All posts
Bulletin No. 8 7 min read

Neptune Bulletin #8

Succinctness is merged, released and hours from activating at block 62,500, but the software that uses the new proofs is not written yet. Covering 26 October to 12 November 2026.

On this page
  1. In brief
  2. Succinctness
  3. What comes next
  4. Mining
  5. Project and community
  6. Elsewhere
  7. Explainer: succinct for whom?
  8. Worth doing

A community-run summary of what is happening across the Neptune Cash ecosystem.


In brief

  • Succinctness is merged, released and about to activate. Soft fork epsilon was merged on 4 November, shipped in neptune-core v0.20.1 on 6 November, and activates at mainnet block 62,500, expected in the early hours of 13 November UTC. Composers, the miners who assemble and prove blocks, must upgrade, and so must exchanges; holders need do nothing.
  • The chain gets its proofs before the software learns to use them. v0.20.1 still downloads and checks every block. Syncing from the latest block plus the mutator set, Neptune’s record of coins created and spent, is an issue opened on 12 November.
  • The activation height moved three times, settling at 62,500. One move followed a late audit report, whose findings were fixed before the merge; the last gave composers a week to upgrade.

Succinctness

Succinctness means every block carries a proof that the block before it was valid, so the newest proof vouches for the chain behind it. It arrives as a soft fork named epsilon: a rule change that older software still accepts.

Merged and released. Pull request #988 was merged on 4 November and released two days later. v0.20.0 carries the fork; v0.20.1, the release with binaries, only repairs the macOS build, so its release page shows one build fix and the epsilon notes sit under 0.20.0 in the changelog. v0.20.0 also repairs the lookups nodes use to find new peers, which had been failing on mainnet.

Who must upgrade. The changelog says older versions accept the new blocks without enforcing the new rules, and that “composers must upgrade before the activation height”. The commit that set the final height adds that, strictly, only composers and exchanges need to, and that guessers, who do the proof-of-work on a composer’s block, should, or they risk wasting their work. The founders’ article, published on 6 November although its page is dated 22 October, tells holders that “nothing is required of you”. Nodes that stay on older versions keep following the chain with the most mining power behind it. Composers should expect between three and a half and four times the proving work, depending on the source.

The height moved from 61,000 to 62,500 in three steps, the second after what the forum announcement called a last-minute report and the last to give composers a week to upgrade. Thorkil Værge, one of Neptune’s two founders, said on the forum that the team checks the code regularly with AI models, and the issues found were fixed before the merge. At 20:11 UTC on 12 November the chain stood at block 62,454, 46 blocks short.

What it does not do yet. The 9 November dev update says the v0.20 releases “will maintain archival validation”, that is, keep checking every block. Issue #997, opened on 12 November, says neptune-core does not yet use the proofs: “Let’s fix that!” Its plan is to sync from the latest block and the mutator set, skip historical block proofs, cut downloads by at least a factor of 1,000 and finish in under a minute on most connections. The article’s line that upgrading lets a node verify the chain near-instantly describes that later software, not v0.20.1. It also gives the eventual saving: historical blocks take about 60 GB today, and about 2 GB without their proofs. Two parts already shipped: nodes can send each other the mutator set (v0.19.3), and can validate a block without its parent (v0.20.0).

Neptune is not the first succinct chain. Mina has carried a recursive proof of its chain since 2021, built on elliptic curves and with a public ledger. Parano1d, a transparent proof-of-work chain launched in August by a pseudonymous developer, with no audit that could be found, also verifies its history with recursive proofs and no trusted setup. Neptune appears to be the first to combine a recursively proven chain with transactions that are shielded by default.

What comes next

A hard fork named zeta. Bulletin #7 noted one gap: some input data is checked only by full validation, not inside the proofs. Issue #996, opened on 12 November, plans to move that data out of blocks; it says the data holds transaction throughput down by a factor of 100 to 1,000. It would also allow inputs in coinbase transactions, the transactions that pay out a block’s reward. The dev update, which calls the fork SATA, targets block 70,000, and a work-in-progress branch carries the same number. Unlike epsilon it is a hard fork: every node will have to upgrade.

