The Decentralist Perspective, or Why Bitcoin Might Need Small Blocks
The block-size limit debate has dominated Bitcoin blogs, forums, chat rooms and meet-ups for months on end, while many of Bitcoin’s brightest minds are gathering in Montreal to discuss the issue face-to-face at the Scaling Bitcoin Workshop this weekend.
So far, however, the two sides of the debate have made little progress coming to a consensus. At least for some part, this seems to result from a difference of visions – visions that are based on a different set of priorities.
One of these visions – represented by Bitcoin XT developers Gavin Andresen and Mike Hearn – is straightforward and clear. For Bitcoin to succeed, they believe it needs to grow, preferably fast. And for Bitcoin to grow fast, it needs to be cheap, accessible and easy to use. This, in turn, means that the block-size limit needs to increase in order for more transactions to fit on Bitcoin’s blockchain, for fees to remain low, and without having to rely on complicated and far-off alternative solutions. This could mean that some aspects of the Bitcoin ecosystem need to specialize further, but that was always to be expected.
On the other end of the spectrum, a majority of Bitcoin’s most active developers think it’s not that simple. For them, Bitcoin’s decentralized nature is sacrosanct, and they believe that an increase of the block-size limit represents a trade-off with this core feature. Some of these developers – perhaps best described as Bitcoin’s “decentralists” – even warn that too big an increase could destroy the system as a whole.
But for many outside this select group developers it still seems unclear why, exactly, big blocks could pose such a grave threat. To find out, Bitcoin Magazinespoke with four of the most prominent of these decentralists: Bitcoin Core developer Dr. Pieter Wuille, Bitcoin Core developer and long-term block-size conservative Peter Todd, hashcash inventor and Blockstream founder and president Dr. Adam Back, and well-known cryptographer and digital currency veteran Nick Szabo.
According to decentralists, the problem of big blocks is essentially twofold. On one hand there is the basic assumption that bigger blocks favor bigger miners – presumably mining pools. On the other hand is the fact that bigger blocks complicate mining anonymously.
The main (though not only) reason bigger blocks favor bigger miners has to do with latency. As the block size increases, so does the time it takes for a newfound block to transmit through the Bitcoin network. That is disadvantageous to all miners, except for the miner that found the block. During the time it takes a new block to make its way through the network, the miner who found the block gets a head start mining on top of the new block, while other miners are still busy mining on top of an older block. So as a miner find more blocks, it gets more head starts. And as a miner gets more head starts, it finds more blocks. Meanwhile on the other end of the equation, smaller miners find fewer blocks and, as a result, have more trouble turning a profit, ultimately causing them to drop off the network. Bigger blocks tend to centralize mining.
This problem is worsened, moreover, if a miner wishes to remain anonymous, and wants to use Tor. Since latency on the Tor network is always higher, the problem as described above is simply magnified. If smaller miners are disadvantaged by bigger blocks to the point where it’s hard to remain profitable on a regular connection, miners using Tor don’t even stand a chance.
Neither of these arguments against bigger blocks is controversial in itself. The controversy lies in the question how big is too big. Andresen and Hearn believe an increase to 8 megabytes or even 20 megabytes should be OK, and assume that it’s fine to grow this limit to 8 gigabytes over a 20-year time period.
Speaking to Bitcoin Magazine, Bitcoin Core developer and Blockstream co-founder Dr. Pieter Wuille explained why this assumption is not shared by decentralists.
“It is obvious that block propagation is too slow already,” Wuille said. “A recent software update revealed that several mining pools maintain arrangements where they share block headers – the minimal required part of an actual block – the moment they find a new block. All miners in on the deal then start mining on top of that block header instead of waiting for the full block to propagate over the network – waiting would cut into their profits too much.”
Explaining why this is a problem, Wuille continued:
“This practice is harmful for Bitcoin, as it requires a lot of trust among miners. They no longer verify the validity of blocks individually, and instead just rely on their peers. But validating blocks is supposed to be a miner’s main task on the Bitcoin network…”
Moreover, decentralists warn there might be no turning back once bitcoin mining has become too centralized. At that point, the remaining miners could have created a deadlock on the mining market, essentially preventing newcomers to compete profitably.
Core developer and long-time block-size conservative Peter Todd told Bitcoin Magazine:
“The big concern we have here is, as we reduce these security margins, we’ll see the already worryingly small number of pools decrease even further. Even more concerning though, is that while currently it’s pretty easy to start a new pool if an existing one ‘goes rogue,’ bigger blocks can make it much more difficult to do so, because from the beginning you’ll be much less profitable than the big pools.”
And while it has been suggested that miners can connect to a VPS if they prefer to remain anonymous rather than connect through Tor, this is not quite satisfying for decentralists either. Speaking to Bitcoin Magazine, hashcash inventor and Blockstream CEO Dr. Adam Back explained why.
