one for random access"); the activate window says it is the DRAM's own limit, and a bank-interleaved mapping does not lift it because tFAW is enforced per channel by the die. The measurement that settles it is one AWS F2 hour (f2.6xlarge, Virtex UltraScale+ VU47P, 16 GB HBM2 in 2 stacks, 32 pseudo-channels, USD 1.98 an hour on demand): the chase kernel of `docs/benchmarks/repro.md` 2.2 ported to a Vitis HLS AXI master over the HBM IP at 1 GiB across all 32 pseudo-channels, 256 to 4,096 lanes in flight, board power at 1 Hz; pass line 15 to 25 M reads/s/W (0.3x to 0.5x of the 5090), alarm 27 (0.5x), over 54 (1.0x) a Counter ASIC 4.0 item. Consequence per tier: none today (no FPGA mines); on the measured row a soft-overlay FPGA mines at an RX 9070 XT's rate per watt for about 7x the price (approximate), so no home or rig tier is displaced. + +Evidence: `docs/analysis/horizon/algorithm.md` section 5.1 (the ceiling table: measured 2.4, tFAW-bound 2.3, tRRD-bound 2.8, bank-bound 11.4, O'Connor 10.7 G reads/s, and the F2 measurement plan), 6 October 2026; JEDEC HBM2 timings as carried by ICCAD 2021 Table I; Shuhai, FCCM 2020, Fig 7. + ## 2. Finality and attacks ### F1. Finality is attackable for the first month @@ -296,6 +314,15 @@ Evidence: design doc, "Proving speed on consumer GPUs" risk item and "Who needs" --- +### F26. "No stake" needs its one sentence: what is at stake, and what strips it +"You write 'no stake' and then run a vote whose weight can be stripped. Either nothing is at stake, in which case equivocation costs nothing, or something is, in which case say what it is and who can take it. One sentence, in the finality section and the summary, not an argument." + +Status: Conceded, stated (6 October 2026, evening; the Horizon lane analysis `docs/analysis/horizon/frontier.md` section 4.1 and item I13 of its incremental list): `site/litepaper.html`, the finality section's "What is not here" paragraph and the "Igneum at a glance" Finality row carry the sentence verbatim: "No coin is staked. The only thing at stake is 30 days of public work: a vote key's weight is its blue blocks over the window, and equivocation strips it for 30 days." The spec sentence for 03 and 05 (I13) follows with the lane's commit. + +Answer: Correct. The weight is a quantity that is at stake, earned by work alone over 30 days, not transferable, not purchasable, and already stripped in full for equivocation (spec 3.6). That is a slashable bond made of blocks, with no coin and no stake class; the "then it is stake" objection and its answer (the weight cannot be bought, borrowed or bridged, so capture by capital is removed; the last-days attack is bounded by the 30-day re-earn) are in the lane file, section 4.1. Extending the strip to execution-layer faults (the lane's 3.2) is an idea, not a rule, and the public text does not claim it. + +Evidence: `docs/analysis/horizon/frontier.md` sections 3.2, 4.1 and the incremental list (I13), 6 October 2026; spec 3.6 (the equivocation strip). + ## 3. Proving and the zkEVM ### P1. The 20-second shard is a number you made up @@ -502,6 +529,15 @@ Evidence: litepaper "Liquidity from the people who are there". --- +### E19. "Proving: a second income" without the arithmetic of how small it is +"You sell proving for other chains as the income that keeps cards on after the subsidy fades. Put a number on it. All of Ethereum L1's proving today costs tens of dollars a day. Your year-1 emission is tens of thousands a day at any price you dare print. Say which one pays the bills." + +Status: Conceded, stated (6 October 2026, evening; the Horizon lane analysis `docs/analysis/horizon/frontier.md` section 3.11, `frontier_model.py` section 7): `site/litepaper.html`, "For miners", under the three-streams table: all of Ethereum L1's proving is about USD 36 a day at the September 2026 tracker cost (USD 0.005 a block x 7,200 blocks; the tracker figure is a secondary source) against about USD 13,700 a day of Igneum's year-1 emission at USD 0.005 per IGN (31.688 IGN a block x 86,400; the price is an input, not a forecast), so external proving is a small second income at launch and the lottery pays the bills; paid demand would have to grow about 1,000x in dollars for proving to become the main income. Figures the lane labels approximate (all rollup proving spend, USD 8,200 to 27,400 a day; Boundless's trailing day, USD 2) are not on the page. + +Answer: Correct. The whole public proving market is three to four orders of magnitude under year-1 emission at any price input (the lane's table: Ethereum L1 at the Sep 2026 cost USD 36 a day, at the Dec 2025 cost 288; year-1 emission 13,700 at USD 0.005, 54,800 at 0.02, 273,800 at 0.10). The cost curve falls 3x to 30x a year, so dollars per proof fall as fast as volume rises. The design's own claim stays the defensible one: a second income that keeps cards on after the subsidy fades (spec 5.10.2), never the main one by 2030. + +Evidence: `docs/analysis/horizon/frontier.md` section 3.11 and the summary (section 7), 6 October 2026; the tracker (ethproofs, "sub-half-cent" fields, September 2026, secondary); the emission schedule (31.688 IGN a block in year 1). + ## 5. Governance and the founders ### G1. No cryptography team diff --git a/site/index.html b/site/index.html index ced727867..317d164a4 100644 --- a/site/index.html +++ b/site/index.html @@ -280,7 +280,7 @@
This hour's program

