Hi Russell,

>I haven't really been following the drivechain discussion; I have found the documentation about how drivechains are supposed to work scattered and difficult to follow. So, without advocating for or against this proposal, I'd also suggest that adding an opcode is not the best way to implement this bribe.

Despite the flaws in this draft BIP, the goal is to start consolidating this information into a more compact format. This BIP is *only*
meant to address the Blind Merging Mining Process of drivechains. It does *not* address the withdrawal process from drivechain -> bitcoin.

>The problem I see is that to send a bribe one must first post a transaction to a script that uses the OP_BRIBE code that fixes the critical hash (and the sidechain id), and then a second transaction is needed to pay the bribe to the miner.

That is intentional, this allows for a competitive process (like bitcoin mining) for a block to be 'found' on the sidechain. The OP_BV output that rewards
the bitcoin miner the most amount of money should be the one that is included in the bitcoin blockchain. If I understand your scheme correctly, you are
assuming the the bitcoin miner is *also* following the sidechain -- Sztorc's scheme does not make this assumption. The *number one goal* of BMM
is to *minimize* the resource burden on bitcoin miners for mining on a drivechain.

To gmaxwell/luke-jr,

I agree my commitment scheme is flawed. Thanks for pointing it out. Is there any way we could manipulate a coinbase transaction
into spending these OP_BV outputs? According to instagibbs, and AFIACT he is right, we cannot have coinbase transactions
spend any outputs in previous blocks without a hard fork. This is unfortunate because it might make more sense for the coinbase transaction
to spend these OP_BV outputs. We could design the coinbase transaction's scriptSig to push the critical hash onto the stack and
place an OP_EQUAL on the OP_BV output to verify they were equal.
If I understand gmaxwell's concern about 'monotone' (or stateless) blocks correctly, I *think*
this solution might fix that as well.

Another way we could fix this is by *fixing* the drivechain indices. Therefore the mining rewards and witness commitments must
*not* occupy one of those indices -- but can occupy any other indice in the coinbase output.
This would give us future flexibility for committing to new soft forks. For instance, we would say
the mining reward must *not* be index 0 of the coinbase transaction, but can occupy index 1 - 256. The same would apply for witness commitments.

-Chris

On Wed, Jun 28, 2017 at 5:49 PM, Russell O'Connor <roconnor@blockstream.io> wrote:

I haven't really been following the drivechain discussion; I have found the documentation about how drivechains are supposed to work scattered and difficult to follow. So, without advocating for or against this proposal, I'd also suggest that adding an opcode is not the best way to implement this bribe.

The problem I see is that to send a bribe one must first post a transaction to a script that uses the OP_BRIBE code that fixes the critical hash (and the sidechain id), and then a second transaction is needed to pay the bribe to the miner.

I suggest instead to use a 0 output value with some currently non-standard OP_RETURN output script that specifies the critical hash (and the sidechain id), similar to ZmnSCPxj's idea.  The difference is that I we would soft-fork a rule that says that such an output is only legal when a miner places the same critical hash suitably in their coinbase output.

OP_RETURN outputs are prunable from the UTXO set.  The special bribe output can be fixed to 0 value because the bribe will be paid using the transaction's fees.  To perform a bribe, a user creates and signs a transaction containing one (or more) of these special bribe outputs.  The fee of this transaction constitutes the bribe, and any change the user has can be sent back to themselves.  This way only a single transaction is required to make a bribe.  I didn't really understand the bribe refund mechanism, but I think the fact that the bribe can be done in a single transaction this way alleviates any need for bribe refunds.

Hopefully I have understood the goal of this proposal.

On Wed, Jun 28, 2017 at 6:20 PM, Paul Sztorc via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:

Hi ZmnSCPxj,

It seems that, in your version, the "bribers" would react to the scheme
in inefficient ways, particularly when the mainchain's tx-fee-rate (ie
fee per Kb) is low.

In short, there would be many bribe-attempts (each of which would take
up space in mainchain blocks), almost all of which would be unsuccessful.

In turn, miners would likely react to this, and try to improve the state
of affairs by offering users the privilege of occupying transaction slot
#2 (ie, the one right after the coinbase). Users would need to trust
miners for this, which introduces a cost friction which is pure
deadweight loss. And, it might be easier for larger/older miners to be
trustworthy than smaller/newer ones.

Your way is actually very similar to mine. Mine _forces_ the bribe to be
in the earliest txn (the coinbase) and to only occur once. Yours doesn't
do anything to refund the briber, if the sidechain (but not the
mainchain) reorganizes (as it can easily do, if an older sidechain
parent is extended while the mainchain proceeds normally). This creates
additional risk.

