Drivechain critiques by gmaxwell revisited, maybe you changed your mind?

[Wind_FURY]

For the newbies, who might not know. Drivechain is a software side-chain project by Paul Sztorc, which may help Bitcoin in scaling, privacy, and other "short-comings" that altcoins are trying to "fix". With Drivechain, we wouldn't need altcoins. Everything will be a side-chain of Bitcoin.

Drivechain allows BTC to travel back-and-forth to other software applications (called “sidechains”). Thus, BTC-owners can opt-in to new features or tradeoffs. Those who don’t opt-in, never need to care what any sidechain is doing.

As with the Lightning Network, DC-users move their coins into a “layer-2” – a zone where BTC can change hands an unlimited number of times. Eventually, just the net effect of these transfers is recorded back on layer-1.

Bitcoin Core can’t observe any layer-2 (by design), so we need a way to discourage fraudulent “netting”. LN counters theft via “justice transactions”; DC via forsaken mining revenues. LN-netting is private and instant; DC-netting is public and VERY slow (once per ~3 months).

Key benefits – only obtainable via Drivechain:

Three existential threats to BTC are neutralized – altcoin-competition, hard fork campaigns, and extension block campaigns.
BTC development becomes anti-fragile with respect to CoreDev mistakes.
BTC maintains hashrate security in the long run.
BTC can scale to credit-card level txn-processing – without changing the CONOP of Bitcoin Core. These cheap txns have optimal fungibility and supply vital pretext to the BTC ecosystem.
BTC gains new, experimental abilities, especially P2P event derivatives.

But this is not what the topic is about. It's about criticisms from the past from respected Bitcoin experts, and a revisitation of if they still believe they hold true.

gmaxwell, you said these in the past. With the dawn of Lightning and Liquid side-chains, and observing their own flaws, and short-comings, have you changed your mind from past criticisims on Drivechain?

Your July, 2017 criticism, https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2017-July/014726.html

December, 2018 criticism with Andrew Poelstra, https://www.youtube.com/watch?v=NA1xSe2nLoY&feature=youtu.be&t=5635

[NotATether]

@Wind_FURY, I found your quote at http://www.drivechain.info/, by the way.

I watched the youtube video and the guy was mentioning how each sidechain has a reward attached to it that they didn't think out through. I don't know how frequently this activity happens in the bitcoin development ecosystem but it's common for a lot of altcoin devs to not think through very important parts of the system through which reflects in the stability of the altcoin.

One problem I can think of right off the bat is what will happen to sidechains that scamcoins are hosted on after their developers abandon maintenance of their sidechain. I think that would be a waste of space.

But I think a much more important problem is that devs can't force everyone to use drivechains due to the decentralized model of bitcoin, by pushing a new bitcoin core release. Still, lots of people/services aren't using native segwit addresses, in some cases they are even using legacy addresses. But that's a different topic.

Now I'm not claiming to know anything about drivechains, but nothing is stopping someone from hard-forking bitcoin even if drivechains were in place. I don't think drivechains can interfere with layer-1 activity which is where the hardfork takes place.

[squatter]

But I think a much more important problem is that devs can't force everyone to use drivechains due to the decentralized model of bitcoin, by pushing a new bitcoin core release.

I wouldn't say that's the real problem of drivechains.

My understanding is that drivechains are soft forked into the main chain consensus. This means two things: Non-upgraded Bitcoin nodes would no longer be fully validating, and fully validating Bitcoin nodes would now have greater bandwidth, latency, and storage overheads. Just like block size increases, this would negatively affect miner and full node decentralization.

Those costs probably aren't worth the gains given how insecure a drivechain is, by Paul Sztorc's own admission:

It is said that “51% of the miners can steal all of the funds on the sidechain”.

It is true that 51% hashrate can overwhelm the 13,150 ACK requirement (ie, the “train metaphor”), and (if unopposed) include any withdrawal they like. Namely, they would include a withdrawal that pays them all of the sidechain’s BTC.

A 51% attack on Bitcoin only allows miners to double spend. On a drivechain, they can just steal everything.