“It is technically possible to mine using a VPS (Virtual Private Server), but miners who do so are not choosing their own transactions,” Back emphasized. “Instead, they connect to a server that does this for them. It’s another form of centralization, at the extreme. And we already see this happening due to bad connectivity in some countries, where miners use VPS services set up in another country to win some time and increase profits…”
Decentralization – and anonymity – might be sacrosanct for decentralists, but that does not mean they are goals in and of themselves. Instead, decentralists cling to these properties because they believe the health of the Bitcoin network relies on them. It’s only through decentralization and anonymity that the system can remain free from outside influence, such as government regulation.
“Bitcoin achieves policy neutrality by decentralization of mining,” Back explained. “If one miner won’t mine your transaction, another will. It’s an additional benefit if miners are many, geographically dispersed and anonymous, since it’s complex to coordinate a policy imposition on many small geographically dispersed miners. And it’s even more complex to impose policy on someone who is anonymous.”
If, on the other hand, Bitcoin reaches the point where only a handful of professional miners will be able to profitably partake in the process of Bitcoin mining, and if these miners are no longer able to do so anonymously, decentralists worry that Bitcoin’s fundamental properties might be at risk.
“It is pretty clear that forcing the Bitcoin protocol to implement anti-money laundering policy and blacklisting of funds is a long-term goal of governments, which can be done by pressuring mining pools,” Todd explained. “Being able to tell regulators that pressure will simply cause pools to leave regulated jurisdictions is important, but without anonymity, there really aren’t that many jurisdictions to run to.”
Furthermore, once more that half of all hashing power is effectively regulated, authorities could even demand a complete freeze of certain funds, Back explained:
“If more than 50 percent of mining is subject to policy, it can actually censor any transaction by ignoring – orphaning – blocks made by other miners. We don’t know if that would happen or not, but given the fact that it would be within their technical power to do it, it should be expected that regulators demand their regulations achieve an effect.”
Moreover, once this sort of regulation does set in, decentralists believe it will probably be too late to fix. Bitcoin would be caught in a regulatory trap without even noticing it – until the trap closed.
“If Bitcoin is already at high policy risk – sort of effectively centralized but not experiencing the side-effects of that yet – and then the policy problem arises, the properties of Bitcoin are lost or eroded. How can you fix it at that point? Suddenly decentralize it? It’s uncertain that the parties who are at that point under central control over the Bitcoin network have the free choice to work to decentralize it. They would have been regulatory captured.”
But even mining centralization and regulation might not be the end of it, decentralists warn. Ultimately, over-sized blocks bear with it another – perhaps even bigger – risk: full-node centralization.
Full-node centralization could be an even bigger risk than mining centralization, decentralists argue, as full nodes effectively verify the consensus rules Bitcoin plays by. These consensus rules enforce that Bitcoin has a 1MB block-size limit, but also that the block reward halves every four years, or that the total supply of bitcoin will not exceed 21 million. And – importantly – being able to verify these rules is what makes Bitcoin a trustless solution. In essence, full nodes allow users to check that Bitcoin does as promised.
But as it becomes expensive to run a full node, decentralists worry that verifying the consensus rules could become reserved to a small elite. This could have several consequences.
An obvious consequence would be that it injects trust in the system. Instead of using trustless full nodes, users would, for instance, use web-wallets – which obviously require a lot of trust in the service. But even solutions such as Simplified Payments Verification (SPV) nodes are no better in this regard, as they do not verify the consensus rules either.
Peter Todd explained:
“SPV nodes and wallets are not a trustless solution. They explicitly trust miners, and do no verification of the protocol rules at all. For instance, from the perspective of an SPV node there is no such thing as inflation schedule or a 21 million bitcoin cap; miners are free to create bitcoins out of thin air if they want to.”
And while the cheating of SPV nodes could be seen as a short-term problem, some decentralists argue that a drop in full nodes might even have more severe consequences in the longer term.
According to Wuille:
“If lots companies run a full node, it means they all need to be convinced to implement a different rule set. In other words: the decentralization of block validation is what gives consensus rules their weight. But if full node count would drop very low, for instance because everyone uses the same web-wallets, exchanges and SPV or mobile wallets, regulation could become a reality. And if authorities can regulate the consensus rules, it means they can change anything that makes Bitcoin Bitcoin. Even the 21 million bitcoin limit.”
It is of vital importance for the health of the Bitcoin network, therefore, that it remains possible to run full node anonymously, Todd urged:
“Like mining, having the option to run full nodes totally ‘underground’ helps change the discussion and gives us a lot of leverage with governments: try to ban us and you’ll have even less control. But if we don’t have that option, it starts looking like regulation efforts have a decent chance of actually working, and gives governments incentives to attempt them.”
Commenting on the block size limit debate itself, Back added:
“I believe that the unstated different assumption – the point at which views diverge – is the importance of economically dependent full nodes. It seems that Gavin thinks a world where economically dependent full nodes retreat to data-centers and commercial operation – and basically all users can only get SPV security – is an OK trade-off and cost of getting to higher transaction volume a year early. But most of Bitcoin’s technical experts strongly disagree and say this risks exposing Bitcoin to erosion of its main differentiating features.”