-        
A new program every hour, so last hour's chip is obsolete. First live swap 4 Oct 2026: Apple, NVIDIA and AMD kept hashing through it, 0 rejected blocks.

+        
A new program every hour: a chip wired for one program is useless. A programmable chip is met by the latency-shadow work and the price per joule, not by surprise. First live swap 4 Oct 2026: Apple, NVIDIA and AMD kept hashing through it, 0 rejected blocks.

       

       

         
Proofs sold to other chains

diff --git a/site/ledger.html b/site/ledger.html
index 3ac462e6f..7b35d0f30 100644
--- a/site/ledger.html
+++ b/site/ledger.html
@@ -4,13 +4,13 @@
 
 
 Igneum ledger: every criticism, answered
-
+
 
 
 
 
 
-
+
 
 
 
@@ -18,7 +18,7 @@
 
 
 
-
+
 
 
 
@@ -51,10 +51,21 @@
 
 
 
 
 
@@ -121,20 +134,20 @@ details{margin-top:8px;font-size:14px;color:var(--ash)}summary{cursor:pointer;co
 
 

 

-
Ledger · 167 entries · regenerated from the repository

+
Ledger · 171 entries · regenerated from the repository

 
Every criticism, answered or conceded

-
This is every criticism the project expects, in the critic's words, with what was done about it and the date. 167 entries since 3 October 2026. Entries are never deleted; a status that changes keeps its history on the line. Where the critic was right the entry says Conceded. Where nothing has been done it says Open and names what settles it. The founder mined through the GPU years. Ethereum's move to proof of stake in September 2022 ended that income and the miners' place in that chain. This is one person building, with AI systems doing the engineering, the coin he wanted to exist for miners: GPU-mined, the miners are the provers, no founder allocation, every cost stated. Help is welcome and a team is wanted: cryptographers, node engineers, miners who will test. This ledger is the application form: pick an open row and write to hello@igneum.network with its id.

+
This is every criticism the project expects, in the critic's words, with what was done about it and the date. 171 entries since 3 October 2026. Entries are never deleted; a status that changes keeps its history on the line. Where the critic was right the entry says Conceded. Where nothing has been done it says Open and names what settles it. The founder mined through the GPU years. Ethereum's move to proof of stake in September 2022 ended that income and the miners' place in that chain. This is one person building, with AI systems doing the engineering, the coin he wanted to exist for miners: GPU-mined, the miners are the provers, no founder allocation, every cost stated. Help is welcome and a team is wanted: cryptographers, node engineers, miners who will test. This ledger is the application form: pick an open row and write to hello@igneum.network with its id.

