The Zcash ArboretumI–IX The Complete Arboretum
Contents
PDF
Theme

1 Scope, sources, and the two double-spends

This volume gives a quantitative account of Nakamoto consensus (proof of work, the most-work rule, and the probabilistic finality they provide), the mechanism that selects the ledger the protocol layers above it assume. The analysis follows Meni Rosenfeld’s Analysis of hashrate-based double-spending (11 December 2012; latest version 12 February 2014; arXiv:1402.2009), an exact counterpart to the approximate analysis in Satoshi Nakamoto’s whitepaper; we reprove its results in full, recompute its tables by script, and instantiate its conclusions for the proposed NU7 parameters.

The term “double-spending” names two distinct attacks, of which this volume treats one. Two spends of the same shielded note reveal the same nullifier, and the protocol specification’s §“Nullifier Sets” forbids a repeated nullifier within a transaction or valid block chain; such a same-chain attempt is therefore invalid under the consensus rules. Consensus must prevent the chain-level attack: the attacker pays a merchant on the public chain, then replaces that chain with another in which the payment does not occur. No zero-knowledge proof (Crypto Guide, §“Interactive proofs, zero knowledge, and SNARKs”) excludes this attack; under the model of this volume, the proof-of-work race bounds its probability of success.

The analysis assumes the Math Guide’s probability section (§“Probability, asymptotics, and computation”), which constructs probability from its definition (finite spaces, conditioning, independence, random variables, expectation) and makes the three extensions used here: countable additivity, continuity along monotone events, and density-defined laws (§“Beyond finite spaces: countable additivity, limits, and densities”). The block and anchor vocabulary uses the Crypto Guide’s hash and commitment sections. Section 3 constructs the race-specific instruments that no lower volume constructs: memoryless clocks, random walks, and the negative binomial law; the algebra used is elementary. The results can be read from Sections 2, 6, and 8 alone, without the proofs.

1.1 Sources and their standing

Rosenfeld’s analysis supplies the probability and economic model, not consensus rules; its two printed tables are reproduced by exact arithmetic before the model is instantiated for Zcash. The protocol specification and the relevant ZIPs define the protocol.

This volume targets the draft NU7 rules as of 6 October 2026: ZIP-259 selects ZIP-218, ZIP-235, ZIP-237, ZIP-2003, Revision 2 of ZIP-207, Revision 3 of ZIP-214, and ZIP-2008. The activation height is assigned on Testnet, the test network, but not on Mainnet, the production network, so the Mainnet heights derived from it are unassigned. These gaps are stated where they affect a calculation; no unassigned constant is treated as a deployed rule.

Remark 1.1 (Defects of the arXiv text).

The arXiv v1 PDF of the paper contains exactly two citations, both rendered as “[?]” (in the abstract and in §1), and no References section. Both refer, by context, to the Nakamoto whitepaper, and we read them so. The paper numbers only two of its displayed equations (its equation (1) is our Theorem 5.2, its equation (2) our equation (3)) and sets its two model assumptions as a bare numbered list; the Assumption environments of Section 4 are ours.