[Wind_FURY]

@Wind_FURY, I found your quote at http://www.drivechain.info/, by the way.

Thanks. Yes, that's the site.

I watched the youtube video and the guy was mentioning how each sidechain has a reward attached to it that they didn't think out through. I don't know how frequently this activity happens in the bitcoin development ecosystem but it's common for a lot of altcoin devs to not think through very important parts of the system through which reflects in the stability of the altcoin.

One problem I can think of right off the bat is what will happen to sidechains that scamcoins are hosted on after their developers abandon maintenance of their sidechain. I think that would be a waste of space.

Waste of space to the people who run the side-chain, just like to the people who run the node of a shitcoin.

But I think a much more important problem is that devs can't force everyone to use drivechains due to the decentralized model of bitcoin, by pushing a new bitcoin core release. Still, lots of people/services aren't using native segwit addresses, in some cases they are even using legacy addresses. But that's a different topic.

That's actually a feature, not a problem/"bug".

Now I'm not claiming to know anything about drivechains, but nothing is stopping someone from hard-forking bitcoin even if drivechains were in place. I don't think drivechains can interfere with layer-1 activity which is where the hardfork takes place.

Warning, I'm just learning about Drivechain too. There will be lots of questions, and shit posts from me. Everyone is welcome to correct me.

BUT, their forked coin won't be as secure, and as valuable as Bitcoin. We could debate that a side-chain-coin might become more valuable than a forked shitcoin.

Let's wait for gmaxwell's post. I'm very curious, especially after Adam Back mentioned "trade-offs" of using Lightning, or Liquid. I believe using Drivechains might also just be a matter of accepting the trade-offs.

https://twitter.com/adam3us/status/1217845788438601733

lightning makes security tradeoffs, liquid makes different security tradeoffs

[squatter]

I'm very curious, especially after Adam Back mentioned "trade-offs" of using Lightning, or Liquid. I believe using Drivechains might also just be a matter of accepting the trade-offs.

https://twitter.com/adam3us/status/1217845788438601733

lightning makes security tradeoffs, liquid makes different security tradeoffs

These are two very different situations.

Lightning and Liquid can be thought of as off-chain or out-of-band. They have no effect on Bitcoin's consensus. The security trade-offs -- like needing to keep private keys online in LN, or trusting Liquid validators not to steal funds -- only affect LN users or Liquid users, respectively. Bitcoin users are unaffected no matter what.

With drivechains, Bitcoin users can't opt out. That's the difference. Drivechains are soft forked into the consensus by miners.

[NotATether]

With drivechains, Bitcoin users can't opt out. That's the difference. Drivechains are soft forked into the consensus by miners.

That may be one of the reasons some people are wary of them. Maybe they don't want to use drivechains since there is no way for them to opt out of it. One of the purported benefits of drivechains are for altcoins to be implemented on them, but with hundreds of alts already, I think it's too late to attempt to clean up that mess. Regarding

BTC maintains hashrate security in the long run.

what is this supposed to mean? Bitcoin already has a high hashrate without any other layers which reflects its mining difficulty, so it's pretty much impossible for anyone to manipulate the blockchain nowadays, compared to say 10 years ago. I'm not sure what drivechain is trying to solve here, unless it's trying to solve its own hashrate problem, which bitcoin is free from having.

[Wind_FURY]

I'm very curious, especially after Adam Back mentioned "trade-offs" of using Lightning, or Liquid. I believe using Drivechains might also just be a matter of accepting the trade-offs.

https://twitter.com/adam3us/status/1217845788438601733

lightning makes security tradeoffs, liquid makes different security tradeoffs

These are two very different situations.

Lightning and Liquid can be thought of as off-chain or out-of-band. They have no effect on Bitcoin's consensus. The security trade-offs -- like needing to keep private keys online in LN, or trusting Liquid validators not to steal funds -- only affect LN users or Liquid users, respectively. Bitcoin users are unaffected no matter what.

With drivechains, Bitcoin users can't opt out. That's the difference. Drivechains are soft forked into the consensus by miners.

With consensus reached, what would be bad about that? Segwit was soft forked into consensus with the backing of full nodes behnd it.