 

 

-
+
-
+

 
CountStatusMeaning

 
 
7Nothing has settled it yet. The entry names what will
53The critic is right. "Stated" means the public text says so; "not yet stated" means it does not yet
57The critic is right. "Stated" means the public text says so; "not yet stated" means it does not yet

 
54A code, spec or text change answers it, with the commit or the page named

 
27A consensus rule or a decision by the owner answers it, dated

 
13A measurement or a simulation exists and is named

 
13A design rule answers it; no measurement is possible yet
167Every entry. The sections: Mining and chips, Finality and attacks, Proving and the zkEVM, Economics and the coin, Governance and the founders, Comparisons, Legal and regulatory, Launch and operations, Builders
171Every entry. The sections: Mining and chips, Finality and attacks, Proving and the zkEVM, Economics and the coin, Governance and the founders, Comparisons, Legal and regulatory, Launch and operations, Builders

 

 

 
Mining and chips

@@ -216,6 +229,18 @@ details{margin-top:8px;font-size:14px;color:var(--ash)}summary{cursor:pointer;co
 
Conceded, stated 5 October 2026, night): a repository file, For miners, Hardware, "Macs mine too, at about a fifth of a flagship card: Measured, 26.7 against 123 million hashes a second" (confirmed by grep tonight; the projected-earnings half is not written because the app shows none, see M29). Was: Conceded, partly stated.

 
The answer as first written
The measured ratio is about 5x in the 5090's favour, so a Mac is a poor miner per dollar. The litepaper says Macs mine; it should say Macs mine at about a fifth of a flagship card and that the one-click app shows projected earnings before it starts.

 
+

+
M32
"Automatic anti-ASIC escalators" overstates what the era draw and the instruction reserve do
6 October 2026

+
You sell the era draw and the reserve unlock as anti-ASIC escalators, as if not knowing next era's parameters stops a chip. A chip that stores the dataset reads every drawn parameter as firmware: an address permute, a rotator, an immediate table. The families, the reserve order, the mixer, the dataset schedule and the class v4 shadow are all public at genesis. So what does the draw actually defend against?

+
Conceded, stated 6 October 2026, evening; the Horizon lane analysis a repository file sections 5.4 and 8, lane 2): the era draw and the instruction reserve are automatic schedule changes against fixed datapaths and against human forks; against the stored-dataset chip every drawn parameter is firmware, and the defence against that chip is the latency-shadow work (class v4) and the price-per-joule model. Stated in a repository file, Mining section ("These are automatic schedule changes ... every drawn parameter is firmware") and the "A chip is impossible" item ("a chip wired for one program is a bad bet ... not the schedule"), the "Every six months" row of the comparison table, and a repository file, the hourly-program note ("a chip wired for one program is useless"). The phrase "automatic anti-ASIC escalators" is withdrawn from public text; it stays in the internal design summary until that is next edited.

+
The answer as first written
Correct. The draw hides (M, R, pos, the op weights within +-2, the fold rotations) until 2 hours before each era, and none of those needs silicon. Biasing the draw is priced at 20 days of 100 percent of the network's hash for one more sample of the same space (lane section 5.4), so the draw is unbiasable at any price that matters and that is its whole job: it is a fairness device and a fork-free schedule, not a chip defence. What a chip wired for one program loses to is the hourly program itself; what the stored-dataset chip loses to is the latency shadow (class v4, 2.1x per joule at k = 1 against the 5090 bench row, 0.9x against the Apple M5 Max) and the price per joule, which is where the public claim now rests.