So what if decentralists are right? Bitcoin mining, and perhaps even running a full node, is reserved to specialists working from data centers. Anti-Money Laundering and Know Your Customer policy might be imposed, and perhaps the protocol rules are regulated to a certain extent. Sure, it would be a blow for drug dealers, CryptoLocker distributors and extortionists, but Bitcoin would still be a global, instant and cheap payments system. In a way, Bitcoin might actually be better of without these outlaws. Right?
Well, not according to decentralists.
Most decentralists maintain that Bitcoin’s distinguishing features are not its global reach, its instant transactions, or its low costs of use. Instead, they argue, Bitcoin’s single most important distinguishing feature is its decentralized nature. Without it, there would be no reason for Bitcoin to even exist.
Well-known cryptographer and digital currency veteran Nick Szabo explained:
“In computer science there are fundamental trade-offs between security and performance. Bitcoin’s automated integrity necessarily comes at high costs in its performance and resource usage. Compared to existing financial IT, Satoshi made radical trade-offs in favor of security and against performance. The seemingly wasteful process of mining is the most obvious of these trade-offs, but it also makes others. Among them is that it requires high redundancy in its messaging. Mathematically provable integrity would require full broadcast between all nodes. Bitcoin can’t achieve that, but to even get anywhere close to a good approximation of it requires a very high level of redundancy. So a 1MB block takes vastly more resources than a 1MB web page, for example, because it has to be transmitted, processed and stored with such high redundancy for Bitcoin to achieve its automated integrity.”
The crucial importance of this trade-off, was seconded by Wuille:
“If we were to allow centralization of mining, we simply wouldn’t need a blockchain in the first place. We could just let a central bank sign transactions. That would allow us much bigger and faster blocks without any capacity problems. No variable block times. No wasted electricity. No need for an inflation subsidy. It would be better in every sense, except that it would involve some trust. Really, if we don’t consider centralization of mining a problem, we might as well get rid of it altogether.”
“These necessary trade-offs, sacrificing performance in order to achieve the security necessary for independent and seamlessly global automated integrity, mean that Bitcoin cannot possibly come anywhere near Visa transaction-per-second numbers and maintain the automated integrity that creates its distinctive advantages versus these traditional financial systems.”
Bitcoin versus bitcoin
This leaves us with one last question. If “Bitcoin cannot possibly come anywhere near Visa transaction-per-second numbers” as decentralists claim, then what is the point of it all? Why even bother building software, investing in startups, and spend time evangelizing Bitcoin, if it inherently doesn’t scale?
The point of it all, for decentralists, lies in a classic distinction: the distinction between Bitcoin the network and bitcoin the currency.
Bitcoin the network, decentralists argue, is fundamentally designed as a settlement system, not as a network for fast and cheap payments. While maybe not the most typical decentralist himself, a recent contribution to the Bitcoin developer mailing list by Core developer Jeff Garzik perhaps explains the decentralist perspective best.
“Bitcoin is a settlement system, by design. The process of consensus ‘settles’ upon a timeline of transactions, and this process – by design – is necessarily far from instant. … As such, the blockchain can never support All The Transactions, even if block size increases beyond 20MB. Further layers are – by design – necessary if we want to achieve the goal of a decentralized payment network capable of supporting full global traffic.”
But, importantly, this vision of Bitcoin as a limited settlement network, does not mean that bitcoin the currency cannot flourish beyond these built-in limits.
As explained by Szabo:
“When it comes to small-b bitcoin, the currency, there is nothing impossible about paying retail with bitcoin the way you’d pay with a fiat currency – bitcoin-denominated credit and debit cards, for example, with all the chargeback and transactions-per-second capabilities of a credit or debit card. And there are clever trust-minimizing ways to do retail payments in the works. Capital-B Bitcoin, the blockchain, is going to evolve into a high-value settlement layer, and we will see other layers being used for small-b bitcoin retail transactions.”
Or as Garzik elaborated:
“Bitcoin payments are like IP packets – one way, irreversible. The world’s citizens en masse will not speak to each other with bitcoin (IP packets), but rather with multiple layers (HTTP/TCP/IP) that enable safe and secure value transfer or added features such as instant transactions.”
Moreover, decentralists contend that even these upper layers could include most of the advantages that the Bitcoin network introduced. Once fully developed, technological innovations such as the Lightning Network and tree-chains should allow users to transact in a decentralized, trustless, and even instant fashion – while ultimately settling on the Bitcoin blockchain. While it is true that on-chain transactions will cost more as room in blocks becomes scarce, decentralists maintain that it is the only way to keep that chain decentralized and trustless – and that that does not need to be a problem.
“Yes, on-chain transaction fees will rise,” Todd acknowledged. “But that changes what you use Bitcoin’s underlying blockchain layer for, and how often – not whether or not you can transact at all. A world where you can send anyone on the planet money directly on the blockchain for five dollars – or for near zero via caching techniques like Lightning – is a very good option, and it will buy us time to develop techniques to make blockchains themselves scalable …”