On this page
A community-run summary of what is happening across the Neptune Cash ecosystem.
In brief
- Succinctness is running on testnet. It lets a node verify the whole chain from the latest block alone. The pull request, opened on 21 October, implements it as a soft fork, and the fork activated on testnet-0 at block 6,420 the same day.
- One payout destination has dominated mining since mid-October. It took more than half of the blocks mined from 22 to 25 October, and by then the pools that used to find most blocks found under 5%.
- Two node releases harden the network. neptune-core v0.19.1 and v0.19.2 bring nine security fixes between them, most of them limits on what a node accepts from its peers. Neither is marked mandatory.
Releases and security
neptune-core v0.19.1 (14 October) speeds up block propagation: a node now announces a new block with a small proof that it meets the mining target, so peers can check it before downloading the block. It brings seven security fixes, mostly limits on what a node accepts from its peers, and its pull request credits three to an outside reporter.
v0.19.2 (16 October) acts on the findings of an audit of v0.17.1 that were “worth acting on”, according to its pull request, which promises “No consensus change. No p2p protocol changes.” It also caps how many blocks a node serves per minute, and rejects a bad block from its header before spending time on its proof. Neither release carries upgrade notes or is marked mandatory.
Succinctness
Every block proves the chain behind it. The code is in pull request #988, “making the blockchain into a proofchain”. Each block must prove that its parent was valid and met the proof-of-work target, so the newest block proves the whole chain (explained below). It must also prove that the mutator set, Neptune’s private record of which coins exist and which are spent, was updated correctly, and that whoever built the block holds a full copy of it. As written, nodes that do not upgrade still accept the new blocks, which makes it a soft fork, “but old nodes cannot mine”. Whether it ships alone or together with smart-contract changes that need a hard fork has not been settled.
Composers carry most of the cost. Neptune splits mining into two jobs: composers build and prove blocks, and guessers search for the winning number with GPUs. Composers would do “about four times as much” proving work, according to the pull request. Under today’s rules, Lentern, which composes blocks for solo guessers, has brought block-proposal time down to 39 seconds, according to the dev update. Guessers keep the same hashing algorithm but must reveal a piece of the mutator set meant to show they hold a full copy, “which all current nodes do”.
No reward for leaving transactions out. Proving the mutator-set update costs more with every input, which would tempt composers to skip transactions. So every block must do a minimum amount of that work anyway, which means “the first 152 inputs come at zero marginal proving cost for the composer”. The pull request states the aim plainly: “To avoid the incentive to mine empty blocks.”
One gap remains. Each spent coin still carries copies of the parts of the mutator set it touches, so nodes can update their records. The new proofs do not cover this data, so only nodes doing full validation check it, and it is bulky: an 8 MB block fits only about 150 to 200 spent coins. A later hard fork is expected to remove it, which Bulletin #6 said should allow 5,000 to 10,000 per block.
New nodes could fetch the mutator set from a peer. A follow-up pull request, opened on 23 October, would let a node fetch the full mutator set from a peer, paying for it with proof of work. On mainnet that set is 72 MB today, of which about 40 MB needs to be sent, according to the pull request.
Activation. Testnet-0 activated at block 6,420 on 21 October. The branch sets mainnet block 61,000, about 1,300 blocks beyond the tip on 25 October; at the current pace of a block every nine and a half minutes, that would be around 3 November. On the forum Thorkil Værge, one of Neptune’s two founders, called the height “still in flux” and suggested ten days’ notice, so with no release out yet, 3 November looks likely to slip. He also warned: “You shouldn’t upgrade your main net node before this PR has been merged and a new release has been made.” A founders’ blog post explaining the fork is drafted and waits on that release.
A wallet that is also a full node
On 21 October, in the Telegram group, Thorkil sketched where succinctness, with more smart-contract groundwork, could lead. A program downloads and verifies the latest block in under a second, gives you an address, and once funded opens a payment channel on its own, taking one or two blocks. From then on it sends NPT or other assets anywhere instantly, while acting as a full node.
He named two open problems. One is the on-ramp, how people buy their first NPT; members suggested swap services. The other is a wallet that is offline when the other side closes a channel unfairly. He thinks this “has been largely solved on the Bitcoin lightning network”, where a delay on closing gives the wronged party time to respond, and watchtowers can respond for a user who is away.