I think mine is also much more space-efficient. Even if ours each had
exactly one h* per sidechain per block, it seems that I only require one
hash to be communicated (plus an indicator byte, and a ~2 byte counter
for the ratchet), whereas you require two. Since its overhead per
sidechain per block, it actually might really add up.

Thanks,
Paul

On 6/28/2017 4:26 AM, ZmnSCPxj via bitcoin-dev wrote:
> Good morning.
>
> I still do not see what this does that cannot be done by:
>
> OP_RETURN <sidechain-id> <sidechain-block-id>
>
> A transaction with such an output would allow sidechain-miners to bribe
> mainchain-miners by paying a transaction fee if the transaction
> containing this OP_RETURN is included in a block and committed to by a
> mainchain-miner in the Merkle tree root.
>
> It would not require a softfork.
>
> --
>
> I have an alternate proposal that sidechains and altcoins that want to
> do "blinded" merge mining can use without a softfork:
>
> 1.  Encode a block header as a simple cons-pair, with the head as the
> block and the tail as the parent cons-pair.
> 1.1.  This can be encoded as a 32-byte hash of the block including its
> header, and the 32-byte hash of the parent cons-pair.
> 1.2.  This is now the actual "chain" in the sidecoin/altcoin blockchain.
> 2.  When a sidechain-node wants to know the consensus, it downloads
> mainchain-blocks and looks for OP_RETURN's.
> 2.1.  Starting with its genesis cons-pair hash (equivalent to the empty
> list) as the current cons-pair, it scans each OP_RETURN transaction.
> 2.1.1.  If an OP_RETURN is 64-byte and has the parent cons-pair equal to
> the current cons-pair, look for the side block indicated and confirm its
> correctness.  If correct, update the current cons-pair for the hash of
> the OP_RETURN data.
> 2.2.  When reaching the latest mainchain block, the current cons-pair is
> now the sidecoin/altcoin latest block.
> 2.3.  Note that if multiple OP_RETURN in a block match the current
> cons-pair, the first one is considered the correct chain.  This property
> means that the sidechain/altchain can only have a chainsplit if the
> mainchain has a chainsplit.
> 3.  When a sidechain-miner wants to create a side-block, it generates a
> new cons-pair and creates an OP_RETURN transaction for it, paying a
> mainchain-miner to include it in the next mainchain-block.
> 3.1.  The sidechain-miner risks that its competitors will outbid it and
> get its OP_RETURN earlier in a mainchain-block (or earlier in the order
> of transactions).  It can mitigate this risk by updating itself to
> become a mainchain-miner, it can then keep its OP_RETURN transaction
> private and put it earlier in the block, ensuring it will "win" the
> sidechain-consensus if it wins the mainchain-consensus.
>
> Regards,
> ZmnSCPxj
>
> -------- Original Message --------
> Subject: [bitcoin-dev] BIP: OP_BRIBVERIFY - the op code needed for Blind
> Merge Mined drivechains
> Local Time: June 28, 2017 8:37 AM
> UTC Time: June 28, 2017 12:37 AM
> From: bitcoin-dev@lists.linuxfoundation.org
> To: Bitcoin Protocol Discussion <bitcoin-dev@lists.linuxfoundation.org>
>
> <pre>
>   BIP: <BIP number>
>   Layer: Consensus (Soft fork)
>   Title: OP_BRIBEVERIFY
>   Author: Chris Stewart <chris@suredbits.com>
>   Status: Draft
>   Type: Standards Track
>   Created: 2017-06-27
> </pre>
>
>
> ==Abstract==
>
> This BIP describes a new opcode, OP_BRIBEVERIFY, for the Bitcoin
> scripting system that allows for a user to bribe a miner to include a hash
> in the coinbase transaction's output.
>
>
> ==Summary==
>
> BRIBEVERIFY redefines the existing NOP4 opcode. When executed, if the given
> critical hash is included at the given vout index in the coinbase
> transaction
> the script evaluates to true. Otherwise, the script will fail.
>
> This allows sidechains to be merged mined against
> bitcoin without burdening bitcoin miners with extra resource requirements.
>
> ==Motivation==
>
> The current political climate of bitcoin is extremely contentious. Many
> community members
> have different visions of what bitcoin is. This op code is meant to
> enable [http://www.truthcoin.info/blog/blind-merged-mining/ Blind Merge
> Mining].
> This enables sidechains in Bitcoin. With OP_BRIBEVERIFY, sidechains
> miners can
