Merge pull request #2196 from fjahr/bip-t5-draft

BIP95: Testnet 5

README.mediawiki

@@ -548,6 +548,13 @@ users (see also: [https://en.bitcoin.it/wiki/Economic_majority economic majority
 | Specification
 | Deployed
 |-
+| [[bip-0095.md|95]]
+| Applications
+| Testnet 5
+| Pol Espinasa, Fabian Jahr
+| Specification
+| Draft
+|-
 | [[bip-0098.mediawiki|98]]
 | Consensus (soft fork)
 | Fast Merkle Trees

bip-0095.md

@@ -0,0 +1,183 @@
+```
+  BIP: 95
+  Layer: Applications
+  Title: Testnet 5
+  Authors: Pol Espinasa <polespinasa@protonmail.com>
+           Fabian Jahr <fjahr@protonmail.com>
+  Status: Draft
+  Type: Specification
+  Assigned: 2026-06-22
+  License: CC0-1.0
+  Discussion: 2026-06-02: https://groups.google.com/g/bitcoindev/c/kGUMTxOvdJA
+  Version: 0.1.0
+  Requires: 54
+  Replaces: 94
+```
+
+## Abstract
+
+Testnet 5 is a new test network intended to replace [Testnet 4][BIP94]. Testnet 5 removes the difficulty
+exception defined in Testnet 4. Sustained exploitation of this exception has made the network difficult
+to use for testing. Additionally, Testnet 5 raises the minimum difficulty and enforces the consensus
+rules specified in [BIP 54][BIP54] from block 1.
+
+## Motivation
+
+Testnet 4 included mitigations for an issue known as the [block storm attack][block-storms] which could render the
+whole network unusable. Block storm attacks led to a depletion of block subsidies, which made it hard to acquire
+coins for testing. However, Testnet 4 still retained a modified version of the difficulty exception rule.
+The difficulty exception permits a block to be mined at the minimum difficulty when more than 20 minutes
+have passed since the previous block. Testnet 4 kept this exception with the aim of allowing CPU users a
+limited path to acquire coins for testing, to mine non-standard transactions that other miners would not
+relay, and to keep the chain moving if a large source of hash power were to leave the network. Shortly
+after Testnet 4's introduction, the exception had been systematically and persistently exploited, which
+prevented the exception from achieving the intended
+goals. While block storms were prevented, the network suffered from constant re-orgs of small
+numbers of blocks due to multiple difficulty-exception blocks competing for the tip. This led
+to discussion about changing Testnet 4 to mitigate this issue (see [Bitcointalk][bitcointalk-thread]
+for analysis and discussion).
+
+In Testnet 5 there is no exception to the PoW rules. This appears to be the logical conclusion,
+since any such exception could be exploited by a motivated attacker. The lack of exception
+ensures the network’s behavior matches mainnet as closely as possible.
+
+BIP 54 is already enforced on signet through [Bitcoin Inquisition][signet-bip54] as of this BIP's creation.
+However, signet does not allow miners to test that their software reliably follows the rules of
+BIP 54. Testnet 5 provides a testing environment for this.
+
+## Specification
+
+Testnet 5 follows the same consensus rules as mainnet, except for the following two changes.
+
+### BIP 54 activation
+
+The rules specified in [BIP 54 version 1.0.0][BIP54] are active on Testnet 5 from block 1.
+
+#### Problem Statement
+
+BIP 54 proposes new consensus rules in order to fix several potential attack vectors. Namely 
+it prevents the timewarp attack, reduces the worst-case block validation time, mitigates Merkle
+tree weaknesses, and avoids duplicate transactions without [bip-0030][BIP30] validation.
+
+#### Rule Specification
+
+See specification in [bip-0054][BIP54].
+
+### Minimum Difficulty
+
+The proof-of-work limit (the maximum target) is `0x1a0fffff` in compact encoding (nBits),
+corresponding to a minimum difficulty of approximately 1,000,000.
+
+### Consensus Rules
+
+All consensus rules active on mainnet as of May 2026 are enforced from block 1, the most
+recent of these being the Taproot softfork. Additionally, the rules of BIP 54 are enforced
+from block 1, as specified above.
+
+### Genesis Block
+
+TODO: Mine the block. The values below are placeholders inherited from Testnet 4. They are
+the genesis block's header fields together with the `Message` and `Pubkey` of its coinbase
+transaction. Notes for the miner:
+
+- `Pubkey` is the public key in the coinbase transaction's pay-to-public-key (P2PK) output. Use a
+  recent Bitcoin mainnet block hash.[^pubkey-methodology]
+- `Message` is an arbitrary string in the coinbase input script, like the newspaper headline in
+  Bitcoin's mainnet genesis block. Testnet 4 used it to carry the anti-pre-mine commitment, but
+  Testnet 5 places that in the `Pubkey` instead, so the `Message` should be left empty, which also