[squatter]

With drivechains, Bitcoin users can't opt out. That's the difference. Drivechains are soft forked into the consensus by miners.

With consensus reached, what would be bad about that? Segwit was soft forked into consensus with the backing of full nodes behnd it.

I can see the analogy you're making, but it's not entirely accurate. Segwit didn't fork a separate protocol into the consensus. I have doubts that Core would merge something like that, especially given the security trade-offs of a drivechain. Segwit was much less contentious than that.

I'm not particularly comfortable with the precedent of miners forking in sidechains, on which users blindly trust those miners. I think economically important drivechains may skew Bitcoin's mining incentives. I'm not sure. I'd like to see more convincing game theory suggesting drivechains are a good idea before considering enforcing them at the consensus level, that much is certain.

[Wind_FURY]

With drivechains, Bitcoin users can't opt out. That's the difference. Drivechains are soft forked into the consensus by miners.

With consensus reached, what would be bad about that? Segwit was soft forked into consensus with the backing of full nodes behnd it.

I can see the analogy you're making, but it's not entirely accurate. Segwit didn't fork a separate protocol into the consensus. I have doubts that Core would merge something like that, especially given the security trade-offs of a drivechain. Segwit was much less contentious than that.

OK. I'm still learning/reading more about Drivechain, and why Paul Sztorc believes strongly in that the trade-offs are worth it.

I'm not particularly comfortable with the precedent of miners forking in sidechains, on which users blindly trust those miners.

But full nodes secure the network, not miners.

I think economically important drivechains may skew Bitcoin's mining incentives. I'm not sure. I'd like to see more convincing game theory suggesting drivechains are a good idea before considering enforcing them at the consensus level, that much is certain.

Why? Mining on the base layer could be worth more because it supports all side-chains. Plus if a side-chain has value, wouldn't the game-theory/incentive-structure be the same as in Bitcoin?

[pooya87]

But full nodes secure the network, not miners.

that's wrong. they both do. you can't just cut out a major part of bitcoin network just like that.
miners provide the work which makes the difficulty go up and ensure the security and immutability of bitcoin blockchain due to expensive cost of 51% attacks thanks to their work. and nodes make sure miners stay in line by enforcing the consensus rules while keeping the network decentralized.

as for side-chain and mining, maybe it could happen alongside bitcoin in a similar fashion that merge mining works. the miner could be working on main net blocks with a much higher difficulty while finding hashes with a lower difficulty that could go into the side chain.

[squatter]

I'm not particularly comfortable with the precedent of miners forking in sidechains, on which users blindly trust those miners.

But full nodes secure the network, not miners.

Bitcoin nodes only SPV validate the drivechain. That's the security model. Full nodes may enforce the drivechain rules at the Bitcoin consensus level, but that can't stop miners from stealing drivechain funds. 51% of miners can always steal all drivechain funds, no matter what. The only thing Bitcoin nodes can do to stop them is a UASF after the fact that reverses the theft.

Mining on the base layer could be worth more because it supports all side-chains.

Drivechains, extension blocks, and similar mechanisms are block size increases by another name. They offload transaction throughput, bringing cheaper fee costs to users. Would the base layer actually be worth more to miners? That depends if overall combined throughput increases enough to account for the cheaper fees. We simply don't know, and it's dangerous to rely on that. This is the block size and fee market debate all over again.

If you had a choice between a block size increase (soft or hard fork) and a drivechain, which would you choose? Why?

[NotATether]

Drivechains, extension blocks, and similar mechanisms are block size increases by another name. They offload transaction throughput, bringing cheaper fee costs to users. Would the base layer actually be worth more to miners? That depends if overall combined throughput increases enough to account for the cheaper fees. We simply don't know, and it's dangerous to rely on that. This is the block size and fee market debate all over again.

Currently the transaction fee is only a small percentage of the money that miners make, the rest are from rewards from mined blocks. BIP-301 is one of the whitepapers for drivechain and the one that connects bitcoin's consensus to drivechains.  By that I mean a miner needs to pay to mine drivechain blocks.