> bribe bitcoin miners to to include their block hash in the bitcoin
> blockchain. If their block
> is included in the coinbase transaction's vout, it is assumed that block
> is a mined block on the sidechain.
>
> This will allow various factions of the community to realize their
> vision on their own separate
> blockchain that is interoperable with the bitcoin blockchain. This
> allows those factions to use
> bitcoin as a 'reserve currency' for their own network.
>
>
> ===Commitment Structure===
>
> A new block rule is added which requires that the miner's coinbase
> reward be at index 0 in the coinbase transaction's output vector.
>
> It also fixes the witness commitment output to be at index 1 of the
> coinbase transaction's output vector.
>
> This is needed so we can reliably tell what vout corresponds to what
> drivechain. For instance, the mimblewimble sidechain
> could correspond to index 2 of the vector outputs on the coinbase
> transaction.
>
> The commitment is recorded in a <code>scriptPubKey</code> of the
> coinbase transaction. It must be at least 34 bytes in size
>    1-byte - OP_RETURN (0x6a)
>    1-byte - Push the following 32 bytes (0x20)
>   32-byte - block hash
>
> the 35th byte and onward have no consensus meaning.
>
> ===OP_BRIBEVERIFY op code===
>
> This op code reads two arguments from the stack. The stack top is
> expected to be a sidechain id for which this user attempting to blind
> merge mine for.
> The next element on the stack is expected to be a block hash. This op
> code looks into the coinbase transaction's output vector at the given
> index (which is derived from the sidechain id) and checks
> to see if the hash in the block matches the hash inside of the
> BRIBEVERIFY program. If the hashes match, the OP_BRIBEVERIFY acts as an
> OP_NOP. If the
> comparison between the two hashes fail, the script fails.
>
> ===BRIBEVERIFY program===
>
> A standard BRIBEVERIFY program has the format:
>   1-byte - Push the following 32 bytes (0x20)
>  32-byte - block hash
>   1 byte - Push operation? (needed if number can't be encoded as OP_0 -
> OP_16)
>   1 byte - sidechain id
>   1 byte - OP_BRIBEVERIFY op code
>
> ==Detailed Specification==
>
> Refer to the reference implementation, reproduced below, for the precise
> semantics and detailed rationale for those semantics.
>
>
>  case OP_NOP4:
>  {
>     //format: block_hash sidechain_id OP_BRIBEVERIFY
>     if (!(flags & SCRIPT_VERIFY_BRIBEVERIFY)) {
>         // not enabled; treat as a NOP4
>         if (flags & SCRIPT_VERIFY_DISCOURAGE_UPGRADABLE_NOPS) {
>             return set_error(serror, SCRIPT_ERR_DISCOURAGE_UPGRADABLE_NOPS);
>         }
>         break;
>     }
>
>     if (stack.size() < 2)
>         return set_error(serror, SCRIPT_ERR_INVALID_STACK_OPERATION);
>
>     const CScriptNum scriptNumSidechainId(stacktop(-1),fRequireMinimal);
>     uint8_t nSidechainId;
>     if (!checker.CheckSidechainId(scriptNumSidechainId,nSidechainId)) {
>         return set_error(serror, SCRIPT_ERR_UNKNOWN_SIDECHAIN);
>     }
>
>     // Check block hash
>     bool fHashCritical =
> checker.CheckCriticalHash(stacktop(-2),nSidechainId);
>     if (!fHashCritical) {
>         return set_error(serror, SCRIPT_ERR_UNSATISFIED_BRIBE);
>     }
>     break;
>  }
>
>
>
> https://github.com/Christewart/bitcoin/blob/94b6f33f2278c42d4d8758a3c8ffe2078e4ec933/src/script/interpreter.cpp#L427
>
> https://github.com/drivechain-project/bitcoin/pull/13
>
> ==Deployment==
>
> TODO
>
> ==Credits==
>
> Credit to Paul Sztorc for the original idea of Blind Merge Mined sidechains.
>
> Credit to CryptAxe for writing the foundational layer of software for
> drivechains so I could implement OP_BRIBEVERIFY.
>
>
> ==References==
>
> Blind Merge Mined Sidechains -
> http://www.truthcoin.info/blog/blind-merged-mining/
> Mailing list discussion -
> https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2017-May/014408.html
>
> ==Copyright==
>
> This document is placed in the public domain.
>
>

> _______________________________________________
> bitcoin-dev mailing list
> bitcoin-dev@lists.linuxfoundation.org
> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>
_______________________________________________
bitcoin-dev mailing list
bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev