From 13925d0ba1d766828671782d2f6c535bf93809d6 Mon Sep 17 00:00:00 2001 From: Shannon Appelcline Date: Wed, 7 Jun 2017 11:02:07 -0700 Subject: [PATCH] Update 11_3_Empowering_Bitcoin_with_Scripts.md --- 11_3_Empowering_Bitcoin_with_Scripts.md | 69 ++++++++++++------------- 1 file changed, 33 insertions(+), 36 deletions(-) diff --git a/11_3_Empowering_Bitcoin_with_Scripts.md b/11_3_Empowering_Bitcoin_with_Scripts.md index 024f049..6870134 100644 --- a/11_3_Empowering_Bitcoin_with_Scripts.md +++ b/11_3_Empowering_Bitcoin_with_Scripts.md @@ -2,23 +2,23 @@ > **NOTE:** This is a draft in progress, so that I can get some feedback from early reviewers. It is not yet ready for learning. -Bitcoin Scripts can go far beyond the relatively simple financial instruments detailed to date. They're also the foundation of most more complex usages of the Bitcoin network, as demonstrated by these real-world examples of off-chain functionality, drawn from the Lightning Network examples in BIP 112. +Bitcoin Scripts can go far beyond the relatively simple financial instruments detailed to date. They're also the foundation of most complex usages of the Bitcoin network, as demonstrated by these real-world examples of off-chain functionality, drawn from the Lightning Network examples in BIP 112. ## Lock for the Lightning Network The [Lightning Network](https://rusty.ozlabs.org/?p=450) is a payment channel that allows users to take funds off-chain and engage in numerous microtransactions before finalizing the payment channel and bringing the funds back into Bitcoin. Benefits include lower fees and faster transaction speeds. -[BIP 112](https://github.com/bitcoin/bips/blob/master/bip-0112.mediawiki) contains a few examples of how transactions could be locked in the Lightning Network. +[BIP 112](https://github.com/bitcoin/bips/blob/master/bip-0112.mediawiki) contains a few examples of how these off-chain transactions could be generated, using Bitcoin locking scripts. ### Lock with Revocable Commitment Transactions -The trick with Lightning is that it's off-chain. The participants jointly lock funds on Bitcoin with an n-of-n multisignature, then they engage in a number of transactions between themselves. Each new "commitment transaction" splits those joint funds in a different way; it's partially signed but _it isn't put on the blockchain_. +The trick with Lightning is the fact that it's off-chain. Too use Lightning, participants jointly lock funds on Bitcoin with an n-of-n multisignature. Then, they engage in a number of transactions between themselves. Each new "commitment transaction" splits those joint funds in a different way; these transactions are partially signed but _they aren't put on the blockchain_. -So how do you keep one of the participants from reverting back to an old transaction that's more beneficial to them? That's where revocation comes in, as is demonstrated in the following example from BIP 112, which was intended as a stepping stone toward Lightning. You give the participant who would be harmed by reversion to a revoked transaction the ability to reclaim the funds himself if the another participant illegitamately tried to use the revoked transaction. +If you have a mass of unpublished transactions, any of which _could_ be placed on the Blockchain, how do you keep one of the participants from reverting back to an old transaction that's more beneficial to them? The answer is _revocation_. A simplified example in BIP 112, which offers one of the stepping stones to Lightning, shows how. You give the participant who would be harmed by reversion to a revoked transaction the ability to reclaim the funds himself if the the other participant illegitimately tries to use the revoked transaction. -For example, presume that Alice updated the commitment transaction to give more of the funds to Bob (effectively: she sent funds to Bob via the Lightning network). As part of this new transaction, she gives Bob a `revokeHash` which can be used to claim the funds from the previous transaction, before Alice gave Bob the new funds. +For example, presume that Alice and Bob update the commitment transaction to give more of the funds to Bob (effectively: Alice sent funds to Bob via this proto-Lightning network). They partially sign new transactions, but they also each offer up their own `revokeCode` for previous transactions. This effectively guarantees that they won't publish previous transactions, because doing so would allow their counterparty to claim those previous funds. -The locking script from that revoked commitment transaction looks as follows: +So what does the old transaction look like? It was a commitment transaction showing funds intended for Alice, before she gave them to Bob. It had a locking script as follows: ``` OP_HASH160 @@ -38,13 +38,13 @@ ELSE ENDIF OP_CHECKSIG ``` -Theoretically, this transaction should never be spent at the point. Bob has no incentive to because he has the newer transaction, after Alice sent him the new funds. Alice has no incentive too, because she loses the funds if she tries. So no one puts the transaction onto the blockchain, and the off-chain transactions continue. +The `ELSE` block is where Alice got her funds, after a 24-hour delay. However now it's been superceded; that's the whole point of a Lightning-style payment channel, after all. In this situation, this transaction should never be published. Bob has no incentive to because he has a newer transaction, which benefits him more because he's been sent some of Alice's funds. Alice has no incentive either, because she loses the funds if she tries because of that `revokeCode`. So no one puts the transaction onto the blockchain, and the off-chain transactions continue. -But what if Alice tried to cheat? She puts the transaction back onto the blockchain even though it's been revoked. Now the locking script comes to the rescue! +It's worth exploring how this script would work in a variety of situations, most of which involve Alice trying to cheat by reverting to this older transaction, which depicts the funds _before_ Alice sent some of them to Bob. #### Run the Lock Script for Cheating Alice, with Revocation Code -Alice could try to use revocation code that she gave to Bob to immediately claim the funds. She sends in ` `: +Alice could try to use revocation code that she gave to Bob to immediately claim the funds. She writes a locking script of ` `: ``` Script: OP_HASH160 OP_EQUAL IF ELSE <+24Hours> OP_CHECKSEQUENCEVERIFY OP_DROP ENDIF OP_CHECKSIG Stack: [ ] @@ -63,13 +63,13 @@ Script: IF ELSE <+24Hours> OP_CHECKSEQUENCEVERIFY OP_DROP OP_EQUAL Stack: [ True ] ``` -The `OP_EQUAL` feeds the `IF` statement. Because Alice uses the hash, she gets into the branch that allows her to redeem the funds immediately, collapsing the rest of the script down to `` (within the conditional) and `OP_CHECKSIG` (afterward). +The `OP_EQUAL` feeds the `IF` statement. Because Alice uses the `revokeCode`, she gets into the branch that allows her to redeem the funds immediately, collapsing the rest of the script down to `` (within the conditional) and `OP_CHECKSIG` (afterward). ``` Script: OP_CHECKSIG Running: True IF Stack: [ ] ``` -Curses! Only Bob can sign immediately with the hash! +Curses! Only Bob can sign immediately using the `redeemCode`! ``` Script: OP_CHECKSIG Stack: [ ] @@ -80,8 +80,7 @@ Stack: [ False ] ``` #### Run the Lock Script for Cheating Alice, without Revocation Code -So what if Alice instead tries to use her own signature, without the revocation code? - +So what if Alice instead tries to use her own signature, without the `revokeCode`? She uses an unlocking script of ` `. ``` Script: 0 OP_HASH160 OP_EQUAL IF ELSE <+24Hours> OP_CHECKSEQUENCEVERIFY OP_DROP ENDIF OP_CHECKSIG Stack: [ ] @@ -100,7 +99,7 @@ Script: IF ELSE <+24Hours> OP_CHECKSEQUENCEVERIFY OP_DROP OP_EQUAL Stack: [ False ] ``` -We now collapse down to the `ELSE` statement and what comes after the conditional +We now collapse down to the `ELSE` statement and what comes after the conditional: ``` Script: <+24Hours> OP_CHECKSEQUENCEVERIFY OP_DROP OP_CHECKSIG Running: False IF @@ -115,10 +114,9 @@ Script: OP_DROP OP_CHECKSIG Running: <+24Hours> OP_CHECKSEQUENCEVERIFY Stack: [ <+24Hours> ] — Script EXITS ``` -#### Run the Lock Script for Virtuous Bob - -What this means is that Bob has 24 hours to reclaim his funds if Alice ever tries to cheat, using the `` and his signature: +#### Run the Lock Script for Victimized Bob +What this means is that Bob has 24 hours to reclaim his funds if Alice ever tries to cheat, using the `` and his signature as his unlocking script: ``` Script: OP_HASH160 OP_EQUAL IF ELSE <+24Hours> OP_CHECKSEQUENCEVERIFY OP_DROP ENDIF OP_CHECKSIG Stack: [ ] @@ -150,15 +148,21 @@ Stack: [ True ] ``` #### Run the Lock Script for Virtuous Alice -All of the commitment scripts are locked with this same transaction, whether they've been revoked or not. That means that the newest commitment script, which is the currently valid one, is locked with it as well. In this situation Alice has never sent a new transaction to Bob and thus never sent him the `revokeCode`. She can now close down the Lightning channel at this point. She puts the transaction on the chain and she waits 24 hours. Bob can't do anything about it because he doesn't have the recovation code. Then, after the wait, Alice reclaims her funds. (Bob does the last thing with his last valid transaction.) +All of Alice's commitment transactions are locked with this same locking script, whether they've been revoked or not. That means that the newest commitment transaction, which is the currently valid one, is locked with it as well. Alice has never sent a newer transaction to Bob and thus never sent him the previous `revokeCode`. + +In this situation, she could virtuously publish the transaction, closing down the proto-Lightning channel. She puts the transaction on the chain and she waits 24 hours. Bob can't do anything about it because he doesn't have the recovation code. Then, after the wait, Alice reclaims her funds. (Bob does the same thing with his own final commtiment transaction.) ### Lock with Hashed Time-Lock Contracts -BIP 112 also offers a slightly more complex mechanism for protecting Lightning-like transactions: a [hashed timelock contract](https://en.bitcoin.it/wiki/Hashed_Timelock_Contracts), or HTLCs. This is what allows singular transactions to actually become a network and is what's actually used as the basis of the Lightning network. +The Revocable Commitment Transactions were just a stepping stone to Lightning. The actual Lightning Network uses a more complex mechanism called a [hashed timelock contract](https://en.bitcoin.it/wiki/Hashed_Timelock_Contracts), or HTLC. + +The main purpose of HTLCs is to create a comprehensive network of participants. Transactions are no longer just between a pair of participants who have entered the network together, but can now be between previously unassociated people. When funds are sent, a string of transactions are created, each of them locked with a `secretHash`. When the corresponding `secretCode` is revealed, the entire string of transactions can be spent. This is what allows singular transactions to actually become a network. + +There's also a bit more complexity in Lightning Network locking scripts. There are separate locks for the sender and the recipient of each transaction that are more widely divergent than the differing commitment transactions alluded to in the previous section. We're going to show both of them, to demonstrate the power of these locking scripts, but we're not going to dwell on how they interact with each other. #### Lock the Recipient's Transaction -Take as an example the following commitment transaction created for new funds that Alice has received: +Once more, we're going to start looking at Alice's commitment transaction, which shows funds that she's received: ``` OP_HASH160 OP_DUP @@ -191,13 +195,15 @@ ENDIF OP_CHECKSIG ``` -The key to this is the `secretHash`, which is what allows a transaction to span the network. Each of several transactions is locked with the `secretHash`, allowing a transmission between two otherwise unconnected people on the Lightning network. When the transaction has spanned from its originator to its intended recipient, the `secretCode` is revealed, which allows all the participants to create a `secretHash` and unlock the whole network of payments: after the `secretCode` has been revealed, Alice can claim the funds 24 hours after the transaction is put on the Bitcoin network. +The key to these new HTLCs is the `secretHash`, which we've seen is what allows a transaction to span the network. When the transaction has spanned from its originator to its intended recipient, the `secretCode` is revealed, which allows all the participants to create a `secretHash` and