I believe the reason the transaction fees become smaller for users is that miners won't have to validate drivechain blocks or do PoW on them. Drivechains are trying to solve a hypothetical scenario where miners are keeping track of future validated blocks. I don't think this situation is actually happening in the wild because the vast majority of miners are owned by mining companies who would have to order the employees to make the miners send extension blocks to each other. So it's motivation seems futile to me, preparing for something that may not be happening.

[mda]

Drivechain is a nebulous structure that few can explain and almost none cares about. There is a simple working construction of multiple altcoins/shards with common PoW and unified interface. Atomic swaps between altcoins/shards can be implemented natively to bypass centralized exchanges:

https://bitcointalk.org/index.php?topic=5109561

[Wind_FURY]

I'm not particularly comfortable with the precedent of miners forking in sidechains, on which users blindly trust those miners.

But full nodes secure the network, not miners.

Bitcoin nodes only SPV validate the drivechain. That's the security model. Full nodes may enforce the drivechain rules at the Bitcoin consensus level, but that can't stop miners from stealing drivechain funds. 51% of miners can always steal all drivechain funds, no matter what. The only thing Bitcoin nodes can do to stop them is a UASF after the fact that reverses the theft.

Mining on the base layer could be worth more because it supports all side-chains.

Drivechains, extension blocks, and similar mechanisms are block size increases by another name. They offload transaction throughput, bringing cheaper fee costs to users. Would the base layer actually be worth more to miners? That depends if overall combined throughput increases enough to account for the cheaper fees. We simply don't know, and it's dangerous to rely on that. This is the block size and fee market debate all over again.

If you had a choice between a block size increase (soft or hard fork) and a drivechain, which would you choose? Why?

On the block size increase, none for now, and on Drivechain, clearly you know my answer is "I don't know", I'm still learning about it the hard way. By debating/playing devil's advocate.

[squatter]

If you had a choice between a block size increase (soft or hard fork) and a drivechain, which would you choose? Why?

On the block size increase, none for now, and on Drivechain, clearly you know my answer is "I don't know", I'm still learning about it the hard way. By debating/playing devil's advocate.

Let me ask that question a different way.

What do you hope to achieve by forking a drivechain into the consensus? Offloading transaction throughput from the mainchain, cheaper fees? Altcoin interoperability?

I think economically important drivechains may skew Bitcoin's mining incentives. I'm not sure. I'd like to see more convincing game theory suggesting drivechains are a good idea before considering enforcing them at the consensus level, that much is certain.

Why? Mining on the base layer could be worth more because it supports all side-chains. Plus if a side-chain has value, wouldn't the game-theory/incentive-structure be the same as in Bitcoin?

I was thinking about this a bit more.

Consider this situation presented by Adam Back, except let's take it to the logical extreme. Say we're a couple decades down the road, the block subsidy is tiny and fees provide almost all block rewards. Mining has a 5% profit margin and drivechains provide 75% of total revenue.

This implies a super majority of transaction activity -- and probably a majority of bitcoin-denominated value -- is on drivechains. Meanwhile, a majority of miners can steal all funds held on drivechains at any time. There is nothing Bitcoin full nodes can do to stop them.

Is the incentive structure still the same as now?

[Wind_FURY]

If you had a choice between a block size increase (soft or hard fork) and a drivechain, which would you choose? Why?

On the block size increase, none for now, and on Drivechain, clearly you know my answer is "I don't know", I'm still learning about it the hard way. By debating/playing devil's advocate.

Let me ask that question a different way.

What do you hope to achieve by forking a drivechain into the consensus? Offloading transaction throughput from the mainchain, cheaper fees? Altcoin interoperability?

Drivechain side-chains can be any of those. Big blocks, privacy, smart-contracts, big blocks-with-privacy-and-smart-contracts. Hahaha.

BUT with the Bitcoin blockchain as the secure, censorship-resistant, base layer.

I think economically important drivechains may skew Bitcoin's mining incentives. I'm not sure. I'd like to see more convincing game theory suggesting drivechains are a good idea before considering enforcing them at the consensus level, that much is certain.

