Bitcoin Forum
November 22, 2026, 12:59:29 AM *
News: Latest Bitcoin Core release: 32.1 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: [ANN][VARV] Varve (formerly BigCoin) - independent PoW chain. yespower CPU. No ICO  (Read 403 times)
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 01, 2026, 08:19:56 AM
Last edit: November 21, 2026, 11:35:31 PM by takamoto_bigcoin
 #1

Notice: the proof of work changed at block 19,500, on 14 November 2026

SHA-256d was the valid proof of work up to and including height 19,499. Every block mined before that and every BIG in a wallet stayed valid - the switch only changed what later blocks have to satisfy.

From height 19,500 the proof of work is yespower 1.0. It needs 2 MiB of memory per hash, so SHA-256 ASICs cannot mine it and an ordinary desktop CPU can. The chain crossed that height on 14 November 2026 at 09:55 UTC and did not split.

A node older than v1.7 will not follow the chain past that height. Upgrade to v1.9: http://133.242.139.204/downloads

I have no intention of changing the proof of work again, and any future consensus change gets at least 30 days between the announcement and the activation height.



Thanks for what was raised in this thread

Thank you to the people who have replied here, including the ones who told me I had got something wrong. The replies have changed what is in this post and what I work on, and that belongs near the top rather than thirty replies down.

- The 30-day commitment in the notice above came out of an objection that five days' notice of a proof-of-work change was too short for people who had pointed hardware at this chain since its first day. That objection was fair. Predictability was the only thing I had left to offer in return, and the commitment stands.
- A question about independent seed nodes, and about how an ordinary user can run a publicly reachable node, produced a page written for exactly that, plus a plan to ship volunteered nodes as built-in seeds in a later release: http://133.242.139.204/node
- Suggestions I did not adopt - a shorter target block interval, with the emission adjusted to match - got a reasoned answer in the thread rather than silence. I would rather explain a no than ignore the question.

If something here is wrong, say so. That has been more useful to this chain than agreement.



The rest of this post is the original announcement from 1 November 2026, kept current where it had gone out of date.

BigCoin is an independent proof-of-work blockchain, forked from Bitcoin Core 31.1, with its own genesis block mined on 21 Oct 2026. It shares no history with Bitcoin. It has been running continuously since launch and is past block 20,000 at the time of writing.

There is no ICO, no presale, no airdrop and nothing for sale. I am announcing it here because this is traditionally where one announces such things.

What it is

- A full fork of Bitcoin Core 31.1 ("BigCoin Core"), MIT licensed, full source published with every release at http://133.242.139.204/downloads
- A fully independent chain: own genesis, own network magic, own ports and address formats
- Bitcoin's economics unchanged: 50 BIG per block, halving every 210,000 blocks, 10-minute target
- Difficulty retargets every block (ASERT, two-day half-life) since height 17,325, rather than every 2016 blocks. On a chain this small that is the difference between tracking the hashrate and lagging a fortnight behind it.
- All Bitcoin softforks (SegWit, Taproot, ...) active from genesis

What it is not

- Nothing is for sale. There is no ICO, no presale, and I will never sell my own coins
- There are no promises. I make no claims about what BIG is worth or will ever be worth
- Nothing technically novel. If you are looking for innovation, this is not it. It is a small, clean, honestly run chain - that is the whole pitch.

Founder reward - full disclosure

Block 1 pays a founder reward of 1,000,000 BIG to my address. This is hard-coded in consensus, documented in the README, and displayed in a permanent banner on the block explorer. Including it, the supply cap is about 22,000,000 BIG. These coins are not for sale - my policy is the original one: publish the software, mine, maintain, and never sell.

Specifications

Code:
Ticker            BIG
PoW               SHA-256d to height 19,499, then yespower 1.0 (CPU)
yespower params   YESPOWER_1_0, N=2048, r=8, pers "BigCoin/yespower1.0"
Supply cap        ~22,000,000 BIG (21M mineable + 1M disclosed founder reward)
Block reward      50 BIG, halves every 210,000 blocks
Block interval    10 minutes target
Difficulty        ASERT every block since height 17,325, two-day half-life
                  (2016-block retarget before that height)
Launch difficulty CPU-minable (genesis nBits 0x1f00ffff)
P2P port          8338
Addresses         bech32 "big1..." / base58 starting with "B"
Genesis (21 Oct 2026)
  000086c5bc212db5bf8760b3fb3a30dfcd6af76257fd529814bc032c8134424d

Verifying your download

The binaries are not code-signed, so Windows SmartScreen will warn about them, and some antivirus products flag an unsigned program that mines on that basis alone. Check the SHA-256 instead of trusting either the download or the warning. These digests are published in this thread as well as on the downloads page on purpose: this thread is not hosted on my server, so if that server were ever compromised the attacker could rewrite the zip and the hash page together, but not this post.

v1.9, the current release:

Code:
BigCoin-win64-v1.9.zip          f3999cc9d6da8a032a4ba8d2023a098e18dfeabe90e0369693b451415c5fd5cb
bigcoin-linux-x64-v1.9.tar.gz   cf85ed4f4b158e3c8c292052025252182c8f80d037a8419f5ef6bea5540b5ba8
bigcoin-source-v1.9.zip         d287776d917d0d22c063ffbc6c6ed91bba3fe96b8c9f385a594a2de1caf790bb
bigcoin-source-v1.9-unix.tar.gz d44066c8bf5d70404b66d21d6689af4d99a13695b1bbf047816c47cc6b99c693
mine-bigcoin-v1.8.cmd           b4973f0c5c6a6f59c276f899a21a7453a60e35e3b916d222e55fa27647c4bf26
mine-bigcoin-v1.8.sh            67d55214e84a5571358c09d2c875f0b49d94a08307ac32e93e79cc56fec574dd

The two mining scripts are worth checking separately even though they also ship inside the archives: they are the files that decide which address a block pays to. The copies inside the archives are byte-identical, just named mine-bigcoin.cmd and mine-bigcoin.sh without the version number, so the digests above match those too.

Earlier releases, recorded here so the history stays verifiable. Nothing published has ever been replaced, and every one of these is still at its original URL with its original digest. Builds older than v1.7 cannot follow the chain past height 19,500; v1.7 and v1.8 can, but v1.9 is the one to run:

Code:
BigCoin-win64.zip (v1.2)        c1d57735ce5e13866543a88c06a2fd898794cfd5a02c0fd108a0f9cb31754dc0
BigCoin-win64-v1.3.zip          b007171b076c57a1c2ac257cdd34906e826099becf2ee0358e65a16847cc2ed9
BigCoin-win64-v1.4.zip          ed3efd7168c25e5e5b1f29a481f8eb39326bc6a29708c924f6e62199c688d306
BigCoin-win64-v1.5.zip          2077bf3e39117992443c3bc70700a8327b0280f790f3682b74fa08e62938d1cf
BigCoin-win64-v1.7.zip          501613b628252dfcac7315dd2256332b0388f19ef44970f0e3794d088c9444d8
bigcoin-linux-x64-v1.7.tar.gz   753399bf4d183ecc94e17404c475b25c5201bb2b7762fb98600332720691e0cb
bigcoin-source-v1.7.zip         954390fb3dd8a3dc6e1feaf0469294ae5ff4957ae8de5117edf01923ba936125
bigcoin-source-v1.7-unix.tar.gz cbae78fca9c0953f3381e6e6d27d7c8dce73fdbb38c1d8e60e780b7d7aa98926
BigCoin-win64-v1.8.zip          dcfebf2f476331f0d83de481379b45f4a523de7f545a456a107b01434642b972
bigcoin-linux-x64-v1.8.tar.gz   66f8b838370410119ed396d6f6cd44da9425bba7b85cedb0232fa4557743a882
bigcoin-source-v1.8.zip         67b32b9f494595400654b11f42eae79a222f35d8b1ebac145d5e817ffbceb472
bigcoin-source-v1.8-unix.tar.gz 422375eafe2e11a7daf29292b1681449599b93d874f4fbe9dc4ac90f692ed6d2

The v1.7-named copies of the two mining scripts are byte-identical to the v1.8-named ones above, so the two script digests in the v1.9 list cover those URLs too.

Windows: certutil -hashfile <file> SHA256 - Linux: sha256sum <file>

Running a node

Windows: download the release zip and run bitcoind.exe - the bundled bitcoin.conf already points at the seed node.
Linux: build from source exactly like Bitcoin Core (the build docs are inside the source archive), or use the static linux-x64 tarball.
Manual connection: addnode=133.242.139.204:8338
Making your node reachable to others, with the privacy trade-offs of each method spelled out: http://133.242.139.204/node

Mining

The chain launched CPU-minable. A miner with real hardware showed up within hours of the v1.0 release and difficulty rose at the maximum allowed rate for weeks, which is why the proof of work changed at block 19,500 - see the notice at the top of this post. Since the switch it is CPU territory again.

Standard solo mining works: getblocktemplate, or the bundled mine-bigcoin script, or a stock cpuminer-opt pointed at your own node. cpuminer-opt's generic yespower algorithm takes the parameters from the command line, so no patched miner is needed:

Code:
cpuminer -a yespower --param-n 2048 --param-r 8 --param-key "BigCoin/yespower1.0" -o http://127.0.0.1:8337 -u anyname -p your-password --coinbase-addr=<your address> -t 8

The mining guide thread has the full setup, a known-answer test vector so you can check your build before spending hours on it, and what a given hashrate should expect at the current difficulty. There is no pool; anyone is free to run one, but it will not be mine.

Known quirks, disclosed

Bitcoin's retarget arithmetic was never designed for a CPU-minable genesis target: the 256-bit multiplication overflows while the target is very large, which made the first retarget jump x21 instead of the expected x1.24. Every node computes the same result, so consensus was never at risk, and difficulty has since risen out of the affected range. v1.1 fixed a header-sync issue this caused, and also pins the hashes of blocks 1 / 1000 / 2000 (replacing the checkpoint system that upstream removed), so settled early history cannot be reorganized. The 2016-block retarget this arithmetic belonged to was replaced by ASERT at height 17,325. Details are in the release notes.

Why

I wanted a chain run the way the original was run at the start: release the code, mine, keep it alive, sell nothing, promise nothing. Whether that is worth anything is not for me to say.

Questions are welcome, including hostile ones. I will answer honestly.

Links

Renamed to Varve on 20 November 2026. The old name was already in use by an unrelated token on another chain, so searching for it did not find this project. Nothing about the chain changed: same genesis block, same network, same port, same addresses, same unit, same binaries, same digests, no fork and nothing to upgrade. Published file names keep the old spelling because their digests are printed in this thread. All of this is also served over HTTPS at https://varve.mine.bz/ - same server, same files, same digests. The IP URLs below are the canonical ones and are not going away.

Code:
Block explorer     http://133.242.139.204/
Downloads          http://133.242.139.204/downloads
Mining status      http://133.242.139.204/mining/status
Faucet             http://133.242.139.204/faucet
Reachable node     http://133.242.139.204/node
Source             git clone http://133.242.139.204/git/bigcoin.git
Seed node          133.242.139.204:8338
Seed node (Tor)    rjxywola2kupk4rp67qzidqiuxwvr76xmri5zimt6ecrvhx4xv5t7wyd.onion:8338
ANN thread         https://bitcointalk.org/index.php?topic=5598514
Mining guide       https://bitcointalk.org/index.php?topic=5599204

These, plus this thread, are the only official channels. There is no Telegram, no Discord, and no token on any other chain. Anything else using the BigCoin name is not me.

- Masashi Takamoto (note: the bitcointalk user "bigcoin77" is not me)
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 01, 2026, 11:06:45 AM
Last edit: November 02, 2026, 12:22:13 PM by Welsh
 #2

v1.2 released - GUI wallet

BigCoin Core v1.2 adds the standard Qt GUI wallet (bigcoin-qt.exe) to the Windows release, so the command line is no longer required to receive, hold and send BIG.

- Download: https://github.com/bigcoin77/bigcoin/releases
- BigCoin-win64.zip SHA256: c1d57735ce5e13866543a88c06a2fd898794cfd5a02c0fd108a0f9cb31754dc0
- No consensus changes. v1.0 / v1.1 nodes remain fully compatible; upgrading is optional.
- First launch asks for a data directory - copy the bundled bitcoin.conf there and it connects to the seed node automatically.

Network status at the time of writing: past block 12,000, two independent miners, and difficulty still rising at the maximum allowed rate every period.

- Masashi Takamoto

Weekly status update — 2026-11-02 (all facts verifiable on-chain)

Network
- Block height: 17,321
- Difficulty: 5.27 — the first upward retarget triggered at block 16,128
  (1.32 -> 5.27, the x4 per-period limit). The adjustment worked as designed.
- Next retarget: block 18,144
- Total supply: 1,866,000 BIG of 21,000,000 (~8.9%)
- Mempool: 0

This week
- An independent miner joined and at peak pushed network hashrate to roughly
  1.5 GH/s, mining several thousand blocks before going quiet. Two distinct
  coinbase addresses appear in the last 400 blocks.
- Independent nodes from several countries connected to the seed and stayed
  in sync.
- Official site is now live: https://bigcoin77.github.io/bigcoin/
- Latest release remains v1.2:
  https://github.com/bigcoin77/bigcoin/releases

Ground rules, unchanged: nothing is for sale, no exchange listings are
sought, no price talk. The founder mines openly; every founder block is
visible on-chain. If you want to participate, run a node or point SHA-256d
hardware at the chain.