+  avoids bike shedding over its content.
+- `Time stamp` time of mining or that of the block which hash is used in `Pubkey`
+- `nBits`: `0x1a0fffff`, per the raised minimum difficulty.
+- `Version: 1`
+
+Testnet 4 placeholders:
+
+> * Message: <code>03/May/2024 000000000000000000001ebd58c244970b3aa9d783bb001011fbe8ea8e98e00e</code>
+> * Pubkey: <code>000000000000000000000000000000000000000000000000000000000000000000</code>
+> * Time stamp: 1714777860
+> * Nonce: 393743547
+> * Difficulty: <code>0x1d00ffff</code>
+> * Version: 1
+> 
+> The resulting Genesis block hash is <code>00000000da84f2bafbbc53dee25a72ae507ff4914b867c565be350b0da8bf043</code>, and the block hex is <code>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</code>.
+
+### Message Start
+
+The message start (also known as the network magic) is defined as <code>0x46495645</code>. These four bytes spell `FIVE` when
+interpreted as ASCII.
+
+### P2P Port
+
+The default p2p port for Testnet 5 is `18335`.
+
+## Rationale
+
+Instead of starting a new Testnet, changing the rules of Testnet 4 [was considered][t4-discussion] as well. The decision
+for a new network had two main reasons:
+
+1. Deploying network changes is hard and typically takes a long time. This effort doesn't seem to be
+   worth it for a Testnet that has the alternative option to start fresh, especially considering that
+   the attackers exploiting Testnet 4's rules may use them to prevent a smooth deployment of new rules,
+   including those of BIP 54.
+2. Any fix to Testnet 4 that goes beyond a band-aid would require a hard fork, which brings its own
+   implementation cost: adding hard-fork support to clients, handling the p2p layer correctly, etc.
+   For a test network this isn't worth taking on, especially since a band-aid would only make
+   Testnet 4 diverge further from mainnet.
+
+Raising the minimum difficulty was dismissed for Testnet 4 because it would have prevented CPU
+miners from utilizing the difficulty exception. With the exception removed in Testnet 5, this
+argument no longer applies. A minimum difficulty of approximately 1,000,000 strikes a balance
+between dampening the quick mining of large numbers of blocks shortly after launch, before the
+difficulty has adjusted to the available hash power, and keeping a low barrier so that a handful
+of at-home miners or a single ASIC can keep the network usable.
+
+## Backward Compatibility
+
+Testnet 5's consensus rules are not compatible with those of Testnet 3 and Testnet 4. The
+consensus rules differ in both directions: Testnet 5 enforces the BIP 54 consensus rules from
+block 1, which is not the case for Testnet 3 or Testnet 4. Testnet 5 also does not apply the
+difficulty exception rule that Testnet 3 or Testnet 4 requires.
+
+Because each network has its own genesis block, their UTXO sets are disjoint and transactions
+can never be replayed across them. Address formats are shared, however, so the same address may
+be reused on different networks.
+
+Any implementation that intends to follow Testnet 5 must add the new network parameters and
+additionally enforce the BIP 54 rules while not permitting any of the difficulty exception rules.
+
+## Reference Implementation
+
+Pull request at ?
+
+## References
+
+- [BIP 30][BIP30]: Duplicate transactions
+- [BIP 54][BIP54]: Consensus Cleanup
+- [BIP 94][BIP94]: Testnet 4
+- [The block storms of Bitcoin's Testnet][block-storms]
+- [Testnet 4 difficulty exception analysis and discussion (Bitcointalk)][bitcointalk-thread]
+- [BIP 54 enforcement on signet via Bitcoin Inquisition][signet-bip54]
+- [Discussion on changing the Testnet 4 rules (Bitcoin Core PR #31117)][t4-discussion]
+
+[block-storms]: https://blog.lopp.net/the-block-storms-of-bitcoins-testnet/
+[bitcointalk-thread]: https://bitcointalk.org/index.php?topic=5569103.0
+[signet-bip54]: https://delvingbitcoin.org/t/bitcoin-inqusition-29-2/2236
+[t4-discussion]: https://github.com/bitcoin/bitcoin/pull/31117
+[BIP30]: bip-0030.mediawiki
+[BIP54]: bip-0054.md
+[BIP94]: bip-0094.mediawiki
+
+## Copyright
+
+This document is licensed under the Creative Commons CC0 1.0 Universal license.
+
+## Changelog
+
+* __0.1.0__ (2026-06-02):
+    * Initial draft
+
+[^pubkey-methodology]: The `Pubkey` field is the public key placed in the coinbase transaction's
+  output as a pay-to-public-key (P2PK) script. Setting it to a recent Bitcoin mainnet block hash
+  serves two purposes: the resulting output is provably unspendable, and the referenced block hash
+  acts as an anti-pre-mine commitment, since the genesis block cannot have been created before that
+  mainnet block existed.