Why? Mining on the base layer could be worth more because it supports all side-chains. Plus if a side-chain has value, wouldn't the game-theory/incentive-structure be the same as in Bitcoin?

I was thinking about this a bit more.

Consider this situation presented by Adam Back, except let's take it to the logical extreme. Say we're a couple decades down the road, the block subsidy is tiny and fees provide almost all block rewards. Mining has a 5% profit margin and drivechains provide 75% of total revenue.

This implies a super majority of transaction activity -- and probably a majority of bitcoin-denominated value -- is on drivechains. Meanwhile, a majority of miners can steal all funds held on drivechains at any time. There is nothing Bitcoin full nodes can do to stop them.

Is the incentive structure still the same as now?

Game-theory. Why would a miner try to "win" the competition by destroying their source of income? Plus, isn't that the situation in most altcoins? What's stopping them?

[aliashraf]

A 51% attack on Bitcoin only allows miners to double spend. On a drivechain, they can just steal everything.

It has been said ever and ever, still a very misleading and wrong assertion.

Although I'm not a fan of two-way-pegged off-chain solutions being drivechain/sidechain or something like LN, It doesn't look a good criticism by any means to me. Honestly, I suspect the people who first brought up this argument, could have little if not zero good faith. No offense,  I understand you are just re-hashing the same assertion they make in any situation when it comes to discussing anything other than their stupid agenda.
Bad criticism is more concerning and dangerous, sometimes, than the original idea being criticized, as it would have worse consequences than simply accepting the idea blindly, agreed?

Ironically, the same people who proposed and advocated two-way-pegged-sidechains in the first place are the ones who are spreading FUD and fake claims about the idea. Why and how? Let's leave this for the next generation of crypto-historians and journalists and stay focused on this miner-phobic assertion you are rehashing here:

1- Resisting double-spend attack is the single privilege of bitcoin. It is merely a decentralized, trustless solution to the double-spend problem:

Abstract.  A purely peer-to-peer version of electronic cash would allow online payments to be sent directly from one party to another without going through a financial institution.  Digital signatures provide part of the solution, but the main benefits are lost if a trusted third party is still required to prevent double-spending. We propose a solution to the double-spending problem using a peer-to-peer network. The network timestamps transactions by hashing them into an ongoing chain of hash-based proof-of-work, forming a record that cannot be changed without redoing the proof-of-work.

Period.

2- At the time of this writing and for the foreseeable future there is no double-spend threat to bitcoin, and it is why in its worst days, bitcoin holds a whopping 100 billion dollars market cap.

3-I'm well known for my strong opposition to the current situation with pools and ASIC manufacturers, but it doesn't imply that I think there is any chance of double-spending in bitcoin, otherwise I wouldn't waste a minute counting on bitcoin and would have been criticizing it as a failed project.

4- A hypothetical collusion by miners against a sidechain in the future, will make my above argument void as it proves the mere existence of a 51% that exposes bitcoin to double-spend and censorship threats. This would be apocalyptic and thanks god, it is not considered feasible as far as the market and the community are concerned.

5- Suggesting such hypothetical miner-collusions as "a security hole" or "something concerning" for sidechains is absurd too as it is applicable to every single two-way-pegged solution, the most distinguished one being LN. e.g. they could selectively censor/nullify anti-cheat punishment transactions in favor of their own fraudulent behaviors in the network and our superhero full nodes would have absolutely no clue about the existence of the problem, forget about being helpful.

I suppose such criticism is political rather than game-theoretical/technical, hence strongly recommend not to follow this trend, no matter who is backing it.

On the other hand, this is meaningless to suggest any kind of proof in such systems, other than (SPV like proof of) work. Securing the "reclaiming phase" in sidechains by relying on or even having involved full nodes, would existentially nullify the off-chain idea, either it is the traditional verification or zk alternatives.
Zero-knowledge proof is decent technology, but once it becomes practical, it could be employed on the mainnet as well, causing a radical shift of concerns.

[squatter]

BUT with the Bitcoin blockchain as the secure, censorship-resistant, base layer.