+

+

+
M33
The FPGA ceiling rests on a tFAW the JEDEC HBM2 table does not give
6 October 2026

+
Your chip model's HBM random-read ceiling takes 8 activates per 12 ns per channel from O'Connor and gets 10.7 G reads/s a stack; the epoch-length page's bank-bound row gets 11.4 and a 12.2 ceiling. JEDEC HBM2 tFAW is 28 ns with 4 activates per channel per window: 2.3 G. The one measured HBM2 FPGA random-read rate (Shuhai, FCCM 2020) is 2.4 G, right on the JEDEC ceiling. Your 1.9x FPGA ceiling is arithmetic on a timing the part does not have.

+
Conceded, stated 6 October 2026, evening; the Horizon lane analysis a repository file section 5.1, the FPGA lane): the public FPGA line carries only the measured row, 2.4 G reads/s per card and 0.30x to 0.39x of the RTX 5090 per watt (Shuhai, FCCM 2020 Fig 7; the tFAW arithmetic from ICCAD 2021 Table I), and the 11.4 G bank-bound row and the 12.2 G ceiling are marked unmeasured until an AWS F2 hour measures them. Stated in a repository file section 5.3 (the activate-bound row marked UNMEASURED with the JEDEC figure beside it, and the FPGA paragraph after the table). The epoch-length analysis's 12.2 row is not on master yet and is corrected when it lands.

+
The answer as first written
Correct. The measured 2.4 G/s had been read on 6 October as a mapping artefact ("the paper's point is that this mapping is the wrong one for random access"); the activate window says it is the DRAM's own limit, and a bank-interleaved mapping does not lift it because tFAW is enforced per channel by the die. The measurement that settles it is one AWS F2 hour (f2.6xlarge, Virtex UltraScale+ VU47P, 16 GB HBM2 in 2 stacks, 32 pseudo-channels, USD 1.98 an hour on demand): the chase kernel of a repository file 2.2 ported to a Vitis HLS AXI master over the HBM IP at 1 GiB across all 32 pseudo-channels, 256 to 4,096 lanes in flight, board power at 1 Hz; pass line 15 to 25 M reads/s/W (0.3x to 0.5x of the 5090), alarm 27 (0.5x), over 54 (1.0x) a Counter ASIC 4.0 item. Consequence per tier: none today (no FPGA mines); on the measured row a soft-overlay FPGA mines at an RX 9070 XT's rate per watt for about 7x the price (approximate), so no home or rig tier is displaced.

+

 
Finality and attacks

 

 
F1
Finality is attackable for the first month
5 October 2026

@@ -295,6 +320,12 @@ details{margin-top:8px;font-size:14px;color:var(--ash)}summary{cursor:pointer;co
 
Answered by design

 
The answer as first written
Full nodes execute natively so users see state in about a second. The proof is for everyone who is not a full node: light clients, bridges, exchanges syncing from a checkpoint, and the external market, which needs a standing prover population with hardware already running. It is also what lets a new node sync from a proven checkpoint instead of replaying history. The 20% is paid for capacity as much as for the proofs themselves, and the litepaper should say that.

 

+

+
F26
"No stake" needs its one sentence: what is at stake, and what strips it
6 October 2026

+
You write 'no stake' and then run a vote whose weight can be stripped. Either nothing is at stake, in which case equivocation costs nothing, or something is, in which case say what it is and who can take it. One sentence, in the finality section and the summary, not an argument.

+
Conceded, stated 6 October 2026, evening; the Horizon lane analysis a repository file section 4.1 and item I13 of its incremental list): a repository file, the finality section's "What is not here" paragraph and the "Igneum at a glance" Finality row carry the sentence verbatim: "No coin is staked. The only thing at stake is 30 days of public work: a vote key's weight is its blue blocks over the window, and equivocation strips it for 30 days." The spec sentence for 03 and 05 (I13) follows with the lane's commit.

+
The answer as first written
Correct. The weight is a quantity that is at stake, earned by work alone over 30 days, not transferable, not purchasable, and already stripped in full for equivocation (spec 3.6). That is a slashable bond made of blocks, with no coin and no stake class; the "then it is stake" objection and its answer (the weight cannot be bought, borrowed or bridged, so capture by capital is removed; the last-days attack is bounded by the 30-day re-earn) are in the lane file, section 4.1. Extending the strip to execution-layer faults (the lane's 3.2) is an idea, not a rule, and the public text does not claim it.

