Bitcoin Kernel (@BitcoinKernel) header
Bitcoin Kernel (@BitcoinKernel) profile picture

Interpreting Bitcoin’s most important discussions — from protocol debates to dev-mailing-list insights.

Joined October 2025
Photo by Bitcoin Kernel (@BitcoinKernel)Photo by Bitcoin Kernel (@BitcoinKernel)Photo by Bitcoin Kernel (@BitcoinKernel)Photo by Bitcoin Kernel (@BitcoinKernel)Photo by Bitcoin Kernel (@BitcoinKernel)Photo by Bitcoin Kernel (@BitcoinKernel)Photo by Bitcoin Kernel (@BitcoinKernel)Photo by Bitcoin Kernel (@BitcoinKernel)Photo by Bitcoin Kernel (@BitcoinKernel)Photo by Bitcoin Kernel (@BitcoinKernel)
One line to close this discussion: The interesting question P2SKH raises is not whether to save 12 bytes — it is whether "hashed Taproot output key" is a coherent design primitive, and whether Bitcoin's output type evolution has room for it. This is @BitcoinKernel — tracking output type design, script path tradeoffs, and durable Bitcoin engineering judgments. Follow for more.
27
7/ Two questions this proposal did not resolve: BIP360 (P2QRH, post-quantum outputs) claims witness version 2. P2SKH also targets version 2. One must move. sashabeton suggested version 3. No consensus formed. hash160 provides 80-bit second-preimage resistance. This has been acceptable since P2PKH in 2009. Whether it is acceptable in a new consensus rule is an open question.
1
196
6/ waxwing (AdamISZ) raised pubkey prefixing — then retracted. P2SKH does prefix: the challenge hash includes hash160(P.x). This satisfies the security requirement. His underlying point stands independently: BIP340 chose H(R, P, m) over H(R, m) to enable MuSig and CISA. Aggregation schemes require pubkey commitment at the protocol level. P2SKH does not break this. It was not the issue here.
1
24
5/ Martin's anonset argument, stated precisely: Taproot was designed to merge single-key and script users into one indistinguishable output format. Every new output type fractures that set. 12 bytes does not justify the fracture. Counter-observation: P2WPKH introduced a new format distinct from P2PKH. It is now the most widely deployed output type on the network. The anonset argument, applied consistently, rules out every output type transition retroactively. The disagreement is not technical. It is a judgment about migration strategy.
1
15
4/ The script path objection has a structural fix not raised in the thread. P2SKH commits to the raw key P. P2TR commits to the tweaked key Q = P + hash(P ‖ merkle_root)·G. If P2SKH committed to Q instead: scriptPubKey = OP_2 <hash160(x(Q))> The Taproot commitment structure is fully preserved. Script path spending works identically to P2TR. Key recovery operates on Q rather than P. The 12-byte saving is retained. As one observer noted: the loss of script path is not inherent to hashed-key outputs. It comes from hashing the wrong object.
1
20
3/ Three objections were raised on bitcoindev. Alex: "You removed all scriptability and opcode upgradeability for 12 bytes." Martin: "This competes with Taproot. Forcing everyone onto P2TR is an intentional privacy design." Saint Wenhao: "With DER signature grinding, P2WPKH can already beat 64 bytes. Schnorr has no equivalent path below 64." sashabeton withdrew the proposal after seven days.
1
20
2/ P2SKH's proposed construction: scriptPubKey = OP_2 <hash160(x(P))> — 22 bytes witness = [<sig>] — 64 bytes, no pubkey Verification uses key recovery: e = TaggedHash("P2SKH/challenge", rx ‖ hash160(P.x) ‖ msg) P = e⁻¹ · (s·G − R) Then: hash160(x(P)) == program The pubkey is never stored on-chain. It is derived at spend time from the signature itself.
1
31
1/ The problem sashabeton identified is real: P2WPKH: 22-byte scriptPubKey. Witness carries full 33-byte pubkey. ~108 witness bytes per input. P2TR: 64-byte witness. scriptPubKey carries full 32-byte x-only pubkey. 34-byte output. Neither achieves both a compact output and a compact witness. No existing single-key type does.
1
29
🧵 Thread: Pay to Schnorr Key Hash (P2SKH) — A New Output Type Proposal This is not a debate about byte counts. It is about whether Bitcoin's output type design space is closed — or still open.
1
1
1
95
🧵 End. Follow @BitcoinKernel for real-time Bitcoin dev updates.
115
5/ What this debate is really about •UTXO growth is a real engineering concern •But consensus is not a cleanup mechanism •Once spendability depends on perceived “acceptable use” → Bitcoin’s neutrality is weakened Bitcoin’s security model depends as much on restraint as it does on code.
1
137
4/ Why confiscation keeps coming up Opponents argue: •Even dust outputs are valid UTXOs •Making them unspendable via consensus •Is effectively protocol-level confiscation Supporters argue it is just deleting abandoned state. This is not a technical disagreement — it is a property-rights boundary.
1
85
3/ Consensus rules vs implementation details Important distinction: •Not storing provably unspendable outputs → implementation detail (no BIP required) •Changing whether an output is consensus-spendable → hard protocol rule Bitcoin Core already drops OP_RETURN outputs from the UTXO set. The Cat goes further than that.
1
116
2/ Core objection (engineering, not preference) As argued by Greg Maxwell: •Dust thresholds are not consensus parameters, only policy heuristics •“Economically irrational to spend” ≠ “protocol-level unspendable” •Deleting UTXOs does not delete tokens; indexers can trivially adapt Conclusion: this encodes subjective judgments into consensus.
1
153
1/ What does The Cat propose? •Classify certain dust UTXOs as “non-monetary” •Mark them permanently unspendable via consensus rules •Remove them from the UTXO set •Claimed benefit: a one-time ~30–50% reduction in UTXO storage Key detail: classification relies on external indexer logic (e.g. Ordinals / Stamps).
1
198
📜 Thread: “The Cat” BIP: A concrete debate about UTXO cleanup, consensus boundaries, and Bitcoin neutrality A recent bitcoindev discussion around the The Cat BIP draft is not really about NFTs or spam. It is about what kinds of judgments are allowed at the consensus layer.
3
1
2
5
30,837
7/ One line to close this debate: Bitcoin isn’t secure because no one can see it — it’s secure because anyone can verify it. When discussions return to protocol definitions, much of the fear fades. This is @BitcoinKernel — tracking protocol disputes, engineering judgments, and durable Bitcoin truths. Follow for more.
1
58
6/ Grok further noted that encryption appears only in ancillary layers, such as: local wallet file encryption node-to-node communication security These sit outside consensus and do not define Bitcoin’s security model.
1
1
69
5/ During the discussion, some asked Grok directly: Where does Bitcoin actually use encryption? Grok’s answer was clear: Bitcoin’s core protocol uses cryptographic signatures for verification, not encryption. Transactions and blocks are public.
1
37
4/ That’s why “quantum decrypting Bitcoin” is the wrong threat model. The theoretical issue is signature unforgeability, not reading the ledger. Bitcoin’s resilience comes from public data, public rules, and upgradeable verification.
1
45