Re: [bitcoin-dev] Extension block proposal by Jeffrey et al

[Luke Dashjr <luke@dashjr.org>]

Recently there has been some discussion of an apparent work-in-progress 
extension block proposal by Christopher Jeffrey, Joseph Poon, Fedor Indutny, 
and Steven Pair. Since this hasn't been formally posted on the ML yet, perhaps 
it is still in pre-draft stages and not quite ready for review, but in light 
of public interest, I think it is appropriate to open it to discussion, and 
toward this end, I have reviewed the current revision.

For reference, the WIP proposal itself is here:
    https://github.com/tothemoon-org/extension-blocks

==Overall analysis & comparison==

This is a relatively complicated proposal, creating a lot of additional 
technical debt and complexity in comparison to both BIP 141 and hardforks. It 
offers no actual benefits beyond BIP 141 or hardforks, so seems irrational to 
consider at face value. In fact, it fits much better the inaccurate criticisms 
made by segwit detractors against BIP 141.

That being said, this proposal is very interesting in construction and is for 
the most part technically sound. While ill-fit to merely making blocks larger, 
it may be an ideal fit for fundamentally different block designs such as 
Rootstock and MimbleWimble in absence of decentralised non-integrated 
sidechains (extension blocks are fundamentally sidechains tied into Bitcoin 
directly).

==Fundamental problem==

Extension blocks are a risk of creating two classes of "full nodes": those 
which verify the full block (and are therefore truly full nodes), and those 
which only verify the "base" block. However, because the extension is 
consensus-critical, the latter are in fact not full nodes at all, and are left 
insecure like pseudo-SPV (not even real SPV) nodes. This technical nature is 
of course true of a softfork as well, but softforks are intentionally designed 
such that all nodes are capable of trivially upgrading, and there is no 
expectation for anyone to run with pre-softfork rules.

In general, hardforks can provide the same benefits of an extension block, but 
without the false expectation and pointless complexity.

==Other problems & questions==

> These outpoints may not be spent inside the mempool (they must be redeemed 
from the next resolution txid in reality).

This breaks the ability to spend unconfirmed funds in the same block (as is 
required for CPFP).

The extension block's transaction count is not cryptographically committed-to 
anywhere. (This is an outstanding bug in Bitcoin today, but impractical to 
exploit in practice; however, exploiting it in an extension block may not be 
as impractical, and it should be fixed given the opportunity.)

> The merkle root is to be calculated as a merkle tree with all extension 
block txids and wtxids as the leaves.

This needs to elaborate how the merkle tree is constructed. Are all the txids 
followed by all the wtxids (tx hashes)? Are they alternated? Are txid and 
wtxid trees built independently and merged at the tip?

> Output script code aside from witness programs, p2pkh or p2sh is considered 
invalid in extension blocks.

Why? This prevents extblock users from sending to bare multisig or other 
various possible destinations. (While static address forms do not exist for 
other types, they can all be used by the payment protocol.)

Additionally, this forbids datacarrier (OP_RETURN), and forces spam to create 
unprovably-unspendable UTXOs. Is that intentional?

> The maximum extension size should be intentionally high.

This has the same "attacks can do more damage than ordinary benefit" issue as 
BIP141, but even more extreme since it is planned to be used for future size 
increases.

> Witness key hash v0 shall be worth 1 point, multiplied by a factor of 8.

What is a "point"? What does it mean multiplied by a factor of 8? Why not just 
say "8 points"?

> Witness script hash v0 shall be worth the number of accurately counted 
sigops in the redeem script, multiplied by a factor of 8.

Please define "accurately counted" here. Is this using BIP16 static counting, 
or accurately counting sigops during execution?

> To reduce the chance of having redeem scripts which simply allow for garbage 
data in the witness vector, every 73 bytes in the serialized witness vector is 
worth 1 additional point.

Is the size rounded up or down? If down, 72-byte scripts will carry 0 
points...)

==Trivial & process==

BIPs must be in MediaWiki format, not Markdown. They should be submitted for 
discussion to the bitcoin-dev mailing list, not social media and news.

> Layer: Consensus (soft-fork)

Extension blocks are more of a hard-fork IMO.