+

 
Proving and the zkEVM

 

 
P1
The 20-second shard is a number you made up
5 October 2026

@@ -405,6 +436,12 @@ details{margin-top:8px;font-size:14px;color:var(--ash)}summary{cursor:pointer;co
 
Conceded, stated

 
The answer as first written
True and stated in the litepaper. Mined coins are the only coins the founders can hold. The addresses are disclosed, so the seeding is visible. Whether to do it at all is a question for counsel (L1, L2).

 

+

+
E19
"Proving: a second income" without the arithmetic of how small it is
6 October 2026

+
You sell proving for other chains as the income that keeps cards on after the subsidy fades. Put a number on it. All of Ethereum L1's proving today costs tens of dollars a day. Your year-1 emission is tens of thousands a day at any price you dare print. Say which one pays the bills.

+
Conceded, stated 6 October 2026, evening; the Horizon lane analysis a repository file section 3.11, frontier_model.py section 7): a repository file, "For miners", under the three-streams table: all of Ethereum L1's proving is about USD 36 a day at the September 2026 tracker cost (USD 0.005 a block x 7,200 blocks; the tracker figure is a secondary source) against about USD 13,700 a day of Igneum's year-1 emission at USD 0.005 per IGN (31.688 IGN a block x 86,400; the price is an input, not a forecast), so external proving is a small second income at launch and the lottery pays the bills; paid demand would have to grow about 1,000x in dollars for proving to become the main income. Figures the lane labels approximate (all rollup proving spend, USD 8,200 to 27,400 a day; Boundless's trailing day, USD 2) are not on the page.

+
The answer as first written
Correct. The whole public proving market is three to four orders of magnitude under year-1 emission at any price input (the lane's table: Ethereum L1 at the Sep 2026 cost USD 36 a day, at the Dec 2025 cost 288; year-1 emission 13,700 at USD 0.005, 54,800 at 0.02, 273,800 at 0.10). The cost curve falls 3x to 30x a year, so dollars per proof fall as fast as volume rises. The design's own claim stays the defensible one: a second income that keeps cards on after the subsidy fades (spec 5.10.2), never the main one by 2030.

+

 
Governance and the founders

 

 
G1
No cryptography team
5 October 2026

diff --git a/site/litepaper.html b/site/litepaper.html
index fc6346bd4..4b67e57e2 100644
--- a/site/litepaper.html
+++ b/site/litepaper.html
@@ -274,7 +274,7 @@ body.all .pager{display:none}
               
               1. Mining lottery
               A random GPU program picks who makes the next block
-              New program every hour, so a chip for last hour's program is useless
+              New program every hour: a chip wired for one program is useless
             
             
               