Explorer: http://133.242.139.204/
Seed node: 133.242.139.204:8338
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 07, 2026, 09:17:56 AM
Last edit: November 07, 2026, 12:27:20 PM by takamoto_bigcoin
 #3

v1.4 released — required upgrade: ASERT difficulty hard fork at height 17,325 (now live)

What happened
- The independent miner that drove difficulty up to 5.27 in early November went quiet. With the original 2016-block retarget, the remaining hashrate faced roughly 21x more difficulty than it could sustain, and the chain stalled at height 17,324 for about five days.
- This is the known weakness of Bitcoin's retarget rule on a small chain, so BigCoin now uses ASERT (aserti3-2d, the per-block difficulty algorithm Bitcoin Cash has run since 2020) with a 2-day half-life.

The fork
- v1.3 (2026-11-02) shipped ASERT scheduled for height 18,000. A stalled chain would never have reached it, so v1.4 moves activation forward to height 17,325 — the first block after the stall point. No existing blocks are reorganized.
- The fork is live and verified on-chain: block 17,325 was mined on 2026-11-07 at ASERT-computed difficulty 5.217, and block 17,326 followed at difficulty 0.881. Difficulty now adjusts every block and converges back to the 10-minute target on its own — if hashrate leaves again, the chain recovers automatically instead of freezing.

Action needed
- Nodes running v1.3 or older do not follow the chain from height 17,325. If you run a node or miner, please upgrade to v1.4. (A couple of older nodes are connected to the seed right now — this post is mainly for you.)

Downloads
- GitHub is temporarily unavailable (account review in progress, support ticket open). Until it is restored, binaries and full source are served from the explorer mirror:
- http://133.242.139.204/downloads
- BigCoin-win64-v1.4.zip — SHA256: ed3efd7168c25e5e5b1f29a481f8eb39326bc6a29708c924f6e62199c688d306
- bigcoin-source-v1.4.zip (full source, tag v1.4) — SHA256: c7680392269ea09060caa01aa871535c7d8c2559c3d0c60a873573ef5da08d44

Unchanged: 50 BIG per block, halving every 210,000 blocks, ~22M cap including the disclosed 1M founder reward, explorer http://133.242.139.204/ , seed node 133.242.139.204:8338. As always: nothing is for sale, no exchange listings are sought, no price talk.

- Masashi Takamoto


v1.5 released — maintenance only, fully compatible with v1.4

No consensus changes, and nothing in the ASERT fork above is affected. v1.4 nodes do not have to upgrade. Two practical fixes:

- The solo-mining helper script (mine-bigcoin.cmd) is now bundled inside the Windows zip. Put your own big1 address in it, double-click, and it creates the data directory, starts the node, creates a wallet and starts mining — no command line needed.
- The shipped binaries no longer carry the build machine's local directory paths. The compiler embeds absolute source and debug paths into every executable; those are now stripped. Nothing about the code itself changed.

Downloads (GitHub is still under account review, so the explorer mirror serves them):
- http://133.242.139.204/downloads
- BigCoin-win64-v1.5.zip — SHA256: 2077bf3e39117992443c3bc70700a8327b0280f790f3682b74fa08e62938d1cf
- bigcoin-source-v1.5.zip (full source, tag v1.5) — SHA256: 5cd3148e75433cdad49ed7a669383511a2635f29287356e0d6c091d0b9303ae6

The binaries are unsigned, so some antivirus products sandbox or flag them on first run — this is not new to v1.5. Check the SHA256 above, or build from the source zip if you would rather not trust a binary.

Network status at the time of writing: height 17,403, difficulty 1.01 after ASERT converged, blocks arriving on schedule again, and three distinct coinbase addresses in the last 300 blocks — including an independent miner that started right after the fork.

- Masashi Takamoto
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 09, 2026, 08:24:08 AM
 #4

Weekly status update - 2026-11-09 (all facts verifiable on-chain)

Network
- Block height: 18,134
- Difficulty: 3.13, recalculated every block
- Total supply: 1,906,650 BIG of 21,000,000 (~9.1%)
- Block spacing: 5.1 min over the last 144 blocks, 4.0 min over the last 500
- Mempool: 0

v1.7 is published today, and it changes the proof-of-work algorithm
Up to height 19,499 BigCoin uses double SHA-256, exactly like Bitcoin. From
height 19,500 it uses yespower 1.0 instead: N=2048, r=8, personalisation
string "BigCoin/yespower1.0", computed over the same 80-byte block header.
Every hash needs 2 MiB of memory.

Why. This chain's difficulty is about 3.1, which is roughly 44 MH/s of
SHA-256. A hobby single-chip miner, the sort sold as a desk ornament at about
1.2 TH/s, is some 27,000 times that; a second-hand datacentre unit runs 13 to
100+ TH/s, which is one to two further orders of magnitude on top. Either
owner could take every block for as long as they cared to, or rewrite recent
history, for the cost of the electricity - and no difficulty rule fixes that
on a chain this small, because difficulty cannot tell one miner from another.
yespower needs memory that a SHA-256 ASIC does not have, so that hardware
cannot participate at all: a block it produces is invalid, not merely late.

What this is not. It is not a claim that BigCoin is ASIC-proof for ever.
Someone can build hardware for yespower. It would have to carry the memory,
which is most of what makes a general-purpose CPU expensive, so the gap would
be a small multiple rather than the four-to-six orders of magnitude a SHA-256
ASIC enjoys here today. That is the whole claim, and I would rather state it
that narrowly than sell you a guarantee I cannot keep.

What it does not fix either. Heights below 19,500 stay double-SHA256 for
ever, so a fork rooted in that range can always be mined at ASIC speed. The
highest pinned block in v1.7 is 17,750, which caps how far back such a fork
can be rooted; a fork from there would still have to out-work the honest
chain from 17,750 on and would arrive as a reorganisation up to 1,750 blocks
deep by the time the switch happens, which no node applies quietly. That is a
bounded residual risk, not a closed hole, and the only thing that shrinks it
further is a later release pinning into the window.

The difficulty at the switch is set by hand, once
The old nBits measures double-SHA256 work and means nothing under yespower,
so the ASERT schedule restarts from a hand-set target at 19,500: 6,000,031
hashes per block, i.e. one block per ten minutes at 10,000 H/s. For scale, the
16-thread desktop this release was built on works out at roughly 5,000 H/s
across all of its threads, so that target assumes something like two such
machines are mining. Read that as an estimate of my hardware rather than a
benchmark of yours: yespower is bound by memory bandwidth, not clock speed, so
the spread between CPUs is wider than their specifications suggest.

That is deliberately on the hard side, and the asymmetry is the reason:
setting it too easy would mint coins quickly and cannot be undone, because
ASERT can only tighten by 0.24% per block. Setting it too high costs slow
blocks that fix themselves. So expect the first blocks after 19,500 to come
slowly - possibly for a day or two - while difficulty finds the real CPU
hashrate. It eases every block, and faster than that if a gap passes six
hours, which is the other change in this release.

Also in v1.7: the stall rule, from height 18,000
ASERT as shipped in v1.4 fixes the next block's target the moment its parent
exists - the elapsed time it measures runs to the PARENT'S timestamp. So
while the chain sits waiting, nothing eases. A difficulty left behind by a
miner who stopped does not decay; it stays where it was until somebody finds
one block at it. That is why height 17,324 froze for 5 days 3 hours at the
start of this month rather than slowly recovering, and v1.4 only cleared it
because I hand-moved the ASERT anchor. The next time it happens I might be
asleep.

From height 18,000 the elapsed time is max(parent's timestamp, candidate's
timestamp - 6 hours). Below six hours of silence that is byte-identical to
the old rule. Past six hours the target eases with real time on its own,
halving every two days, so a stranded difficulty clears itself. A three-day
visit from an ASIC followed by it unplugging used to leave a difficulty whose
expected wait for the next block was about 140 days; it now recovers in about
11 days with no intervention from anyone, including me.

What it costs you:
- Block timestamps are already capped at 2 hours ahead of real time, so the
  most anyone can gain by dating a block forward is 2^(2h / 2 days) = 2.9%
  less work. It does not compound, because the ASERT anchor is fixed, and it
  is a level shift available equally to everybody rather than a per-block
  subsidy to whoever lies hardest.
- Mining software must stop rolling the timestamp while keeping the old
  nBits, because from 18,000 the required nBits depends on that timestamp.
  getblocktemplate no longer advertises "time" as mutable at that height. Any
  miner built on generatetoaddress, including the scripts I publish, needs no
  change: it asks for a fresh template each round, which is also how it picks
  up the easing during a stall.

Height 18,000 is behind the tip by the time you read this, which is on
purpose: for every gap shorter than six hours the new rule computes exactly
the same nBits as the old one, so v1.7 validates the chain as it already
stands, block for block. I check that before publishing rather than assuming
it - if any gap at or above 18,000 had reached six hours, this release would
have been held back instead of shipped.

Both mining scripts are new, and worth taking
Every version before v1.7 waits for the node to report that its initial block
download has finished. A node started while the chain itself is stalled never
reports that, so the one moment the chain most needs another miner was the one
moment the old script refused to mine. That is fixed.

They also ask the node which algorithm the next block needs instead of
assuming one, and size each mining call by timing the previous one - which
they have to, because a yespower hash takes milliseconds where a double-SHA256
hash takes a fraction of a microsecond, and no fixed setting suits both sides
of 19,500.

One consequence is worth knowing before you judge the switch on per-hash costs
alone. These scripts drive the node's own generatetoaddress, which rebuilds and
re-serialises the block for every nonce it tries. That overhead is trivial
beside a yespower hash and dominant beside a double-SHA256 one, so today the
scripts reach well under what your CPU could do in a tight SHA-256 loop, and
after 19,500 the same code runs at close to the hardware's full yespower rate.
A CPU miner using what I publish therefore gains more at the switch than the
two algorithms' raw costs by themselves imply.

Unlike last week, the scripts are NOT byte-identical to the previous ones.
Replace them. The standalone downloads are versioned from now on
(mine-bigcoin-v1.7.cmd, mine-bigcoin-v1.7.sh) so the older files keep
working for anyone who linked to them; inside the packages the names are
unchanged.

What you have to do
Upgrade to v1.7 before block 19,500. At the pace of the last 144 blocks that
is about five days out, on the order of 14 November, but that estimate moves
with hashrate, so watch the height and not the date:

  bigcoin-cli getblockcount

Nodes on v1.5 or older stop at 19,500. Because it is the algorithm that
changes and not a difficulty rule, they do not merely fall behind: the hashes
they compute are no longer the ones consensus asks for, so they can neither
follow the chain nor mine one valid block past that height. The same is true
of any external miner or stratum setup speaking SHA-256d.

Upgrading is a binary swap and nothing else. The chain data on disk does not
change, so there is no reindex and no resync: stop the node, replace the
binaries, start it again. That holds even if you only notice after 19,500. A
node that rejects those blocks rejects them at the header, before anything is
written to its block index, so it has no bad verdict stored to undo - it
simply carries on from where it stopped once the binary understands what it is
being sent. You can read that for yourself in src/validation.cpp,
AcceptBlockHeader().

There is no v1.6 download, and the gap in the numbering is not an oversight.
v1.6 was built with the stall rule but without yespower; publishing it would
have handed you one more binary that dies at 19,500, so it was replaced by
v1.7 before release and never distributed. The published versions are v1.3,
v1.4, v1.5 and v1.7.

Your node will tell you where the switch is and which algorithm it expects
next, so you never have to take my word for it:

  bigcoin-cli getblockchaininfo     ->  pow_algo, yespower_activation_height

The seed node runs v1.7. The source is published with the binaries, so you can
read the change instead of trusting the description: src/pow.cpp,
src/crypto/yespower/ (vendored 1.0.1, with its provenance and local patches in
that directory's README.md) and src/kernel/chainparams.cpp.

Mining distribution
Coinbase addresses over the last 200 blocks, 6 distinct:

  179  89.5%  one address        not mine
   15   7.5%  one address        mine
    6   3.0%  four addresses     none mine

Over the last 500 blocks, 11 distinct addresses: the largest holds 86.6% and I
hold 8.6%. One address therefore accounts for roughly six blocks in seven,
which is enough to reorganise the chain at will. It is not me, and me saying so
does not make it safe. If you are considering mining here, that is the single
most important number on this page, and it is the reason independent hashrate
matters more than anything else I can ship.

I publish no addresses but my own; every figure above is reproducible from the
explorer in a few minutes. From 19,500 every miner needs yespower, whatever
hardware they are using today. I do not know what that will do to this table
and I am not going to pretend otherwise.

My own holdings, unchanged and still disclosed
My wallet holds 1,008,001 BIG, which is 52.9% of everything in existence,
across two addresses:
  big1qe2v5u9aqhafy85pxet8cpzzxyqgxxykqykj5gh  101 blocks, heights 1-101
  big1qycz9vmq49wuxkvl6v2tdnudglnfaxy7jd84xc3   80 blocks, heights 102-18,132
That total includes 200 BIG of rewards I cannot spend yet, because a coinbase
output matures 100 blocks after it is mined. I count them anyway: the supply
figure at the top of this post counts unmatured coinbases too, and the honest
way to state your own share is the one that does not quietly leave part of it
out of the numerator.
The first address includes the block 1 reward of 1,000,000 BIG, which was
disclosed from day one and is visible to anyone:
http://133.242.139.204/block/1
I have mined 1,009,000 BIG in total. The 999 BIG difference is the faucet:
1,000 BIG out to fund it, and 1 BIG back in, which is the single test claim I
made yesterday to confirm it actually pays. Nobody else has used it yet.
Nothing has been sold. Nothing is for sale. There is no exchange, no listing
application, and no mechanism by which I take money from anyone.