> License: Public Domain

BIPs may not be "public domain" due to non-recognition in some jurisdictions. 
Can you agree on one or more of these? 
https://github.com/bitcoin/bips/blob/master/bip-0002.mediawiki#Recommended_licenses

> ## Abstract
> 
> This specification defines a method of increasing bitcoin transaction 
throughput without altering any existing consensus rules.

This is inaccurate. Even softforks alter consensus rules.

> ## Motivation
> 
> Bitcoin retargetting ensures that the time in between mined blocks will be 
roughly 10 minutes. It is not possible to change this rule. There has been 
great debate regarding other ways of increasing transaction throughput, with 
no proposed consensus-layer solutions that have proven themselves to be
particularly safe.

Block time seems entirely unrelated to this spec. Motivation is unclear.

> Extension blocks leverage several features of BIP141, BIP143, and BIP144 for 
transaction opt-in, serialization, verification, and network services, and as 
such, extension block activation entails BIP141 activation.

As stated in the next paragraph, the rules in BIP 141 are fundamentally 
incompatible with this one, so saying BIP 141 is activated is confusingly 
incorrect.

> This specification should be considered an extension and modification to 
these BIPs. Extension blocks are _not_ compatible with BIP141 in its current 
form, and will require a few minor additional rules.

Extension blocks should be compatible with BIP 141, there doesn’t appear to be 
any justification for not making them compatible.

> This specification prescribes a way of fooling non-upgraded nodes into 
believing the existing UTXO set is still behaving as they would expect.

The UTXO set behaves fundamentally different to old nodes with this proposal, 
albeit in a mostly compatible manner.

> Note that canonical blocks containing entering outputs MUST contain an 
extension block commitment (all zeroes if nothing is present in the extension 
block).

Please explain why in Rationale.

> Coinbase outputs MUST NOT contain witness programs, as they cannot be 
sweeped by the resolution transaction due to previously existing consensus 
rules.

Seems like an annoying technical debt. I wonder if it can be avoided.

> The genesis resolution transaction MAY also include a 1-100 byte pushdata in 
the first input script, allowing the miner of the genesis resolution to add a 
special message. The pushdata MUST be castable to a true boolean.

Why? Unlike the coinbase, this seems to create additional technical debt with 
no apparent purpose. Better to just have a consensus rule every input must be 
null.

> The resolution transaction's version MUST be set to the uint32 max (`2^32 - 
1`).

Transaction versions are signed, so I assume this is actually simply -1. 
(While signed transaction versions seemed silly to me, using it for special 
cases like this actually makes sense.)

> ### Exiting the extension block

Should specify that spending such an exit must use the resolution txid, not 
the extblock's txid.

> On the policy layer, transaction fees may be calculated by transaction cost 
as well as additional size/legacy-sigops added to the canonical block due to 
entering or exiting outputs.

BIPs should not specify policy at all. Perhaps prefix "For the avoidance of 
doubt:" to be clear that miners may perform any fee logic they like.

> Transactions within the extended transaction vector MAY include a witness 
vector using BIP141 transaction serialization.

Since extblock transactions are all required to be segwit, why wouldn't this 
be mandatory?

> - BIP141's nested P2SH feature is no longer available, and no longer a 
consensus rule.

Note this makes adoption slower: wallets cannot use the extblock until the 
economy has updated to support segwit-native addresses.

> To reduce the chance of having redeem scripts which simply allow for garbage 
data in the witness vector, every 73 bytes in the serialized witness vector is 
worth 1 additional point.

Please explain why 73 bytes in Rationale.

> This leaves room for 7 future soft-fork upgrades to relax DoS limits.

How so? Please explain.

> A consensus dust threshold is now enforced within the extension block.

Why?

> If the second highest transaction version bit (30th bit) is set to to `1` 
within an extension block transaction, an extra 700-bytes is reserved on the 
transaction space used up in the block.

Why wouldn't users set this on all transactions?

> `default_witness_commitment` has been renamed to 
`default_extension_commitment` and includes the extension block commitment 
script.

`default_witness_commitment` was never part of the GBT spec. At least describe 
what this new key is.

> - Deployment name: `extblk` (appears as `!extblk` in GBT).

Should be just `extblk` if backward compatibility is supported (and `!extblk` 
when not).

> The "deactivation" deployment's start time...

What about timeout? None? To continue the extension block, must it be 
deactivated and reactivated in parallel?