Forking a drivechain into Bitcoin's consensus makes it part of the base layer. That's the issue.

Game-theory. Why would a miner try to "win" the competition by destroying their source of income? Plus, isn't that the situation in most altcoins? What's stopping them?

Altcoins get 51% attacked all the time, and that's just to perform limited double spending attacks.

Why don't we fully trust SPV wallets? Because they trust miners to be honest. The same logic applies in this case, since Bitcoin nodes only SPV validate drivechains.

Such an attack would only destroy faith in drivechains, not Bitcoin, so the "Miners won't 51% attack because of long term incentives" argument doesn't apply. Presumably, transaction activity would not only return to Bitcoin, but at even higher fee rates.

A 51% attack on Bitcoin only allows miners to double spend. On a drivechain, they can just steal everything.

It has been said ever and ever, still a very misleading and wrong assertion.

Paul Sztorc admits it is true here.

I didn't realize that he later addressed this by increasing the withdrawal requirement to 13,150 ACKs.

That's one way of sidestepping the whole problem -- nobody will be interested in using this sidechain since it takes 3-6 months to withdraw your bitcoins! Therefore, there will never be enough value to on the drivechain for any of this to matter.

Thus, mis-withdrawals are possible – they just take 3-6 months to go through (as do valid-withdrawals).

A hypothetical collusion by miners against a sidechain in the future, will make my above argument void as it proves the mere existence of a 51% that exposes bitcoin to double-spend and censorship threats.

Sidechain theft is much easier than targeted double spend attacks. That's something you aren't accounting for.

Suggesting such hypothetical miner-collusions as "a security hole" or "something concerning" for sidechains is absurd too as it is applicable to every single two-way-pegged solution, the most distinguished one being LN. e.g. they could selectively censor/nullify anti-cheat punishment transactions in favor of their own fraudulent behaviors in the network and our superhero full nodes would have absolutely no clue about the existence of the problem, forget about being helpful.

The rules for sidechain withdrawals aren't enforced by Bitcoin full nodes. That's the difference. Miners can collude to steal all drivechain funds; it's simply a matter of waiting. This situation does not apply to Bitcoin or LN.

What you're suggesting is just attempted censorship and is probably virtually impossible already. It's completely impossible once we're transacting with LN via Taproot.

[aliashraf]

A 51% attack on Bitcoin only allows miners to double spend. On a drivechain, they can just steal everything.

It has been said ever and ever, still a very misleading and wrong assertion.

Paul Sztorc admits it is true here... increasing the withdrawal requirement to 13,150 ACKs.

No, he doesn't and it wouldn't matter if he does. But you are misreading his faq comment.  I'm not discussing Drivechain project or its devs' opinions, anyway. As of 13,150 ACKs, they became poisoned by the "51% theft security whole" hoax and ruined their project by such stupid decisions. My proposal: let's put it on 200 ACKs and observe that there will be no theft for the next couple of decades.

A hypothetical collusion by miners against a sidechain in the future, will make my above argument void as it proves the mere existence of a 51% that exposes bitcoin to double-spend and censorship threats.

Sidechain theft is much easier than targeted double-spend attacks. That's something you aren't accounting for.

What do you mean by "easier"
Computers are doing the job
It is easy as long as you got the magical 51% relative power!

And FYI, from a game-theoretic perspective, it is a hell "harder" to steal a penny from sidechains compared to the mainnet. Double-spending is a covert operation and mainnet full-nodes are absolutely blind about it, but sidechain full nodes will detect the theft at the moment it is happening.

What you're suggesting is just attempted censorship and is probably virtually impossible already.

It is impossible because of the game theory behind bitcoin and its network hash rate and the market cap. Taproot has nothing to do with it.

Sadly, you are not reading my comments.

[Explodicle]

13,150 ACKs aren't a major problem because we have cross-chain atomic swaps. If you're going to hodl for 6+ months anyways, then why not collect some fees on top of that like a JoinMarket maker?

200 ACKs seems a little bit low. If I was betting on chain split tokens for a UASF, I would want more time to install the sidechain and verify the alleged theft.