New this week - faucet
The explorer now has a faucet: http://133.242.139.204/faucet pays 1 BIG per
address, once per 24 hours, up to 50 BIG a day across everybody, funded with
the 1,000 BIG noted above - so you can see a transaction land in your own
wallet before you commit a CPU. If it tells you the daily budget is spent,
that is what has happened; it refills the next day.

v1.7 downloads
  BigCoin-win64-v1.7.zip           26.1 MB
  sha256 501613b628252dfcac7315dd2256332b0388f19ef44970f0e3794d088c9444d8
  bigcoin-linux-x64-v1.7.tar.gz    9.6 MB
  sha256 753399bf4d183ecc94e17404c475b25c5201bb2b7762fb98600332720691e0cb
  bigcoin-source-v1.7.zip          15.4 MB
  sha256 954390fb3dd8a3dc6e1feaf0469294ae5ff4957ae8de5117edf01923ba936125
  bigcoin-source-v1.7-unix.tar.gz  14.0 MB
  sha256 cbae78fca9c0953f3381e6e6d27d7c8dce73fdbb38c1d8e60e780b7d7aa98926
  mine-bigcoin-v1.7.cmd            21.6 KB
  sha256 b4973f0c5c6a6f59c276f899a21a7453a60e35e3b916d222e55fa27647c4bf26
  mine-bigcoin-v1.7.sh             17.4 KB
  sha256 67d55214e84a5571358c09d2c875f0b49d94a08307ac32e93e79cc56fec574dd

Ground rules, unchanged: nothing is for sale, no exchange listings are
sought, no price talk. If you want BIG, mine it.

Downloads:    http://133.242.139.204/downloads
Mining guide: http://133.242.139.204/mining
Explorer:     http://133.242.139.204/
Faucet:       http://133.242.139.204/faucet
Seed node:    133.242.139.204:8338
solarsurfer
Jr. Member
*
Offline

Activity: 97
Merit: 2


View Profile WWW
November 09, 2026, 08:57:30 AM
 #5




BigCoin (BIG) – no longer mined at Shears Mining

From block 19,500 (expected around 14 November), BigCoin changes its proof-of-work from SHA-256 to yespower, a CPU-only algorithm. After that,
blocks found by SHA-256 hardware – Bitaxe, NerdMiner, ESP32 and ASIC miners – are no longer valid on the BigCoin chain.

As a SHA-256 pool, we have closed our BigCoin port (3790) and removed BigCoin from the pool, effective today.

Thank you to everyone who mined BigCoin with us – SHA-256 miners secured the chain through its first 18,000 blocks. Any BIG you mined is safe in your own
wallet.

Looking for another coin your ESP32 or Bitaxe can find real blocks on?
See our low-difficulty coin list: https://sha256-mining.shears.co.uk/coins.php


I would like to state however, we like supporting coins that start up, by mining ourselves, and setting up a mining system that also allows other people to mine by SHA256 (Bitaxes esp32 etc). If dev teams find our help to promote a coin useful, then I find it alarming that coin dev teams take away the help we provided to them during launch day of their coin, by removing SHA256 from mining, and not only that, but an early stage in the coins development. It also upsets people who mined this coin ! DEV TEAMS NOTE  !

Just my thoughts that doesn't need a reply !

takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 09, 2026, 09:58:48 AM
Last edit: November 12, 2026, 06:43:14 PM by Welsh
 #6

Re: the notice above

Fair criticism, and I am not going to argue with it.

First, the facts, for anyone reading this thread who mines with SHA-256 hardware:

SHA-256d is still the valid proof of work right now. The switch happens at block 19,500, expected around 14 November. Until that height, SHA-256 hardware mines BigCoin exactly as it always has. Every block already found with it stays valid for ever, and every BIG already mined stays spendable for ever. The switch changes which algorithm future blocks have to satisfy. It does not touch the history, the coins, or anyone's wallet.

What I got wrong is the notice. I announced a proof-of-work change five days before it activates, twenty-four days after this chain launched, to miners who had pointed hardware at it since the first day. Five days is too short for a change of that kind, and the people who supported the launch are the ones who absorbed the cost. That was my decision and nobody else's, and I am not going to dress it up as anything other than a mistake in how I handled it.

Why I made the change anyway, stated plainly: a chain this small cannot survive one SHA-256 participant deciding to reorganise it, and the window where that is cheap to do does not close on its own as the chain grows - it closes when the proof of work stops being something rented by the terahash. I would rather take the credibility hit now than explain a deep reorganisation later. That reasoning does not give anyone their setup time back, and I am not pretending it does.

Since the only thing I can usefully offer is predictability, two commitments:

- I have no intention of changing the proof of work again. The switch at 19,500 is meant to be the last one.
- Any future consensus change gets at least 30 days between the announcement and the activation height, with that height published in the release notes and in this thread before it is reached. No more five-day notices.

For miners who want to keep going after 19,500: the chain becomes CPU-mineable again. An ordinary desktop CPU finds blocks on it - no special hardware, no pool, and no permission from me required. The v1.7 release also adds a rule that eases difficulty automatically if no block arrives for six hours, so the chain recovers by itself instead of freezing the way it did for five days earlier this month.

Thank you for carrying the first 18,000 blocks. That was real work, done before there was any reason to bother, and it is why this chain is still running.

- Masashi Takamoto
Correction: block 19,500 is about a day away, not four

Earlier today I said the switch at block 19,500 was expected around
14 November. That is no longer accurate. Since I am asking people to
upgrade before that height, I would rather correct it immediately than
let it stand.

The network hashrate has risen about sevenfold in the last nine hours.
Blocks that were arriving every five or six minutes are now arriving
roughly every ninety seconds, and difficulty is still climbing to catch up.

Measured from the chain itself:

Code:
height now                18,629
blocks until 19,500          871
recent block interval      ~1.5 min   (the target is 10)
difficulty                  8.35     rising every block

Turning that into a date is unreliable, because difficulty rises with every
block and the interval lengthens as it does. 871 blocks at the current rate
is about a day; as difficulty catches up it will stretch beyond that. My
honest estimate is late on 10 November or during 11 November UTC, and it
would be dishonest to make it sound more precise than that.

So watch the height, not the date:

Code:
bitcoin-cli getblockcount

What you need to do

If you run v1.7, nothing. It already knows about the switch, and the
bundled mining script asks the node which algorithm the next block needs,
so it changes over on its own.

If you run v1.5 or older, upgrade now. Those builds do not contain the new
algorithm at all. They will reject the first block above 19,499 as invalid
proof of work and stop there, on a chain that no longer grows, without
reporting anything as wrong. v1.6 was never released, so v1.5 is the newest
affected build.

Code:
http://133.242.139.204/downloads

501613b628252dfcac7315dd2256332b0388f19ef44970f0e3794d088c9444d8  BigCoin-win64-v1.7.zip
753399bf4d183ecc94e17404c475b25c5201bb2b7762fb98600332720691e0cb  bigcoin-linux-x64-v1.7.tar.gz
954390fb3dd8a3dc6e1feaf0469294ae5ff4957ae8de5117edf01923ba936125  bigcoin-source-v1.7.zip
cbae78fca9c0953f3381e6e6d27d7c8dce73fdbb38c1d8e60e780b7d7aa98926  bigcoin-source-v1.7-unix.tar.gz
b4973f0c5c6a6f59c276f899a21a7453a60e35e3b916d222e55fa27647c4bf26  mine-bigcoin-v1.7.cmd
67d55214e84a5571358c09d2c875f0b49d94a08307ac32e93e79cc56fec574dd  mine-bigcoin-v1.7.sh

Those digests are also in the first post of this thread and in the mining
guide thread. Neither is hosted on my server, so if that server is ever
tampered with, the copies here will disagree with it.

This is not a new consensus change

The activation height has not moved. It is still 19,500, the same height
published in the first post of this thread, in the mining guide and in the
v1.7 release notes. Nothing about the rules has changed; only my estimate
of when the chain reaches that height. The commitment I made earlier today
still holds: any future consensus change gets at least 30 days between the
announcement and the activation height.

What to expect at the switch

Difficulty does not carry across. The SHA-256d difficulty the chain has now
would be unmineable with CPUs, so the new algorithm starts from a fixed
value chosen for it. Expect the blocks right after 19,500 to be irregular,
and possibly slower than ten minutes, while difficulty finds its level.
v1.7 also eases difficulty automatically if no block arrives for six hours,
so a stall resolves itself rather than freezing the chain the way it did
for five days earlier this month.

The history and the coins are untouched. Every block mined with SHA-256d
stays valid for ever, and every BIG already mined stays spendable.

- Masashi Takamoto
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 12, 2026, 12:52:25 PM
 #7

v1.8 released: the Windows package is no longer half in Japanese

v1.8 is a documentation release. Every program in it is byte-for-byte the
one in v1.7 and still reports v31.1.7. No consensus rule, no network rule
and no default has changed, and nothing was rebuilt. If you already run
v1.7, there is nothing here you need to install.

What was wrong

The Windows archive shipped README.txt and bitcoin.conf in Japanese only.
Those are exactly the two files a new user cannot skip: no seed nodes are
compiled into the binaries, so a wallet that never reads bitcoin.conf never
finds the network at all. The Linux archive has shipped an English
README.txt with README.ja.txt beside it since v1.5. Windows was left behind,
and I did not notice until I went looking for why visitors were not coming
back.

v1.8 fixes that. The Windows archive now carries English README.txt and
bitcoin.conf, with README.ja.txt and bitcoin.conf.ja alongside. The source
tree gains doc/BigCoin.ja.md, so the English and Japanese specifications
sit side by side instead of one of them being the only copy, and a new
VERIFYING.md walks through checking an archive against a digest and against
a git tag.

Do not take "identical programs" on trust

Unpack v1.7 and v1.8 side by side and compare. All 19 programs in the
Windows archive, and every binary in the Linux tarball, have the same
digests:

Code:
Windows:  certutil -hashfile bitcoind.exe SHA256
Linux:    sha256sum bigcoin/bitcoind

The only files that differ between the two archives are README.txt and
bitcoin.conf, and the only files added are README.ja.txt and
bitcoin.conf.ja.

Digests

Code:
http://133.242.139.204/downloads

dcfebf2f476331f0d83de481379b45f4a523de7f545a456a107b01434642b972  BigCoin-win64-v1.8.zip
66f8b838370410119ed396d6f6cd44da9425bba7b85cedb0232fa4557743a882  bigcoin-linux-x64-v1.8.tar.gz
67b32b9f494595400654b11f42eae79a222f35d8b1ebac145d5e817ffbceb472  bigcoin-source-v1.8.zip
422375eafe2e11a7daf29292b1681449599b93d874f4fbe9dc4ac90f692ed6d2  bigcoin-source-v1.8-unix.tar.gz
b4973f0c5c6a6f59c276f899a21a7453a60e35e3b916d222e55fa27647c4bf26  mine-bigcoin-v1.8.cmd
67d55214e84a5571358c09d2c875f0b49d94a08307ac32e93e79cc56fec574dd  mine-bigcoin-v1.8.sh

The two miner scripts did not change either. They carry v1.8 names so that
the quick start does not ask you to download a v1.8 archive and a v1.7
script in the same two steps, but they are byte-for-byte copies, which is
why their digests are the same as the v1.7 ones published earlier in this
thread. The v1.7 names stay published and keep working.

Every v1.7 file is still at the same URL with the same digest. Nothing
published has been replaced.

The source is also a git repository

Code:
git clone http://133.242.139.204/git/bigcoin.git
git remote add upstream https://github.com/bitcoin/bitcoin.git
git fetch --depth=1 upstream tag v31.1
git diff --ignore-cr-at-eol --stat v31.1 v1.8

That is the whole of what BigCoin changes in Bitcoin Core v31.1: 52 files,
two more than v1.7 because of VERIFYING.md and doc/BigCoin.ja.md. Tag v1.8
holds the published source archive file for file. One caveat, since v1.8 is
the release that invites you to check: the archives are packed on Windows,
so contrib/bigcoin/mine-bigcoin.cmd has CRLF line endings there and LF in
git. Nothing else differs, and VERIFYING.md gives the one extra flag that
accounts for it. A clone proves nothing by itself, though. The digests in
this thread are the anchor, because they are not hosted on my server.

One number in the bundled READMEs was wrong

The passage explaining the upcoming algorithm change said this chain's
difficulty was around 2.3, roughly 40 MH/s. Measured on 12 November it was
12.3, roughly 88 MH/s. The point it was making is unchanged -- a single
second-hand ASIC still does tens of TH/s, more than a hundred thousand
times the whole chain -- but the figure now matches what a node reports.
The correction is written into both changelogs rather than applied quietly.

Block 19,500 has not moved

Code:
height now                19,159
blocks until 19,500           341
difficulty                 12.18

At 19,500 the proof of work changes from SHA-256d to yespower 1.0
(N=2048, r=8, personalisation BigCoin/yespower1.0). Every hash needs 2 MiB
of memory, which no SHA-256 ASIC has. The height has not changed and this
release does not change it; v1.8 contains the same consensus code as v1.7.

If you run v1.7 or v1.8, you are ready and need do nothing. If you run v1.5
or older, upgrade before that height. Those builds do not contain the new
algorithm at all: they will reject the first block above 19,499 as invalid
proof of work and stop there, on a chain that no longer grows, without
reporting anything as wrong. v1.6 was never released, so v1.5 is the newest
affected build.