@@ -318,7 +318,7 @@ body.all .pager{display:none}
             Mining lotteryA new program every hour on Apple, NVIDIA and AMD cards, compiled ahead, no pause and no rejected block at the boundary4 Oct 2026, first live swap
             BlocksAbout one a second with the full fleet; 0.65 a second over the hour to 16:00 UTC on 6 Oct 2026 with one PC off (the live page's hour count, 2,325 blocks)genesis, 3 Oct 2026
             DifficultyRule v2, a 600-second reference window, switched on by height under the running chain with no fork and no restart of the chainDAA 33,000, 4 Oct 2026
-            FinalityRule v2: a checkpoint every 30 s of chain, locked at two thirds of all 30-day weight. First live lock: checkpoint 242 at 77.4% of all weight, 17 vote keys4 Oct 2026
+            FinalityRule v2: a checkpoint every 30 s of chain, locked at two thirds of all 30-day weight. First live lock: checkpoint 242 at 77.4% of all weight, 17 vote keys. No coin is staked. The only thing at stake is 30 days of public work: a vote key's weight is its blue blocks over the window, and equivocation strips it for 30 days4 Oct 2026
             Provingv0 active: shards are assigned to miners' keys, proven on their cards, and the records are carried in blocksDAA 84,100, 5 Oct 2026
             EmberThe one-click miner on the fleet, version 0.3.13 (6 Oct 2026); the app window still says Igneum Miner4 Oct 2026, first install
             WalletIgneum Wallet 0.1.4 on macOS (0.1.1 first shipped 5 Oct 2026)6 Oct 2026
@@ -339,11 +339,11 @@ body.all .pager{display:none}
             Every hashThe 128 dataset addresses depend on the nonce, so every hash reads different memory. The one-bit select inside the maths costs a chip nothing and is not a defence; the random reads areNo
             Every hourA new random programNo, the miner compiles whatever arrives
             Every dayA new datasetNo
-            Every six monthsA new instruction mix and memory pattern drawn by the chain from rules fixed at genesis, and a new family of instructions unlocked from a reserve written at genesis, so the program space widens every eraNo
+            Every six monthsA new instruction mix and memory pattern drawn by the chain from rules fixed at genesis, and a new family of instructions unlocked from a reserve written at genesis, so the program space widens every era. A schedule change against fixed datapaths and human forks, not a surprise: a programmable chip reads every drawn parameter as firmwareNo
             ContinuouslyThe dataset grows on a schedule fixed at genesis, slowly enough that consumer cards keep up for years. A chip is built with fixed memory, so it is on a countdown from the day it ships. Ethereum's growing dataset ran Bitmain's E3 out of memory in 2020 this way, approximate, with nobody doing anythingNo
           
         

-        
Three ideas carry the chip resistance. The hash rewrites itself. A new program every hour, drawn from the chain. Its memory pattern changes with it. The rules change on a schedule fixed at launch. No release, no vote. It waits on memory, not maths. Every hash is a chain of random reads into a table too big for a chip to carry. The wait is the same physics for everyone. Miners hold the switch. Spare defences are written into the rules, switched off. A 90% miner signal turns one on. No fork.

+        
Three ideas carry the chip resistance. The hash rewrites itself. A new program every hour, drawn from the chain. Its memory pattern changes with it. The rules change on a schedule fixed at launch. No release, no vote. These are automatic schedule changes: they defeat a chip wired for one datapath and they need no human fork. Against a chip that stores the dataset every drawn parameter is firmware, and what meets that chip is the latency-shadow work (class v4) and the price per joule (the Horizon lane analysis, 6 October 2026, section 5.4; ledger M32). It waits on memory, not maths. Every hash is a chain of random reads into a table too big for a chip to carry. The wait is the same physics for everyone. Miners hold the switch. Spare defences are written into the rules, switched off. A 90% miner signal turns one on. No fork.

         
No hash has stayed free of chips forever. Igneum does not claim to. It states the gain its own model finds, the response takes a week, and both are measured. The model is public: the numbers; the claim is tested by paid independent cryptanalysis and the public benchmark. Monero has run on RandomX since 2019 with no chip publicly shipped, approximate; that is precedent, not proof.

         
One thing takes a person, here and on every chain that exists: writing new code. A chain cannot safely write its own generator, and it cannot safely tell a chip from a wave of honest new cards by hashrate alone. If the design above ever failed, anyone could publish a new generator and miners would switch it on by signalling, as Monero's community can fork. Igneum is built to make that day unlikely, and does not depend on avoiding it.

       
@@ -389,7 +389,7 @@ body.all .pager{display:none}
         
A miner's vote weight is simply the blocks it has mined over the trailing 30 days, measured by work, so splitting into many keys buys nothing and joining a pool costs nothing. Hashrate that arrived today holds almost none of it. Even an attacker producing every block on the chain, with honest miners gone, would need ten days of mining in public to hold a third of the weight, and twenty to hold two thirds. An attacker matching the honest network needs twenty days for a third and never reaches two thirds while the honest miners keep mining. Rental is priced by the hour. The only route left is to drive honest miners off the chain and hold two thirds for a month on the public hashrate charts, which is the same limit Bitcoin lives with, with a month's warning attached. Pools carry their hashers' votes, so vote concentration equals pool concentration, and it is public.