Mining
Most blocks now pay one destination. Before the period, three pools found about nine blocks in ten. From about 13 October a single guesser digest took over. The digest is the fingerprint each block records of where the guesser’s reward goes, and this one appeared on more than half of the blocks mined from 22 to 25 October, according to block-by-block attribution. Dozens of smaller digests found most of the rest. The old pools found under 5%, though the 19 October dev update still counted NPTmining.com at nearly 15%. One digest could be one machine or many.
Empty blocks kept appearing. For several days many blocks carried nothing but the miner’s reward. A core developer said in the Telegram group on 19 October that some were mined “while fee-paying transactions were in the mempool”, and on 21 October that the problem “seems to have been fixed completely”. Such blocks still appeared from 21 to 23 October, from the large miner and from smaller ones. Whether transactions were waiting each time cannot be told from the chain. PR #988’s minimum-work rule is meant to end the incentive, but only once the fork activates.
Project and community
CoinGecko now shows NPT’s circulating supply, after a community member’s correction request on 12 October, so the listing no longer reports a market cap of zero. It still presents NPT as a token on the Base network: that is the wrapped wNPT contract, not the native coin, and the correction is still to be resolved. A tracker that shows a Base contract address for NPT is describing the wrapped token.
$NPT is now a cashtag on X, added after a request from another community member. Typing $NPT in a post now offers Neptune Cash, with its logo and market data, among the suggestions. The ticker is shared: Texxon Holding, listed on Nasdaq, also trades as NPT and comes first in the list, so pick the entry marked Crypto.
Elsewhere
Pirate Chain rolled back an attack that inflated its supply by a fifth. On 14 October two validation flaws let someone create about 39.8 million ARRR and move it into the shielded pool, where it could not be traced. Pirate said on 22 October that the chain had been rolled back to just before the exploit, after first announcing and then withdrawing a rollback. The flaws were not in the zero-knowledge proofs: the fix notes describe coins imported from other chains without proof that they had been burned there. It is the third privacy coin since August, after Firo and Zano, to disclose a flaw that could create coins.
New proposals from Microsoft Research and Vitalik Buterin rely on hash-based recursion. On 15 October Microsoft Research introduced Magnetar, a zkVM, or system for proving that any program ran correctly. In its announcement’s words, it needs “no elliptic curves, no lattices, no special-purpose hash functions, no trusted setup”. On 19 October Vitalik Buterin argued for EIP-8288, which would fold Ethereum users’ signatures and proofs into a single STARK. Neptune already merges transaction proofs into one proof per block, though its Tip5 hash is one of the special-purpose kind Magnetar avoids.
Germany sets deadlines for moving to post-quantum cryptography, and Coinbase proposes ideas for Bitcoin’s move. On 12 October Germany’s government and its cybersecurity agency, the BSI, issued a national guideline: high-risk uses should move to quantum-safe methods by the end of 2030, and the whole transition should be largely complete by 2035. A day later Coinbase set out seven points for moving Bitcoin to post-quantum signatures. Neptune’s consensus and spending already rest on hashes, so it has no such migration ahead, apart from the encryption of payment details covered in Bulletin #6.
Starknet votes to study leaving Ethereum. Starknet, a network on top of Ethereum that already uses hash-based STARK proofs, closed a vote on 22 October in which 93.52% of voting power backed researching a quantum-resistant chain of its own. StarkWare, the company behind it, names Ethereum’s bridge and blob-data cryptography as the parts a quantum computer could break. The vote is non-binding.
Explainer: how one proof can cover a whole chain
To know which block is really the latest, a Bitcoin node downloads every block since 2009 and checks each one. Neptune nodes do much the same today.
Checking a block is a program: it takes a block and its parent and answers valid or invalid. Neptune already uses Triton VM to prove that programs ran honestly, so that check can be proved too. The trick is to make each block’s proof also verify its parent’s proof. The newest proof then covers its parent, whose proof covered the grandparent, and so on down the chain. Checking one proof covers every block it reaches.
A rival chain could carry valid proofs of its own, so each proof also checks that its parent met the proof-of-work target, and nodes still follow the chain with the most work. Because succinctness arrives on a running chain, the recursion starts at an activation height rather than at genesis; how nodes will treat the blocks before it has not yet been documented.
Worth doing
- Upgrade. Node operators and miners: neptune-core v0.19.2, for its networking and security fixes. Miners, composers especially: watch for the release that carries soft fork epsilon, and do not run the pull request on mainnet before it ships.
Corrections and additions welcome. Discussion in the Neptune Bulletin thread on the forum.