Watch the height, not the date:

Code:
bitcoin-cli getblockcount

The commitment from earlier in this thread still holds: any future
consensus change gets at least 30 days between the announcement and the
activation height. This release is not one -- it does not touch src/ at
all.

- Masashi Takamoto
tadashy
Newbie
*
Offline

Activity: 4
Merit: 0


View Profile
November 13, 2026, 03:04:03 PM
 #8

So, my proposal for BigCoin would be one block every 2 minutes, following the activation of Yespower.
2 minutes
Proposed target interval
Feature   Result
Blocks per hour   30
Blocks per day   720
Blocks per year   262,800
Compared to 10 minutes   5 times more blocks


However, there are two significant consequences:
1. If the reward remains at 50 BIG per block, the daily issuance would increase from 7,200 to 36,000 BIG.
2. Block propagation and orphaned blocks would need to be reviewed to avoid stability issues.
My proposal: 2 minutes per block, while adjusting the reward to control monetary issuance.
EL MAOGO
Newbie
*
Offline

Activity: 96
Merit: 0


View Profile
November 13, 2026, 03:29:37 PM
 #9

Hi developer,

I have a suggestion regarding the Yespower activation at block 19,500.

Have you considered reducing the target block interval from 10 minutes to 2 minutes?

A 2-minute block interval would make BigCoin more responsive, provide faster transaction confirmations, and give solo miners more frequent opportunities to find blocks.

Of course, I understand that this change would require reviewing the difficulty adjustment, network security, block propagation, orphan rates, and coin emission.

If the block reward remains at 50 BIG, daily issuance would increase fivefold, so the reward could also be adjusted to preserve the intended monetary emission.

What do you think about exploring a 2-minute block interval for BigCoin?

Best regards!
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 14, 2026, 11:04:23 AM
 #10

yespower is live - the switch at block 19,500 is done

Block 19,500 was mined at 2026-11-14 09:55:37 UTC. From that block on,
BigCoin's proof of work is yespower 1.0 instead of SHA-256d. Nothing else
about the rules changed, and the activation height is the one published in
the first post of this thread.

Code:
block        19,500
hash         9337cce12090a21bbeec58fb488cc47d776a686591db40f104678a81ea68eabc
time         2026-11-14 09:55:37 UTC
height now   19506   (6 blocks under the new algorithm)

Difficulty did not carry across

The SHA-256d difficulty the chain had reached was far out of reach for a CPU,
so it is discarded at the switch and the new algorithm starts from a fixed
value chosen for it. That is by design, not a reset of anything else.

Code:
block 19,499   difficulty 14.4924    SHA-256d   (bits 1c11aa05)
block 19,500   difficulty 0.001397   yespower   (bits 1e02cbd3)
right now      difficulty 0.001403

The starting value is one I had to pick in advance, so I will say plainly how
I picked it. I measured yespower on one ordinary desktop CPU and set the
anchor so that machine would be in the right order of magnitude for ten-minute
blocks - deliberately a little above it, not far below. On eight cores and
sixteen threads I measure about 8,250 hashes per second. (An earlier post in
this thread said "roughly 5,000 H/s"; that was my own conservative guess
before I measured properly, and the measured figure is the one to use.)

Since difficulty now adjusts every block, the first blocks after the switch
are irregular while it finds its level. So far 6 blocks have been found under
yespower, averaging 8.6 minutes apart, which is already close to the
ten-minute target.

If you run v1.7 or v1.8, there is nothing to do

Your node already knew about the switch and crossed it without being told.
The bundled mining script asks the node which algorithm the next block needs
and resizes its own work accordingly, so it changed over on its own.

If you run v1.5 or older, your node has stopped

It stopped at block 19,499 and it is not broken - it is correctly enforcing
the rule it was built with, and that rule says block 19,500 has invalid proof
of work. It will sit there on a chain that no longer grows. v1.6 was never
released, so v1.5 is the newest affected build.

Stop the old node, replace the binaries, start it again. That is the whole
procedure - no reindex, no rescan, no command to un-stick it.

Code:
Windows:  close BigCoin, unzip BigCoin-win64-v1.8.zip over the old folder,
          start it again
Linux:    stop bigcoind, unpack bigcoin-linux-x64-v1.8.tar.gz, start it again

Your wallet, your blocks and your bigcoin.conf live in the data directory,
not in the program folder, so replacing the binaries does not touch them.

The reason nothing else is needed: when the old build rejected block 19,500
it discarded the header instead of recording it as bad, so the new build asks
for it again on the next start and accepts it. The node catches up from
19,499 by itself.

Code:
http://133.242.139.204/downloads

dcfebf2f476331f0d83de481379b45f4a523de7f545a456a107b01434642b972  BigCoin-win64-v1.8.zip
66f8b838370410119ed396d6f6cd44da9425bba7b85cedb0232fa4557743a882  bigcoin-linux-x64-v1.8.tar.gz
67b32b9f494595400654b11f42eae79a222f35d8b1ebac145d5e817ffbceb472  bigcoin-source-v1.8.zip
422375eafe2e11a7daf29292b1681449599b93d874f4fbe9dc4ac90f692ed6d2  bigcoin-source-v1.8-unix.tar.gz
b4973f0c5c6a6f59c276f899a21a7453a60e35e3b916d222e55fa27647c4bf26  mine-bigcoin-v1.8.cmd
67d55214e84a5571358c09d2c875f0b49d94a08307ac32e93e79cc56fec574dd  mine-bigcoin-v1.8.sh

These are the same digests as in the v1.8 release post earlier in this thread.
The v1.7 files are all still published at the same URLs with the same digests,
and those are in the first post and in the mining guide thread. None of those
posts is hosted on my server.

If you want to read the switch itself

The clone URL and the upstream comparison are in the v1.8 release post
earlier in this thread, so I will not repeat them here. What is worth
pointing at is where this change lives: five modified files - src/pow.cpp,
src/pow.h, src/consensus/params.h, src/kernel/chainparams.cpp and
src/validation.cpp - plus the added src/crypto/yespower, which is the
yespower 1.0 reference code. If you want to check what the switch actually
does, those are the files.

One thing to flag before you read them. The comment above yespowerAnchorBits
in src/kernel/chainparams.cpp still carries my first measurement, 4,910 H/s,
and calls the anchor "about twice" that rate. The figure further up this
post - 8,250 H/s, which makes the anchor about 1.2 times it - is the correct
one. I have left the released source exactly as it was published rather than
quietly editing it, since its digests are in this thread; it is still there
in v1.8, which does not touch src/ at all. It gets corrected in the first
release that changes the source. The code itself is unaffected - the anchor
is 0x1e02cbd3 either way - but I would rather point that out myself than
have you find it and wonder what else is stale.

On my timing estimates

The first post of this thread put the switch around 14 November. Partway
through I talked myself out of that: the hashrate rose about sevenfold in nine
hours, blocks started arriving every ninety seconds, and I said at the time
that it would be late on 10 November or during 11 November. The hashrate did
not stay: it fell back, the interval stretched out again, and the height was
reached on 14 November - which is where the first estimate had it.

I am not going to claim that as a good forecast. It was right by accident,
after I had replaced it with one that was wrong. Difficulty here adjusts every
block, so the block interval tracks whoever happens to be mining at the time,
and any date derived from it is only as good as the assumption that they keep
mining. I should have said that instead of giving a date twice. Watch the
height rather than the calendar:

Code:
bigcoin-cli getblockcount

(The binary is bigcoin-cli - an earlier post in this thread mistyped it as
bitcoin-cli, which does not exist in any release.)

On the suggestion to move to two-minute blocks

Two people in this thread suggested that the target interval drop from ten
minutes to two at this switch. The short answer is that it stays at ten
for now and I will come back to it: nothing about the interval changes
here, and I would rather revisit it with measurements from yespower in
hand than decide it today. It is a fair thing to ask, so it is worth
being specific about why rather than just saying no.

It could not have been in this one. The height was published on 1 November
and the binaries that enforce it have been in people's hands for days; a
second rule change bolted onto the same height would have meant every node
already running v1.7 or v1.8 rejecting the chain on the spot. And the
commitment in this thread is at least 30 days between announcing a
consensus change and the height that activates it. Even if I decided today
that two minutes was right, the earliest honest activation is a height
about a month out.

Both suggestions said the block reward would need adjusting to keep
issuance the same. That is right for daily issuance - 50 BIG every ten
minutes and 10 BIG every two minutes are both 7,200 BIG a day - but the
reward is only half of it. The halving interval is 210,000 blocks, not a
span of time, so five times as many blocks pulls every halving five times
closer. From where the chain is now, the first halving moves from about
three and a half years away to about eight months. Holding the emission
curve steady needs that interval moved to 1,050,000 as well. That is the
parameter that decides the money supply, and it is not one to change in
the same breath as the reward.

Three other things in this code are written as spans of time or as fixed
block counts, and shortening the interval quietly rescales all of them:

Code:
                                   at 10 min      at 2 min
stall grace (6 h)                  36 intervals   180 intervals
future timestamp allowance (2 h)   12 intervals   60 intervals
median time past (11 blocks)       110 min        22 min
coinbase maturity (100 blocks)     16.7 h         3.3 h

The first is the v1.7 mechanism that eases difficulty when the chain falls
silent - at two minutes it would sit through 180 missed blocks before it
helped. The second matters more than it looks, because ASERT sets the
target from block timestamps: an allowance that spans twelve intervals
today spans sixty of them at two minutes, which hands anyone with a large
share of the hashrate a much longer lever on the difficulty. The last two
are not wrong, only weaker in wall-clock terms than the block counts make
them sound - and the maturity one cuts both ways, since it also means a
miner waits three hours rather than seventeen to spend what it found.

What would not break, for the record, is ASERT itself. Its half-life is
two days in seconds, not in blocks, so its response to a change in
hashrate is the same in wall-clock terms either way. It can tighten by at
most 0.24% per block now and 0.048% per block at two minutes - five times
less per block, five times as many blocks, the same curve.

The part I weigh heaviest is what it would do to the people already
mining here, and it is not where I expected to find it. It is not the
network. Blocks on this chain are nearly empty, so they propagate fast and
the orphan rate from propagation alone is a fraction of a percent at
either spacing. It is the mining script.

The published script mines by handing the node a nonce budget and letting
it grind, and that call does not notice a new tip while it is running. It
sizes itself to about thirty seconds, and the reason is written in the
script's own comments: at a 600-second target spacing, a 30-second call
carries roughly a 5% risk that a block it finds is built on a parent that
is no longer the tip. At a 120-second spacing that risk is roughly 25%.

That is one constant in one file and I would of course change it. But the
change only reaches whoever downloads a new script. The files I publish
keep their digests, so mine-bigcoin-v1.7.cmd and v1.8.cmd would stay up
exactly as they are, and anyone still running them would quietly lose
about a quarter of what they found. Anyone who wrote their own miner, or
who checks the tip every couple of seconds, would not notice at all. So
the cost lands on precisely the casual solo miner the suggestion was meant
to help, and it lands through the tooling rather than through consensus.
That is fixable - but it has to be fixed before such a change, not after.

So the decision for now is the dull one: the target spacing stays at ten
minutes, and this switch changes the algorithm and nothing else. That is
not a refusal. Two minutes is defensible and ten is defensible; ten is
what BigCoin inherited rather than something I chose for it, and the
argument in its favour is better than one of the ones I have just made
against it - five times as many blocks at a fifth of the size makes a
solo miner's income much less lumpy, and less lumpy is worth something
real when you are mining alone on a CPU.

What I will do is look at it properly rather than leave it as a thread
reply. Once yespower has been running long enough to show what the real
block interval looks like, I will post what the measurements say about it
here. I have already put a date on this thread twice by reasoning ahead
of the measurement - one of them missed and the other was right by
accident - and I would rather not do that a third time with a consensus
parameter. And if that ever turns into a decision to change the
interval, it gets its own announcement, its own height, and at least 30
days, like anything else here - not a line buried in an update.

Nothing else is changing

Every block mined with SHA-256d stays valid for ever, and every BIG already
mined stays spendable - the switch changes how new blocks are found, not what
the old ones are worth as history. I have no further consensus change
scheduled, and the interval question above is something I intend to measure
and report on rather than something already queued. If a change ever does
become necessary, the commitment I made in this thread stands: at least 30
days between the announcement and the activation height, with the height
written in the release notes and here.

- Masashi Takamoto
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 15, 2026, 11:52:56 AM
Last edit: November 16, 2026, 07:53:27 AM by takamoto_bigcoin
 #11

v1.9: a node that joins the network with no config file -- and how to let others reach yours

Two things in one post. First the release, which is small and mostly one fix.
Then the part that has no version number attached to it: how to accept inbound
connections over Tor, IPv6 or a forwarded port, and what each of those
publishes about you. Links and digests are at the foot of this post, and they
will be at the foot of every release post from here on.

What v1.9 fixes

Until v1.8 no seed node was compiled into the binaries. A node that never read
bitcoin.conf never found a single peer -- and copying that config file is
exactly the step a newcomer skips. The seed is now built in. The 60-second
wait that stood in front of it is gone as well: Bitcoin Core only falls back
to its built-in seeds after giving DNS seeds a minute to answer, and BigCoin
has no DNS seeds for that minute to be worth anything.

Measured on an empty data folder, with no config file, no -addnode and no
-seednode:

Code:
                          v1.8     v1.9
  seeds loaded             +61 s     +0 s
  first peer connected     +61 s     +1 s

In the v1.9 run the node went from the genesis block to the then current tip
at height 19,694 in 84 seconds from launch. Unzipping the archive and starting
the node is now enough. Copying bitcoin.conf is still worth doing -- it is
where your own settings go -- but it is no longer what gets you onto the
network.

Both the Windows and the Linux build carry the seed and report v31.1.9. v1.8
reported v31.1.7, because its binaries were byte-identical to v1.7's; this
time they are not. bitcoind -version, or Help -> About in the GUI, is how you
tell which one you are running.

What v1.9 does not change

No consensus rule, no activation height, no default. A v1.9 node and a v1.7
node accept exactly the same blocks, and the node I run has been on this build
since the morning of 15 November, with v1.7 peers at the other end of every
connection. If you
run v1.7 or v1.8 there is nothing here you need to install.

If you still run v1.5 or older, you are already stuck. Block 19,500 was mined
on 2026-11-14 at 09:55 UTC and the proof of work changed to yespower 1.0
there. Those builds do not contain the algorithm at all, so they rejected the
first block above 19,499 as invalid proof of work and stopped, on a chain that
no longer grows, without reporting anything as wrong. Upgrading is the only
fix. v1.6 was never released, so v1.5 is the newest affected build.

Documentation corrections

The readme gained a section on encrypting the wallet, all of it measured on
this build. The part that catches people out: encrypting a wallet generates a
new HD seed, so a backup taken before you encrypt does not cover the addresses
you create afterwards. There is also no way to decrypt a wallet again, and
mining carries on while the wallet is locked, because getnewaddress needs no
private key.

The backup advice in the same readme gained the other thing measured on this
build: a normal Windows restart or shut down is safe for a running node.
Windows sends a console shutdown event, the node's handler takes the same path
as the stop command, and on this build that takes about a second -- inside the
time Windows waits. Task Manager's End task and a power cut send no such
signal, and those are the ones that can cost you a reindex.

Four other things were wrong rather than out of date. All four were written
for a chain that had not reached height 19,500 yet:

- The getting-started steps gave the nonce ceiling for one mining call as
  20000000. That was the right order of magnitude for SHA-256d. A yespower
  hash costs about 3 ms, so the same call now runs for hours before it
  returns. It reads 4000.
- Those same steps said v1.1 or later for the command line and v1.2 or later
  for the GUI. Both of those builds stop dead at 19,500. They now say v1.7
  or later.
- One page of the project site still described the proof of work as the same
  as Bitcoin's, so that Bitcoin mining hardware would work as-is. That
  stopped being true at 19,500 and the page now says so. The mining document
  is in the past tense for the same reason.
- The project document said the next release would shrink the 1,749-block
  range above height 17,750 that an ASIC could in principle re-mine, by
  pinning blocks inside it. v1.9 does not do that. The document now says no
  release has done it, rather than naming the next one.

The comment in src/kernel/chainparams.cpp that explains the yespower
difficulty anchor quoted 4,910 H/s and called the anchor "about twice" that
rate. Measured on the same machine it is 8,250 H/s idle and about 5,300 H/s
while the machine is being used for something else, which makes the anchor
about 1.2x the idle rate. The anchor value itself is untouched -- changing it
would be a consensus change.

Source

Code:
git clone http://133.242.139.204/git/bigcoin.git
git remote add upstream https://github.com/bitcoin/bitcoin.git
git fetch --depth=1 upstream tag v31.1
git diff --ignore-cr-at-eol --stat v31.1 v1.9

55 files, three more than v1.8: contrib/seeds/nodes_main.txt,
src/chainparamsseeds.h and src/net.cpp, which is the whole of the change
described above. Tag v1.9 holds the published source archive file for file.
The archives are packed on Windows, so contrib/bigcoin/mine-bigcoin.cmd has
CRLF line endings there and LF in git; VERIFYING.md gives the flag that
accounts for it. Only --stat honours that flag, so --name-only and
--name-status list fourteen files whose contents are identical and whose line
endings are not, and report 69 entries for the same diff. A clone proves
nothing by itself: the digests in this thread are the anchor, because they are
not hosted on my server.

The miner scripts keep their v1.8 names

There is no mine-bigcoin-v1.9.cmd or .sh. Those scripts are byte-for-byte the
ones published for v1.7, renamed once for v1.8, and a third name for the same
file would only add another digest to check. The quick start in the mining
guide thread is unchanged.

A page for telling a slow miner from a broken one

Published alongside this release and not part of the program: a page that
answers "is my miner actually doing anything" for someone who has not found a
block yet.

Code:
http://133.242.139.204/mining/status

Put your mining address in and it reports blocks found, the reward still
maturing, your share of the last 500 blocks, how many blocks you found in the
last 24 hours, the interval that share implies, and the network's current rate
per hour. It reads the chain: nothing is stored and nothing is submitted.
Bookmark the result and it stays your own page.

It will not tell you about anyone else. There is no list and no ranking --
only the address you typed and network-wide totals.

Why it exists: a machine doing a few hundred hashes a second can go a day or
more between blocks at the current network rate with nothing wrong, which is
exactly what a miner that has quietly stopped also looks like. The share
figure is what separates the two. If you used the script and do not know which
address is yours, run this from the folder you unzipped. The script creates one
address per run, labelled mine1, mine2 and so on, so try the later labels too
if you have started it more than once:

Code:
bigcoin-cli.exe -datadir=data -rpcwallet=miner getaddressesbylabel mine1

If an antivirus removes a download

The Windows binaries are not code-signed, so SmartScreen warns about them, and
some antivirus products will delete a freshly downloaded unsigned executable
on reputation alone, before it has ever run. The answer is not to switch
anything off. Check the archive against the digest at the foot of this post,
and restore the file from quarantine if it matches. If it does not match, do
not run it, and say so in this thread.

Part two -- letting other nodes reach yours

Everything above gets you a node that mines. This is the other half: letting
other nodes reach yours, and what that costs in privacy. None of it is
required in order to mine, and none of it changes a consensus rule.

A node with default settings makes outbound connections and accepts none. That
is enough to validate the chain and mine on it. What such a node cannot do is
accept anybody else, and a new wallet can only reach nodes that listen. More
reachable nodes make the network easier to join and harder to interrupt, which
makes this the most useful thing a participant can add that is not hashrate.

There are three ways to be reachable. They differ mainly in what they publish
about you.

First, what peers already know

Read this part before you open anything.

A node that accepts no connections still gives its IP address to every peer it
dials out to. Not listening is not anonymity.

If you do listen on IPv4 or IPv6, your address is passed around the network in
ADDR messages, kept in the address database of every node that hears it, and
collected by anyone crawling the network. Treat it as public and permanent.

The address you mine to is in the chain forever. If the same line also runs a
listening node, the two can be set side by side. A faucet payout works the
same way: it is a permanent public link between the address you gave the
faucet and everything that address later touches.

Port 8337 is the RPC port. Never forward it, and never bind it to anything but
localhost. The RPC cookie sits in your data directory, and a wallet with no
passphrase can be emptied by anyone who reaches that port. The only port that
is ever forwarded is 8338.

My own machines are not exempt. The explorer and the faucet are plain HTTP, so
they see your IP address the way any website does. The onion address at the
foot of this post is a node, not a web server; for the web pages, use Tor
Browser if that matters to you.

Where the config file lives

Code:
mine-bigcoin.cmd / mine-bigcoin.sh    data\bitcoin.conf  (next to the script)
bitcoind or bigcoin-qt with -datadir  bitcoin.conf in that folder
bitcoind or bigcoin-qt on Linux       ~/.bitcoin/bitcoin.conf

The miner scripts write data\bitcoin.conf once and never rewrite it, so lines
you add there survive. If you started bigcoin-qt by double-clicking it and
gave it no -datadir, it is using a default folder in your user profile, and
the program will tell you which: Help -> Node window -> Information, the line
labelled Datadir. That folder is named Bitcoin rather than BigCoin, because
the fork inherits the path from Bitcoin Core unchanged -- so if you also run
Bitcoin Core on the same machine, give BigCoin its own -datadir, or the two
will try to share one folder and one bitcoin.conf. Restart the node after
editing.

Option 1 -- Tor: reachable without publishing an address

You get an .onion address, inbound connections, and no IP address shown to
peers. The node builds the hidden service itself; there is nothing BigCoin
specific to put in Tor's config.

Tor needs a control port. These are the settings:

Code:
SocksPort 9050
ControlPort 9051
CookieAuthentication 1
CookieAuthFileGroupReadable 1
DataDirectoryGroupReadable 1

On Linux they are usually already in /etc/tor/torrc, commented out. Uncomment
them, restart Tor (sudo systemctl restart tor), and put your own user in
Tor's group so it can read the control cookie:

Code:
sudo usermod -aG debian-tor YOUR_USER

then log out and back in. On Windows, use the Tor Expert Bundle rather than
Tor Browser: Tor Browser listens on 9150 and opens no control port. Put the
first three lines in a torrc file of your own and start it with
tor.exe -f C:\path\to\torrc.

Then two lines in bitcoin.conf:

Code:
proxy=127.0.0.1:9050
addnode=rjxywola2kupk4rp67qzidqiuxwvr76xmri5zimt6ecrvhx4xv5t7wyd.onion:8338

That is the whole of it. The node asks Tor for a hidden service, writes the
key to onion_v3_private_key in its data directory, and starts accepting
inbound connections through it. -proxy applies to every network, not just
onion, so outbound IPv4 and IPv6 connections go through Tor as well and your
address is not shown to peers at all.

Check it with:

Code:
bigcoin-cli getnetworkinfo    -> localaddresses contains ...onion:8338
bigcoin-cli getpeerinfo       -> "network": "onion"

If nothing shows up, restart with -debug=tor and read the log. A permissions
error on /run/tor/control.authcookie is the usual cause.

Three things worth knowing:

onion_v3_private_key is your node's identity. Delete it and you get a new
address; give it to someone else and they can be your node. Keep it out of any
backup you share.

Do not set listen=0 in order to "only use Tor". listen=0 switches the hidden
service off with everything else. The way to be reachable over Tor only is
bind=127.0.0.1:8338, which keeps the listener on loopback where nothing but
Tor can reach it, and as a side effect stops the node from advertising your
real addresses at all. (With no bind line the node opens a second loopback
listener on 8339 for the hidden service to point at. Either arrangement
works.)

onlynet=onion is the strict version: no clearnet peers and no Tor exits. Until
more onion nodes exist that leaves you with very few peers, and a node with
one or two peers only sees what they choose to show it. proxy= on its own
gives you a normal set of peers with your address still hidden, which is the
better trade for most people right now.

Tor hides your address from peers. It does not hide from your ISP that you are
using Tor.

Option 2 -- IPv6

If your ISP gives you IPv6, this is the least work of the three: there is no
port mapping to set up, only a firewall rule to add. The node already listens
on :: port 8338.

In the router, allow inbound TCP 8338 to that one machine's IPv6 address --
not the whole prefix, and not 8337. On Windows, let bitcoind or bigcoin-qt
through Windows Firewall as well; the prompt on first run does this. The
address then appears under localaddresses in getnetworkinfo once a peer
confirms it.

The privacy cost is specific: an IPv6 address carries your prefix, which
identifies your line, and the address you open has to be a stable one to be
useful -- so what gets published is a stable identifier for your connection
rather than a temporary one. An IPv4 address shared behind carrier NAT is
vaguer than that.

Option 3 -- IPv4 port forward

Give the machine a fixed LAN address first (a DHCP reservation), then forward
TCP 8338 to it. Without the reservation the rule ends up pointing at a
different machine after a reboot. Let bitcoind or bigcoin-qt through the local
firewall too.

If the WAN address shown in your router is inside 100.64.0.0/10, or simply
does not match what a "what is my IP" page tells you, you are behind carrier
NAT and no amount of forwarding will work. Use Tor instead.

upnp=1 and natpmp=1 ask the router to open the port for you. Both are off by
default, and both are a request to a device that will then be forwarding a
port on your behalf: set them deliberately or leave them alone.

Verifying is where people fool themselves:

Code:
bigcoin-cli getpeerinfo    -> look for "inbound": true

An actual inbound peer is the only proof. The reachable flags in
getnetworkinfo describe what the node believes about its own configuration,
not whether anything got through from outside. Give it an hour or so; inbound
connections arrive when other nodes get round to your address.

Which one

Code:
want to help, address stays private    Tor
have IPv6, address may be public       IPv6 firewall rule on 8338
IPv4 only, not behind carrier NAT      forward 8338
not sure                               do nothing; mining works either way

Risk, stated plainly

None of this turns a node into a privacy product. Opening a port invites
unsolicited traffic from the whole internet, including scanners that have
never heard of BigCoin. Tor costs latency and depends on the Tor network
being up. Running a reachable node means peers learn either your address or
your onion, and I run the seed and the explorer, so I see connections to
those. This is a young chain, so if yours is one of the first reachable nodes,
that is visible. Decide with all of that in front of you.

BIG has no promised value. The only way to get it is to mine it.

Links and digests

Code:
Block explorer     http://133.242.139.204/
Downloads          http://133.242.139.204/downloads
Mining status      http://133.242.139.204/mining/status
Faucet             http://133.242.139.204/faucet
Source             git clone http://133.242.139.204/git/bigcoin.git
Seed node          133.242.139.204:8338
Seed node (Tor)    rjxywola2kupk4rp67qzidqiuxwvr76xmri5zimt6ecrvhx4xv5t7wyd.onion:8338
ANN thread         https://bitcointalk.org/index.php?topic=5598514
Mining guide       https://bitcointalk.org/index.php?topic=5599204

f3999cc9d6da8a032a4ba8d2023a098e18dfeabe90e0369693b451415c5fd5cb  BigCoin-win64-v1.9.zip
cf85ed4f4b158e3c8c292052025252182c8f80d037a8419f5ef6bea5540b5ba8  bigcoin-linux-x64-v1.9.tar.gz
d287776d917d0d22c063ffbc6c6ed91bba3fe96b8c9f385a594a2de1caf790bb  bigcoin-source-v1.9.zip
d44066c8bf5d70404b66d21d6689af4d99a13695b1bbf047816c47cc6b99c693  bigcoin-source-v1.9-unix.tar.gz
b4973f0c5c6a6f59c276f899a21a7453a60e35e3b916d222e55fa27647c4bf26  mine-bigcoin-v1.8.cmd
67d55214e84a5571358c09d2c875f0b49d94a08307ac32e93e79cc56fec574dd  mine-bigcoin-v1.8.sh

Every v1.7 and v1.8 file is still at the same URL with the same digest.
Nothing published has been replaced.

- Masashi Takamoto
EL MAOGO
Newbie
*
Offline

Activity: 96
Merit: 0


View Profile
November 16, 2026, 04:05:27 AM
 #12

I appreciate the work behind v1.9, especially the integrated seed nodes and the improvements to the initial synchronization process.

I've been testing BigCoin mining on Linux using an Intel i7-13700, and I find the transition to Yespower 1.0 particularly interesting.

I have one important question regarding the long-term decentralization of the network:

Would you consider adding more independent seed nodes and encouraging community members to operate publicly reachable nodes?

Having multiple independent nodes would reduce reliance on a single infrastructure provider and make the network more resilient if the main seed server becomes unavailable.

I also believe that making it easier for ordinary users to run a full node and contribute to network connectivity is just as important as increasing hashrate.

Thanks for continuing to develop and maintain BigCoin. It's encouraging to see an actively maintained project with a focus on solo mining and independent participation.
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 16, 2026, 07:19:48 AM
Last edit: November 16, 2026, 07:54:01 AM by takamoto_bigcoin
 #13

Reachable nodes: a page of their own, and how to get yours into the next release

Both of those were the right things to ask for, and the second one is the part I
cannot do by myself.

The instructions have a page now

Last week's reachability instructions were at the bottom of a long release post,
which is the wrong place for something nobody reads on release day and several
people need a week later. They now have a page of their own, kept up to date
separately from any release:

Code:
http://133.242.139.204/node

What peers already know before you open anything, where the config file actually
lives, Tor on Windows and on Linux, IPv6, an IPv4 port forward, how to tell
whether it worked, and what each option publishes about you. No new binaries and
no new version: this is a web page, nothing to download and nothing to install.

It is longer than the release post on one point in particular -- Tor on Windows,
step by step -- because that is the arrangement running on a Windows machine
here, and the release post skipped the parts that actually caught me out:

Tor Browser will not do. It listens on 9150 and opens no control port, so
the node cannot ask it for a hidden service at all. The Tor Expert Bundle is the
one you want, with a torrc of your own.

Tor can idle itself out, and a dormant Tor stops publishing the onion
descriptor.
The node then silently becomes unreachable again, with nothing in
its own log to say so, and getnetworkinfo still showing the onion address.
DormantTimeoutEnabled 0 in torrc is the fix. I would not have guessed at that
one.

The reachable flags in getnetworkinfo are not a check on any of this. They
report whether the node is willing to use a network, which only -onlynet
changes. They say nothing about whether anything got through from outside. An
inbound peer in getpeerinfo is the only proof, and I had this wrong in my own
notes for longer than I would like.

The page also keeps the warning that matters more than any of the above: a node
that accepts no connections still gives its address to every peer it dials out
to. Not listening is not anonymity, and Tor is the option that actually fixes
that rather than the option for people who want extra work.

On more seed nodes

The built-in list takes as many entries as it is given. One address is in it
today, and nothing in the program has to change to add more. It is not a
consensus rule, so the 30-day notice I committed to in this thread does not
apply to a seed address and would only slow it down.

The useful thing to be clear about: a second seed run by me would still be run
by me. It would be a second address, not a second operator, and it would not
answer the question you asked. What changes that is other people's nodes.

So, concretely: if you run a reachable node and want it compiled into the next
release, say so in this thread.
What is needed:

- Reachable in fact, proved by "inbound": true in getpeerinfo -- not by a
  config flag and not by the reachable flags above.
- An address you expect to answer for months. An onion address is stable by
  construction, which is one more reason to prefer it.
- Onion, IPv6 or IPv4 are all usable. An onion address publishes nothing about
  your line: it is the node's identity, not your connection's.
- You accept that it is permanent. A seed address is compiled into released
  binaries and cannot be recalled from copies that are already downloaded.
  Think about that before offering an IP address rather than an onion.

Built-in seeds only matter on a node's first start, while its address book is
still empty. After that the node learns peers from the network and never reads
the list again. That first minute of a newcomer's first run is the part nobody
can route around, which is exactly why the list wants more than one entry in it.

No list of reachable nodes is published anywhere, and there is no ranking. If
you volunteer an address for the seed list, that address goes into the software
and nothing else about you does.

Two things from the mining side, since Linux came up

Both of these are easy to get wrong and neither is in the readme yet.

Raise rpcthreads if you raise the thread count. The miner holds one RPC
connection per worker, and bitcoind's default is rpcthreads=4. Past about four
workers the workers start queueing on the node rather than hashing, and the
symptom is a hash rate that stops improving as you add threads -- which reads
exactly like a CPU limit. Put rpcthreads=N+2 in bitcoin.conf for N workers.

On a recent desktop Intel part, the thread count the OS reports is not the
number you want.
Those chips have two different kinds of core and the count
includes both. yespower at these parameters needs 2 MB per thread and is limited
by memory latency rather than arithmetic, so threads that land on the smaller
cores, which share much less cache between more of them, do not contribute what
the number suggests. I would not predict the right figure -- measure it. Run at
the full thread count, then at the number of large cores only, and keep whichever
gives more blocks per hour over a few hours. The mining status page is the
quickest way to compare the two runs:

Code:
http://133.242.139.204/mining/status

Risk, stated plainly

None of the node page turns a node into a privacy product. Opening a port
invites unsolicited traffic from the whole internet, including scanners that have
never heard of BigCoin. Tor costs latency and depends on the Tor network being
up, and it does not hide from your ISP that you are using Tor. Running a
reachable node means peers learn either your address or your onion, and I run
the seed and the explorer, so connections to those are visible to me. This is a
young chain, so if yours is one of the first reachable nodes, that is visible
too.

BIG has no promised value. Nothing is for sale. The only way to get it is to
mine it.

Links

Code:
Run a node         http://133.242.139.204/node
Block explorer     http://133.242.139.204/
Downloads          http://133.242.139.204/downloads
Mining status      http://133.242.139.204/mining/status
Faucet             http://133.242.139.204/faucet
Source             git clone http://133.242.139.204/git/bigcoin.git
Seed node          133.242.139.204:8338
Seed node (Tor)    rjxywola2kupk4rp67qzidqiuxwvr76xmri5zimt6ecrvhx4xv5t7wyd.onion:8338
ANN thread         https://bitcointalk.org/index.php?topic=5598514
Mining guide       https://bitcointalk.org/index.php?topic=5599204

No digests this time, because nothing published has changed. Every v1.7, v1.8
and v1.9 file is still at the same URL with the same digest given in the release
posts above.

- Masashi Takamoto
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 19, 2026, 11:03:54 AM
 #14

yespower is live: what the switch actually did

The notice at the top of this thread said the proof of work would change at block 19,500. It did. This is the follow-up with numbers, including the parts that have not settled yet.

What happened

Block 19,500 was mined under yespower 1.0 and the chain carried on without a split. Everything up to and including height 19,499 is still SHA-256d and still valid. No block was reorganized, no BIG was invalidated, and no wallet needed to do anything.

Nodes older than v1.7 stopped following the chain at 19,499, as announced. If you have one, it is still sitting there and will not catch up until you upgrade. v1.9 is the current release.

The numbers

Code:
yespower active since   block 19,500
blocks since the switch 1105
elapsed                 5.0 days

work per block, at the switch     6.02 Mhash
work per block, now              14.88 Mhash   (x2.47)

mean spacing, whole era          395 s
mean spacing, last 144 blocks    514 s
mean spacing, last  72 blocks    566 s
mean spacing, last  24 blocks    570 s
target spacing                   600 s

network hashrate, last 144       28711 H/s
network hashrate, last  72       26333 H/s

The switch deliberately reset the target to something a CPU could solve, and ASERT has been pulling it back toward the 10-minute target ever since. That is the whole story of the table: work per block has more than doubled since the switch, and the mean spacing is climbing toward 600 seconds from below.

The hashrate figure is work divided by elapsed time, taken from the chainwork in the block headers. I am not quoting getnetworkhashps, which averages over a window and reads badly right after a retarget.

What a CPU actually gets

At the work per block above, a CPU doing about 6,000 hashes per second averages roughly 41 minutes per block. That figure is measured, not modelled: 6,000 H/s is what a 16-thread CPU produces here while the machine is also being used for other things.

"Averages" is carrying a lot of weight in that sentence. Block finding is a Poisson process: if your expected interval is forty minutes, then a three-hour gap with nothing is unremarkable, and so are two blocks ten minutes apart. Judge your hashrate over days, not over one dry spell. If you want to check it against something, the mining status page records what the chain actually did.

The honest part: this gets harder. Every CPU that joins raises the difficulty for everyone already here, and ASERT will keep tightening until the spacing sits at the target. Nothing about yespower makes CPU mining permanently cheap. It only makes it possible, which SHA-256d no longer was.

How the difficulty adjusts

The specifications block in the opening post used to say "retarget every 2016 blocks". It had been wrong since height 17,325, and I corrected it on 17 November. An edit to a first post is easy to miss, so here is the whole schedule in one place.

Code:
to 17,324    legacy Bitcoin retarget, every 2016 blocks
from 17,325  ASERT (aserti3-2d), every block, half-life 2 days
from 18,000  once the chain has been silent for more than six hours,
             ASERT measures the elapsed time against the candidate
             block's own timestamp, so a stalled chain gets easier
             without having to wait for a block to prove the stall
from 19,500  yespower 1.0 replaces SHA-256d, re-anchored to a
             CPU-solvable target

Per-block adjustment is why the difficulty could more than double in a few days instead of waiting two weeks for a retarget, and the real-time part is why a chain that goes quiet recovers on its own instead of freezing. Both are in pow.cpp in the published source, with the reasoning in comments.

Links and digests

Code:
Block explorer     http://133.242.139.204/
Downloads          http://133.242.139.204/downloads
Mining status      http://133.242.139.204/mining/status
Faucet             http://133.242.139.204/faucet
Source             git clone http://133.242.139.204/git/bigcoin.git
Seed node          133.242.139.204:8338
Seed node (Tor)    rjxywola2kupk4rp67qzidqiuxwvr76xmri5zimt6ecrvhx4xv5t7wyd.onion:8338
Running a node     http://133.242.139.204/node
ANN thread         https://bitcointalk.org/index.php?topic=5598514
Mining guide       https://bitcointalk.org/index.php?topic=5599204

f3999cc9d6da8a032a4ba8d2023a098e18dfeabe90e0369693b451415c5fd5cb  BigCoin-win64-v1.9.zip
cf85ed4f4b158e3c8c292052025252182c8f80d037a8419f5ef6bea5540b5ba8  bigcoin-linux-x64-v1.9.tar.gz
d287776d917d0d22c063ffbc6c6ed91bba3fe96b8c9f385a594a2de1caf790bb  bigcoin-source-v1.9.zip
d44066c8bf5d70404b66d21d6689af4d99a13695b1bbf047816c47cc6b99c693  bigcoin-source-v1.9-unix.tar.gz
b4973f0c5c6a6f59c276f899a21a7453a60e35e3b916d222e55fa27647c4bf26  mine-bigcoin-v1.8.cmd
67d55214e84a5571358c09d2c875f0b49d94a08307ac32e93e79cc56fec574dd  mine-bigcoin-v1.8.sh

These digests are published here as well as on the downloads page on purpose: this thread is not hosted on my server, so if that server were ever compromised the attacker could rewrite an archive and the hash page together, but not this post. Every v1.7, v1.8 and v1.9 file is still at the same URL with the same digest. Nothing published has been replaced.

- Masashi Takamoto
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 20, 2026, 11:13:35 AM
 #15

The coin is now called Varve, and the downloads have a certificate: https://varve.mine.bz/

Two announcements, in one post because they happened together.

1. The name

BigCoin was already taken, and not by something dead. There is a token on another chain using the same name, with a nearly identical supply figure, the same halving story and the same mine-it-yourself pitch, and there is a separate ERC-20 using the ticker. I am not going to argue about who was first or what any of them are worth. The practical problem is simpler: someone who hears about this project and searches for it does not find it. They find those. A name that sends readers to somebody else's token is a liability, not a brand, and the longer this chain runs under it the more expensive the fix gets.

So, from today, this chain is called Varve.

A varve is a layer of lake-bed sediment laid down in a single year. They stack in order, one layer per year, and the way you date a sediment core is by counting them. Nothing can be inserted between two layers once they have settled. That is the shape of a chain of blocks, described by a thing that actually exists rather than a word invented for a coin. As far as I can find, no project and no ticker is using it.

2. Nothing about the chain changed

This is a rename, not a release and not a fork. There is no upgrade, no activation height, no deadline, and nothing to install. If you are running v1.9 and mining today, do nothing at all.

Code:
Genesis block        unchanged
Network magic        unchanged - old and new nodes are the same nodes
P2P port 8338        unchanged
Addresses            unchanged - still big1... and B...
Unit                 unchanged - still BIG
Block/tx validation  unchanged
The binaries         unchanged - not rebuilt, same bytes, same digests
Every URL            unchanged and still working

The ticker is still BIG and addresses still start with big1 or B, because those are in the software you are running and in files on your disk. They do not change because a web page did. If the unit label ever moves, the old one keeps being accepted indefinitely and it gets announced before it ships rather than after. Worth saying plainly, since it is the question I would ask: changing how an address is spelled is not a consensus change - block and transaction validation rules are untouched by it, so it cannot split the network - but I am not doing it today either way.

3. The file names keep the old spelling, on purpose

The downloads page is now titled Varve and the files on it are still called BigCoin-win64-v1.9.zip and bigcoin-source-v1.9.zip. That is not something I forgot to finish.

Those are the files that were published, and their SHA-256 digests are printed in this thread next to those names. Renaming a published file invalidates every check anyone has already made against it, and it would mean this post disagreed with the posts above it. So the archives, the digests, the mine-bigcoin scripts, the git clone path, the program names and the data directory path all keep the spelling they shipped with. The rule I am following is that anything which exists on disk or in the protocol keeps its name, and only prose changes.

4. The genesis block still says BigCoin, permanently

The coinbase of block 0 contains the text BigCoin genesis 21/Oct/2026. It is inside the block that every node hashes and checks, so it can never be edited by anybody, including me. The old name is part of this chain's first block for as long as the chain exists. I would rather write that down than have somebody discover it and wonder what else was quietly rewritten.

5. The downloads have a certificate

Since the first release, everything this project serves has been on a bare IP over plain HTTP. From a stranger's point of view that is the worst thing on the page: you are asked to download an unsigned program that mines, over a connection nobody can authenticate. There is now a name with a real certificate, and it carries the new spelling.

Code:
Block explorer     https://varve.mine.bz/
Downloads          https://varve.mine.bz/downloads
Mining status      https://varve.mine.bz/mining/status
Faucet             https://varve.mine.bz/faucet
Running a node     https://varve.mine.bz/node
Source             git clone https://varve.mine.bz/git/bigcoin.git

Same server, same files, same seed node. The certificate is from Let's Encrypt and renews itself. The clone path is still bigcoin.git for the reason in section 3.

6. Every old address still works

http://133.242.139.204/ and every path under it still work, and they will keep working. Those URLs are printed in release notes that have already shipped and in the bitcoin.conf bundled inside the archives. I am not going to break a URL that is already sitting in a file on someone else's disk.

So there is no HTTP-to-HTTPS redirect. The usual way to add TLS is to redirect everything to the new name, and that would have killed the IP URLs, because no certificate authority issues certificates for a bare IPv4 address. The name was added beside the old URLs rather than in front of them. Use whichever you prefer.

The name resolves to the address that has been published from the beginning - it is another way to the same server, not another server:

Code:
nslookup varve.mine.bz   ->  133.242.139.204

The digests have not moved either. Every digest at the bottom of this post was published in this thread before the name existed. I downloaded all six files over the new URL and hashed them rather than assuming: byte for byte the same files. Whichever address you use, the check is the one it has always been - the SHA-256 of what you downloaded, against a digest posted in this thread.

7. What the certificate does not do

TLS authenticates the transport. It says nothing about whether the program is worth running. The binaries are still not code-signed, SmartScreen will still warn about them, and some antivirus products still flag an unsigned program that mines on that basis alone. The digest is still the thing that matters.

8. If a name does not resolve for you

The name is a free subdomain under a shared domain at freedns.afraid.org. Some corporate networks and filtering resolvers block free dynamic-DNS domains wholesale, so there will be people for whom it simply does not work. That is part of why the IP URL is staying: it is the canonical one and the name is a convenience. If the name is ever lost, nothing breaks except the padlock.

The Tor address is unchanged.

Links and digests

Code:
Block explorer     https://varve.mine.bz/
Downloads          https://varve.mine.bz/downloads
Mining status      https://varve.mine.bz/mining/status
Faucet             https://varve.mine.bz/faucet
Running a node     https://varve.mine.bz/node
Source             git clone https://varve.mine.bz/git/bigcoin.git

The same things by IP, unchanged and permanent
                   http://133.242.139.204/
                   http://133.242.139.204/downloads
                   http://133.242.139.204/mining/status
                   http://133.242.139.204/faucet
                   http://133.242.139.204/node
                   git clone http://133.242.139.204/git/bigcoin.git

Seed node          133.242.139.204:8338
Seed node (Tor)    rjxywola2kupk4rp67qzidqiuxwvr76xmri5zimt6ecrvhx4xv5t7wyd.onion:8338
ANN thread         https://bitcointalk.org/index.php?topic=5598514
Mining guide       https://bitcointalk.org/index.php?topic=5599204

f3999cc9d6da8a032a4ba8d2023a098e18dfeabe90e0369693b451415c5fd5cb  BigCoin-win64-v1.9.zip
cf85ed4f4b158e3c8c292052025252182c8f80d037a8419f5ef6bea5540b5ba8  bigcoin-linux-x64-v1.9.tar.gz
d287776d917d0d22c063ffbc6c6ed91bba3fe96b8c9f385a594a2de1caf790bb  bigcoin-source-v1.9.zip
d44066c8bf5d70404b66d21d6689af4d99a13695b1bbf047816c47cc6b99c693  bigcoin-source-v1.9-unix.tar.gz
b4973f0c5c6a6f59c276f899a21a7453a60e35e3b916d222e55fa27647c4bf26  mine-bigcoin-v1.8.cmd
67d55214e84a5571358c09d2c875f0b49d94a08307ac32e93e79cc56fec574dd  mine-bigcoin-v1.8.sh

These digests are published here as well as on the downloads page on purpose: this thread is not hosted on my server, so if that server were ever compromised the attacker could rewrite an archive and the hash page together, but not this post. Every v1.7, v1.8 and v1.9 file is still at the same URL with the same digest. Nothing published has been replaced, and the rename did not replace anything either.

- Masashi Takamoto
visibleplayer
Jr. Member
*
Offline

Activity: 166
Merit: 5


View Profile
November 20, 2026, 12:53:34 PM
 #16

Make new topic with new coin name?
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 20, 2026, 03:57:41 PM
 #17

No new topic, but you are right about the title. I have put the new name in the subject of this topic instead.

Same topic, same URL, same post history, same digests - only the title line changed. A second thread would have left every published digest behind under the old title, and I would have had to answer in both places from now on.
yoshikiazuma
Member
**
Offline

Activity: 168
Merit: 10


View Profile WWW
November 20, 2026, 10:27:45 PM
 #18

hi there takamoto

ive got a little idea

by using default yespower 1.0 im getting a little bit of problems with running along side my other miners

like multi mining on one pc with dozens of other coins algorithm mining alongside

i would suggest in the future to switch the algorithm to randomx or argon2

or maybe if you want to keep on the yespower algorithm (i like yespower but needs a little work done on it so it performs a little better)

could you give me a little idea on what bitcoin version is this or what bigcoin is about and are you looking to continue the works with this type of bitcoin protocol

thanks
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 20, 2026, 11:25:28 PM
 #19

Thanks for reading it closely. Short answers, in order.

The algorithm is not changing. I wrote in this thread that I do not intend to
change the proof of work again, and that stands. If any consensus change ever
does happen there will be at least 30 days between the announcement and the
activation height, with the height printed in the release notes and here, so
nobody wakes up on the wrong side of a fork. There is also a floor under how
cheap such a change could be: the personalisation string
"BigCoin/yespower1.0" is an input to the hash function itself, not a label, so
editing it would reject every block already on the chain. It keeps that
spelling permanently, rename or no rename.

On the actual problem - yespower not sharing a box well with your other
miners - switching to either of the algorithms you named would make that
worse rather than better:

Code:
yespower N=2048 r=8   2 MiB per hash, per thread
RandomX               2080 MiB dataset in fast mode, 256 MiB in light mode
Argon2 variants       tens to hundreds of MiB per lane, depending on the coin

2 MiB per thread is the smallest footprint of the three, and it is the reason
this one can run next to other work at all. The contention you are hitting is
cache and memory bandwidth, not clock, so the knob that helps is thread count,
not process priority: run fewer yespower threads than you have hardware
threads and leave the rest of your miners their share of cache. I have not
published a thread-scaling curve, so measure it on your own machine rather
than trusting a number from mine - the miner takes --threads N and the mining
guide has the command line.

On yespower needing work to perform better: a 2x CPU optimisation would be
welcome, since it helps everyone equally and changes nothing about validation.
But yespower was chosen precisely because no implementation is dramatically
faster than a plain CPU one. If a 10x speedup turned up, that would not be a
tuning win, it would mean the algorithm had failed at the only job it was
picked for. For reference, my own 8-core machine measures 7,000-8,250 H/s.

What this is: Varve (unit BIG, formerly BigCoin) is an independent Bitcoin
Core 31.1 fork - own genesis block dated 21 October 2026, own network magic,
own P2P port 8338, own address prefixes. You mine solo against your own node
and the whole block reward goes to your own address. It ran SHA-256d until
height 19,500 and has been yespower 1.0 since 14 November 2026.

And yes, I intend to keep it a Core fork and keep following upstream releases
rather than drifting away from them. The consensus differences are deliberately
small and listed in the source. Keeping validation recognisably Bitcoin's is a
feature, not laziness: anyone who can audit Core can audit this, and I would
rather inherit review from thousands of people than ask you to trust mine.

Parameters, a known-answer test vector to check your build against, and the
exact command line are in the yespower thread, which is the better place for
miner detail:

Code:
yespower parameters  https://bitcointalk.org/index.php?topic=5600188
Mining guide         https://bitcointalk.org/index.php?topic=5599204
Downloads            https://varve.mine.bz/downloads

- Masashi Takamoto
takamoto_bigcoin (OP)
Newbie
*
Offline

Activity: 16
Merit: 0


View Profile
November 21, 2026, 11:34:16 PM
 #20

v1.11: the first binaries that call the unit VARV

Also in this post: the swap toolkit published as v1.10, which never had an
announcement of its own, and a second place to read the source


This project was renamed from BigCoin to Varve on 20 November, and the source
has said Varve ever since. No compiled program did. v1.10 added a toolkit and
rebuilt nothing, so every binary published before this one labels amounts BIG.
These are the first that do not.

That is the whole of the release. No consensus rule, no network rule, no
default and no RPC method changed.
Every release from v1.7 onward accepts
exactly the same blocks and sits on the same network, and all of them stay
published. If your wallet saying BIG does not bother you, there is nothing here
you have to install.

Links and digests are at the foot of this post, as always.

What is in it, and what is not

Code:
Wallet amounts        BIG -> VARV, mBIG -> mVARV   the visible change
bigcoin-cli -version  Varve Core RPC client version v31.1.11
Node user agent       /Satoshi:31.1.11/            was /Satoshi:31.1.9/
Documentation         English only; three translations in doc/ are gone
Everything else       unchanged

There is no new feature here and no fix anybody is waiting on. The diff from
v1.10 is 43 files, and with one exception every line of it is text a human
reads: comments, log messages, the help in the two mining scripts, the readme,
three deleted translation files, and a screenshot retaken from this build. The
exception is the string that names the unit, which is one line of code:

Code:
-const std::string CURRENCY_UNIT = "BIG";
+const std::string CURRENCY_UNIT = "VARV";

That string is what the wallet prints next to an amount, and what some RPC
help text and error messages spell out. No RPC method, argument or result
changed shape, and no amount anywhere is a different number.

Upgrading without resyncing

Worth spelling out, because the folder is named differently now and the easy
mistake costs you a full resync.

The release archives unpack into Varve-win64 and varve, where
v1.9's unpacked into BigCoin-win64 and bigcoin. The bundled mining scripts
keep the chain and the wallet in a data folder next to themselves, so a
fresh folder means a fresh empty chain:

Code:
Windows   stop the node first:  bigcoin-cli.exe -datadir=data stop
          then copy the .exe and .dll files out of Varve-win64\ into the
          folder you are already running, keeping your existing data\
          (or unpack the zip and move your data\ folder into Varve-win64\)

Linux     stop the node, then the same thing: the binaries are new, the
          data directory is yours and does not move

Your wallet file is in data\wallets. Nothing in this release touches it
and nothing reads it differently. There is no migration step, no reindex and no
database change: the same data directory works under v1.9 and v1.11, in both
directions.

The label is the only thing about your coins that changed, and it changed
nothing


Balances, transaction history, addresses, the data directory and bitcoin.conf
are untouched. No coin moved when the name on the amount changed.

Addresses are still big1... and B... and they are staying that way.
Renaming an address prefix would make every address already written down, in
every file and every message, unusable. That is not a trade worth making for
tidiness, and I am not making it.

What was said in advance

When I announced the rename I wrote that if the unit label ever moved, the old
one would keep being accepted indefinitely and it would be announced before it
shipped rather than after. The first half is kept exactly, and more simply than
I expected: no part of this software has ever taken a unit name as input. RPC
amounts are numbers, addresses did not change, and the label exists only where
the wallet prints it. There is nothing that could stop accepting anything.

The second half, honestly: the specifications table on the site has said since
20 November that the label changes in the next release, and the build landed on
the 21st, so the notice was published before the binaries were -- but not here
first, and it should have been. None of this is a consensus matter. The
30-day promise is about changes a node validates; it still stands, and
this is not one.