         
Two further rules close the gaps. A lock needs two thirds of all 30-day weight, so finality pauses whenever less than two thirds of that weight is connected and signing, until it returns or ages out of the window, up to 30 days, and the chain runs on proof of work meanwhile. The node reports the pause. A key that stops signing is reported as absent within two hours, which is how operators see a pause coming. Beneath the latest lock the depth to rely on is the finality depth: a node never switches to a chain forked more than 12 hours of median time back, and a certified checkpoint shortens that to its own age. Kaspa's one-hour merge depth is a limit on which old blocks a new block may merge, not a reorganisation bound. Signing two different checkpoints at the same height is equivocation, provable by anyone, and it strips the key of its vote for 30 days.

         
What is not here

-        
No stake. No coin-holder class votes on anything. No anchoring into Bitcoin or any other chain. Nothing in Igneum's consensus depends on anything outside Igneum.

+        
No coin is staked. The only thing at stake is 30 days of public work: a vote key's weight is its blue blocks over the window, and equivocation strips it for 30 days. No coin-holder class votes on anything. No anchoring into Bitcoin or any other chain. Nothing in Igneum's consensus depends on anything outside Igneum.

       
 
       

@@ -494,6 +494,7 @@ body.all .pager{display:none}
             External proving jobsRollups and apps on other chains, priced in their moneyNo, but the market is small today and is upside, not a promise
           
         
+        
The size of that third stream today, in numbers: all of Ethereum L1's proving is about USD 36 a day at the September 2026 tracker cost (USD 0.005 a block, 7,200 blocks a day; the tracker figure is a secondary source), against about USD 13,700 a day of Igneum's year-1 emission at USD 0.005 per IGN (31.688 IGN a block, 86,400 blocks a day; the price is an input, not a forecast). So external proving is a small second income at launch and the lottery pays the bills; for proving to become the main income the paid demand would have to grow about 1,000x in dollars (the Horizon lane analysis, 6 October 2026, section 3.11; ledger E19).

         
The honest bear-market case rests on cost. A miner's card is already running and the power is often domestic, so Igneum miners' marginal cost in the proving market is close to power, which is an edge over data-centre provers and nothing more. Which of the two in-chain streams pays more per GPU-second depends on the size of the fleet: on the devnet of 4 October 2026, three machines at 275 million hashes a second, a second of hashing paid about 4.9x a second of proving the pool share; at 10,000 cards the same arithmetic favours proving by about 930x. That is arithmetic on measured devnet rates, approximate, not a market measurement.

         
Hardware

         
The dataset starts at 2 GB and grows (the proposed schedule, fixed at the testnet genesis: 2 GB, doubling at years 4, 12 and 28, the average of half a gigabyte a year), so a 4 GB card mines for about four years and an 8 GB card for about twelve, approximate. Every NVIDIA card from 8 GB proves; 12 GB and up mine and prove; 24 GB on the stock server (eleven rented cards, RTX 3060 to RTX 5090, 6 October 2026). NVIDIA and AMD both work, because the mining program is generated for the architecture both share and the proof system is hash-based. Apple's chips are GPUs with unified memory, so Macs mine too, at about a fifth of a flagship card: Measured, 26.7 against 123 million hashes a second, an Apple M5 Max beside an RTX 5090 on the live devnet, 4 October 2026. A Mac is a poor miner per dollar. There is no CPU mining lane, on purpose, because CPU mining is what botnets farm. Nodes, wallets and exchanges need no GPU at all.

@@ -681,7 +682,7 @@ body.all .pager{display:none}
         
Here are the limits, stated before anyone else states them.