DeFi groundwork. Draft pull request #980 grew to 62 commits. It implements standing orders to buy time-locked NPT, and adds vaults that lock NPT to issue “vaulted NPT” one to one; paying orders in vaulted NPT is designed but not yet built. It depends on a hard fork that allows coinbase inputs, which zeta plans, and none of it is in a release.

Mining

CPU guessing became competitive. v0.19.4 (28 October) made the built-in guesser about 17 times faster on a 12-core processor with AVX-512. On the forum Thorkil compared a CPU and a graphics card costing about 500 dollars each and found the card 3.2 times faster. The hashing library behind it, twenty-first, was relicensed on 26 October from GPL-2.0 to MIT or Apache-2.0, so miner authors can use its code in closed software.

The proof-of-work hash stays. A 6 November post commits the founders to keeping Tip5 as the guessing hash unless it is catastrophically broken, and says FPGA and ASIC mining, meaning purpose-built hardware, would be welcome. A member then claimed in the Telegram group that such hardware had already replaced graphics cards. Thorkil answered on the forum on 12 November: “I have no reason to believe that to be the case.” No public evidence of FPGA or ASIC miners was found.

Project and community

The forum was unusually busy: 13 new topics against two active ones last time, three dev updates in three weeks, and 130 registered users, up from 119.

CoinGecko and CoinMarketCap both list NPT in their quantum-resistant categories. Both still present it as a token on Base, which is the wrapped wNPT, and Coinranking’s list omits it.

Impersonators were reported twice more, one using a core developer’s name and photo and one calling a member as “GROUP MODERATOR”. Nobody legitimate will ask for your seed phrase.

Elsewhere

Iron Fish shut down after a counterfeiting bug. On 27 October the supply of its bridged token on Base rose from about 3.9 million to nearly 3 billion. The team traced the cause to its own chain: a padding rule in the note tree, dating from a 2019 prototype, let the owner of a note spend it a second time at a position that did not exist. The fix works only for a new chain (“This PR does not provide an upgrade path for existing chains.”), and on 30 October the team announced the end of support for the network, which had been in maintenance mode since April. Because positions are private, the post-mortem says chain data alone cannot attribute an amount. It does not say how the bug was found. It is the fourth privacy coin since August, after Firo, Zano and Pirate Chain, to disclose a flaw that created coins.

MEXC is dropping Monero and Dash. The exchange suspends XMR and DASH spot trading on 13 November and delists their perpetual futures the same day. Deposits are already closed, withdrawals stay open for 30 days, and the notice gives no reason. MEXC was one of a handful of large exchanges still listing Monero for spot trading.

A rival design measures itself against Neptune. ANNULAR, a proposal for post-quantum shielded payments built on isogenies, a different branch of post-quantum mathematics, calls Neptune Cash “the closest prior scheme” and credits it with resting on hash functions alone. It also records the costs: proving on a workstation with 768 GB of memory, and a stated 160-bit security level that, it notes, a contributor reads as 160 bits conjectured or 80 provable. Its source for that reading is a community proposal for machine-checked proofs, opened on neptune-core in October.

Explainer: succinct for whom?

“Succinct” can describe three different jobs, and the documentation separates them.

Validating the chain. This is what epsilon adds: each block proves its parent was valid, so the latest proof vouches for everything since activation.

Mining. Neptune “intentionally does not feature succinct mining”. If miners could build on a block knowing only its proof, one of them could publish a block while withholding the one before it. Everyone would accept the chain, but nobody could update the data that keeps their coins spendable. So composers must prove they hold the full mutator set.

Managing coins. Partly. A wallet keeps its coins spendable by following each new block. One that was offline for long, or caught in a reorganisation, still needs a peer that keeps history.

After activation, checking the chain can become cheap once software uses the proofs. Mining stays heavy by design, and holding coins still means keeping up with blocks.

Worth doing

  • Upgrade. Composers and exchanges: neptune-core v0.20.1 before block 62,500. Guessers and other node operators should follow. Holders need do nothing.

Corrections and additions welcome. Discussion in the Neptune Bulletin thread on the forum.