Three strings keep the old name, permanently

Code:
Network magic          BIGC
Genesis coinbase text  BigCoin genesis 21/Oct/2026
yespower pers string   BigCoin/yespower1.0

These are deliberately outside the rename and will stay outside it. The first
is how a node recognises its own network, the second is inside the block that
every node hashes and checks, and the third is an input to the proof-of-work
function itself -- it is hashed into every block from the activation height
onward, so editing it would reject the entire chain rather than relabel it. The
source says so where the string is defined, in a comment put there to stop a
future tidy-up: it is not prose, and a rename must never touch it.

The file names on the downloads page follow the same rule in the other
direction: BigCoin-win64-v1.9.zip and the rest keep their names and their
digests forever, because those digests are printed in this thread next to
those names, and renaming a published file invalidates every check anyone has
already made against it.

One consequence of that rule is worth stating before somebody hits it. The
standalone mine-bigcoin-v1.8.cmd on the downloads page is unchanged and keeps
its digest. The copy of that script inside the v1.11 zip is not the same bytes
any more: it prints Varve and VARV where the standalone one prints BigCoin and
BIG. Nothing else about it differs, and I would rather say so than publish a
fifth standalone file for a difference nobody has to care about.

There was never a 31.1.10: v1.10 published no program, and the release number
and the program version have been the same number since v1.7, so the gap stays
unused rather than breaking that.

English only from here

The three Japanese translations in doc/ and the Japanese page on the website
are gone. A translation that quietly disagrees with the English it was made
from is worse than no translation, and those had started to drift. The readme's
wallet screenshot was retaken from this build, so the window in the picture
says VARV like the one you will be looking at.

v1.10: swapping coins for BTC with nobody holding either side

This went up on 20 November and never got a post of its own. It is seven Python
files, it adds no node and no consensus rule, and it is inside the v1.11 source
archives at contrib/bigcoin/swap/ -- there is no separate download and nothing
to install twice. The two v1.10 archives stay published, with the digests at
the foot of this post.

Ground rules first, because they matter more here than in any other release:
nothing is for sale. There is no market in this thread, no price, no
rate and no price talk, I am not a counterparty, and every amount in the
documentation is a placeholder. No exchange listings are sought.

What it does. Each side pays into one output on their own chain, and
that output can be spent two ways:

Code:
claim   whoever knows a 32-byte secret, and holds the claim key
refund  the person who funded it, but not before a deadline

Both chains use the same secret, so the moment one side claims, the
secret is sitting in a public block and the other side can claim too. If
anybody walks away, both deadlines pass and both sides recover their own coins.
At no point does either party hold the other's money, and there is nobody to
run off with it, because there is no third party in the protocol at all. The
contract is one miniscript descriptor:

Code:
wsh(or_i(and_v(v:pk(CLAIM),sha256(H)),and_v(v:pk(REFUND),after(T))))

That is a plain P2WSH output. Your counterparty does not have to trust this
program's description of what they are paying into -- the script is visible on
a block explorer, including one that is not mine. No consensus change was
needed: CHECKLOCKTIMEVERIFY and CHECKSEQUENCEVERIFY have been buried since
height 1 and segwit has been active from height 0.

No wallet passphrase, ever. Swap keys are generated locally and live
only in your swap file. Signing goes through descriptorprocesspsbt, which takes
a private key inline in the descriptor, so your wallet is never unlocked and
nothing is imported. The flip side, stated plainly: the swap file is a
bearer secret.
Between funding and settling, whoever has that file can move
the money. Back it up somewhere you control, and delete it when you are done.

What you need: Python 3.8 or newer, standard library only; a Varve node
and a Bitcoin Core node, both with -txindex=1. That flag is not optional -- the
tool fills in each input's witness_utxo from the node, and once an output has
been spent the UTXO set no longer knows anything about it.

Verify it before you use it. Every module checks itself when you run it
directly, against published test vectors and the BIP380 checksum vector. Then
there is an end-to-end rehearsal that starts two throwaway private chains, runs
a complete swap through both the claim path and the refund path, and checks 28
things about it -- including that a party without the secret cannot claim, that
a swap file with an address edited into it is refused rather than funded, and
that a refund is refused before its deadline and succeeds after it. It uses its
own temporary data directories and ports and touches nothing of yours:

Code:
python keys.py
python htlc.py
python psbtfield.py
python varveswap.py selftest
python rehearse.py --bigcoind /path/to/bigcoind

The Bitcoin half was tested against a stock Bitcoin Core. Both chains in
the rehearsal are Varve regtest chains, so that script alone cannot answer the
question this release rests on: does an unmodified Bitcoin Core sign and
finalise our contract? It was tested separately against a released Bitcoin Core
32.1 binary -- unmodified, signature-verified, reporting /Satoshi:32.1.0/ --
with Varve regtest on one side and Bitcoin regtest on the other: 45 checks,
both the claim path and the refund path.
Core parses the descriptor and
reports it solvable, derives the same address the swap file recorded, signs and
finalises the claim from the preimage carried in the PSBT with no patch and no
custom build, and refuses the time-locked refund before the deadline while
citing the chain's median time.

Every spend in that run also had to cross a P2P connection and be accepted by a
second node that holds no wallet and has never seen the descriptor. That is
worth more than the broadcasting node's opinion of its own transaction, because
-acceptnonstdtxn defaults to 0 in Bitcoin Core 32.1 on every chain,
regtest included: the node passing judgment is applying mainnet's relay policy.

What is still untested: mainnet, on either chain, and one thing no test
network can supply -- a live fee market. Fees in the rehearsals come from
-fallbackfee, so nothing here has paid a fee anyone competed for.
Treat your first mainnet swap as the experiment it is, and make it a small
one.


Put Varve on the initiator's leg. Either direction works and the tool
asks each node what chain it is on rather than assuming, but the six-hour gap
between the two deadlines is checked once against the wall clock, while the
deadlines themselves are enforced by each chain's median time past, which
trails the wall clock by roughly five and a half block intervals. Measured on
20 November, Varve's median time past was 1h43m behind the wall clock where
Bitcoin's sits near an hour, and Varve's difficulty takes days to answer a fall
in hashrate -- the ASERT half-life is two days -- so intervals stay long and
the lag stays wide. At an average interval above about 65 minutes the whole six
hours is gone. Put Varve on the initiator's leg and that lag works for the
responder instead of against them. It is also the only arrangement the
rehearsal tests.

Risk, stated plainly. A reorganisation deeper than the confirmations you
waited for breaks this, and on a chain this size that is cheap to buy. One
ordinary 8-core desktop does about 7,000 H/s, and this network has never run at
more than a few multiples of that. Three or four machines like that outrun
everything else on it combined, which is a few hours of rented cloud CPU. So
waiting for confirmations here buys you exactly that much and no more: six
confirmations is not six confirmations' worth of security, it is about an hour
of somebody's rental bill. Size what you trade to what you would tolerate
losing outright. That is a property of a young proof-of-work chain and not
something a tool can fix, and it works in your favour on the Bitcoin leg and
not at all on the Varve leg -- whoever is receiving coins on this chain carries
it.

The other things that can go wrong:

- You miss your own claim deadline. Then the counterparty refunds and
  you do not. Claim early; do not leave it to the last block.
- You lose the swap file between funding and settling. The money is then
  unspendable by anyone, forever.
- You fund before checking the fingerprint. Both sides print the same
  fingerprint of the agreement before anyone pays; if the two differ, stop. Do
  not fund. One changed character changes the fingerprint.
- Either node is not fully synced. A node behind the chain reports a
  healthy-looking depth for a transaction the rest of the network has not seen.
- This coin has no promised value and no market. There is no exchange,
  no market maker and no buyer of last resort. If you acquire some and later
  want out, your only exit is another private trade with someone who wants in,
  and there may not be one. Anything you pay could be a total loss, and nobody
  needs to buy any: mining is open to everyone.

And what the toolkit is not:

Code:
not a service  no server, no website, no API, no operator
not custody    no key generated here ever leaves your machine
not anonymous  both legs are ordinary on-chain transactions, and the claim
               publishes the secret that links the two

One detail for anyone reading the files at tag v1.11: they are the v1.10 files
with one change, the readme's examples now spelling the unit VARV where they
said BIG. The tool's own help still shows BIG. Both work -- a leg label is a
short name you choose when the swap is proposed, and the tool takes either.

Source, and a second place to read it

Code:
git clone https://varve.mine.bz/git/bigcoin.git
cd bigcoin
git diff --stat v1.10 v1.11

That prints the whole of this release: 43 files changed. To review the fork
against Bitcoin Core rather than just this release, the comparison from the
earlier posts, with the tag moved on:

Code:
git remote add upstream https://github.com/bitcoin/bitcoin.git
git fetch --depth=1 upstream tag v31.1
git diff --ignore-cr-at-eol --stat v31.1 v1.11

59 files. The flag matters because the source archives are packed on Windows,
so 14 files that upstream keeps as CRLF are LF in this repository; without it
the same command says 73 and the extra 14 differ by nothing but line endings.
VERIFYING.md inside the archives spells that out.

Tag v1.11 holds exactly the published archives, file for file. Its history
continues from the v1.10 commit, so no published history moved and every tag
this repository has already served still resolves to the object it always did.

The source now has a second home, because one host is one host:

Code:
https://codeberg.org/takamoto/varve

It is a mirror and not a second source of truth. The repository above, on the
same server as the downloads, is the one I push from; the mirror never pushes
back, and the clone URL that has always been published is unchanged and is not
going away. The reason it exists is simple enough: if that one server is down,
unreachable from where you are, or filtered by something between you and it,
the source is still readable somewhere.

A mirror you cannot check is worth very little, so here is how to check this
one. A commit id pins the entire tree, so two lines settle it:

Code:
git ls-remote https://codeberg.org/takamoto/varve.git refs/heads/main refs/tags/v1.11

6ac9e40b9140cd04e06fe03db70b1599854bd15b  refs/heads/main
1d118d0b9bc1a4e6ccbf0a8fe22f20faabf4502f  refs/tags/v1.11

If those match what is printed here, the mirror is this repository, down to the
byte. If they ever do not, believe this post and the digests below -- they are
on a forum I do not host, which is the whole point of printing them here.

On the software

Any future consensus change still gets at least 30 days between the notice and
the activation height, and the height goes in the release notes and in this
thread. This release is not one: it changes nothing a node validates.

The coin has no promised value. The only way to get it is to mine it.

Links and digests

Code:
Block explorer   https://varve.mine.bz/
Downloads        https://varve.mine.bz/downloads
Mining status    https://varve.mine.bz/mining/status
Faucet           https://varve.mine.bz/faucet
Running a node   https://varve.mine.bz/node
Source           git clone https://varve.mine.bz/git/bigcoin.git
Source (mirror)  https://codeberg.org/takamoto/varve

The same things by IP, unchanged and permanent:
  http://133.242.139.204/
  http://133.242.139.204/downloads
  http://133.242.139.204/mining/status
  http://133.242.139.204/faucet
  http://133.242.139.204/node
  git clone http://133.242.139.204/git/bigcoin.git

Seed node        133.242.139.204:8338
Seed node (Tor)  rjxywola2kupk4rp67qzidqiuxwvr76xmri5zimt6ecrvhx4xv5t7wyd.onion:8338
ANN thread       https://bitcointalk.org/index.php?topic=5598514
Mining guide     https://bitcointalk.org/index.php?topic=5599204
yespower thread  https://bitcointalk.org/index.php?topic=5600188

v1.11, published 21 November 2026:
347a8f51d7b1770e8d2d82b8e99d80f82e432f3258ff159b7f8595719e589788  Varve-win64-v1.11.zip
40c90c692431ec0fb45d62a102340dd23501ee4219c446b6753479e7482ccae1  varve-linux-x64-v1.11.tar.gz
4b76899fb8be633e60d4621917b9c82a2112a92e96343c21865216bf99cbdd8c  varve-source-v1.11.zip
0f705a4eec25b49919ab98f8f422adaf724fd7d3ddbb07a177abd1a5689af5d2  varve-source-v1.11-unix.tar.gz

v1.10, the swap toolkit, published 20 November 2026:
6bd8fa23735f799ad2444aebf9224bafa90b48f09073922b318fabc04a304641  varve-swap-v1.10.zip
c31d74e4637e940d0df6fabe0a495f4823c14a200392936ab504c13745de1f72  varve-swap-v1.10.tar.gz

v1.9, still published at the same URLs with the same digests:
f3999cc9d6da8a032a4ba8d2023a098e18dfeabe90e0369693b451415c5fd5cb  BigCoin-win64-v1.9.zip
cf85ed4f4b158e3c8c292052025252182c8f80d037a8419f5ef6bea5540b5ba8  bigcoin-linux-x64-v1.9.tar.gz
d287776d917d0d22c063ffbc6c6ed91bba3fe96b8c9f385a594a2de1caf790bb  bigcoin-source-v1.9.zip
d44066c8bf5d70404b66d21d6689af4d99a13695b1bbf047816c47cc6b99c693  bigcoin-source-v1.9-unix.tar.gz
b4973f0c5c6a6f59c276f899a21a7453a60e35e3b916d222e55fa27647c4bf26  mine-bigcoin-v1.8.cmd
67d55214e84a5571358c09d2c875f0b49d94a08307ac32e93e79cc56fec574dd  mine-bigcoin-v1.8.sh

These digests are published here as well as on the downloads page on purpose:
this thread is not hosted on my server, so if that server were ever compromised
the attacker could rewrite an archive and the hash page together, but not this
post. Nothing published has been replaced, and the rename did not replace
anything either.

- Masashi Takamoto
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!