unlock the whole network of payments. -However, as with the previous example, the hash could alternatively be a `revokeHash` that was supplied after the transaction was supplanted by a new one. Bob can relcaim the funds in that situation _or if_ an absolute timeout has occurred. +After the `secretCode` has been revealed, the `IF` branch opens up: Alice can claim the funds 24 hours after the transaction is put on the Bitcoin network. + +However, there's also the opportunity for Bob to reclaim his funds, which appears in the `ELSE` branch. He can do so if the transaction has been revoked (but Alice puts it on the blockchain anyway), _or if_ an absolute timeout has occurred. #### Lock the Sender's Transaction -Due to the additional complexity of HTLCs, an additional commitment transaction is required for the sender of every HTLC transaction: +Here's the alternative commitment transaction locking script used by the sender: ``` OP_HASH160 OP_DUP @@ -241,11 +247,6 @@ Running: OP_DUP Stack: [ ] Initial Script: OP_EQUAL OP_SWAP OP_EQUAL OP_ADD -Running: OP_DUP -Stack: [ ] - -Initial Script: OP_EQUAL OP_SWAP OP_EQUAL OP_ADD -Running: OP_DUP Stack: [ ] Initial Script: OP_SWAP OP_EQUAL OP_ADD @@ -263,19 +264,15 @@ Initial Script: OP_ADD Running: OP_EQUAL Stack: [ ] -Initial Script: OP_ADD -Running: OP_EQUAL -Stack: [ ] - Initial Script: Running: OP_ADD Stack: [ ] ``` -Running through the script, it becomes obvious that the initial checks determine if the hash was either the `secretCode` or the `revokeCode`. If so, Alice can take the funds in the first block. If not, Bob can take the funds, but only after Alice has had her chance, and both the 24 hour timeout and the absolute timeout have passed. +Running through the script reveals that the initial checks, above the `IF`/`ELSE`/`ENDIF` determine if the hash was either the `secretCode` _or_ the `revokeCode`. If so, Alice can take the funds in the first block. If not, Bob can take the funds, but only after Alice has had her chance, and both the 24 hour timeout and the absolute timeout have passed. #### Understand HTLCs -HTLCs are quite complex, and you may not entirely understand them from just this overview. Rusty Russell's [overview](https://rusty.ozlabs.org/?p=462) of them has more and there's even more in his [Deployable Lightning](https://github.com/ElementsProject/lightning/blob/master/doc/deployable-lightning.pdf) paper. But don't worry if some of the intricacies still escape you, particularly the interrelations of the two scripts. +HTLCs are quite complex, and this overview doesn't try to explain all of their intricacies. Rusty Russell's [overview](https://rusty.ozlabs.org/?p=462) explains more. and there's even more detail in his [Deployable Lightning](https://github.com/ElementsProject/lightning/blob/master/doc/deployable-lightning.pdf) paper. But don't worry if some of the intricacies still escape you, particularly the interrelations of the two scripts. For the purposes of this tutorial, there are two important lessons for HTLCs: @@ -286,6 +283,6 @@ It's worth your time running each of the two HTLC scripts through each of its pe ### Summary: Empowering Bitcoin with Scripts -We're closing our examination of Bitcoin Scripts with a look at how truly powerful they can be. In 20 opcodes or less, a Bitcoin Script can form the basis of an entire off-chain payment channel. Similarly, two-way pegged sidechains are the product of less than twenty opcodes, as briefly noted in [BIP 112](https://github.com/bitcoin/bips/blob/master/bip-0112.mediawiki). +We're closing our examination of Bitcoin Scripts with a look at how truly powerful they can be. In 20 opcodes or less, a Bitcoin Script can form the basis of an entire off-chain payment channel. Similarly, two-way pegged sidechains are the product of less than twenty opcodes, as also briefly noted in [BIP 112](https://github.com/bitcoin/bips/blob/master/bip-0112.mediawiki). -If you've ever seen complex Bitcoin functionality or Bitcoin-adjacent systems, they were problem built on Bitcoin Scripts. And now you have all the tools to do the same yourself. +If you've ever seen complex Bitcoin functionality or Bitcoin-adjacent systems, they were probably built on Bitcoin Scripts. And now you have all the tools to do the same yourself.