Is the Blockchain Too Expensive?
For at least a year now, the magic word in technology circles has been “blockchain” – the accretive cryptographic system behind Bitcoin and other virtual currencies.
A distributed ledger
A blockchain serves as a distributed ledger, a record of transactions or other information that is secure from alteration and that operates without any central authority. Its security derives from the digital signatures required to affect the record. It foregoes a central authority by residing simultaneously on very large numbers of computers on its network (which is what “distributed” means in this context, in contrast with “centralized”.) Each such computer contains the whole chain of transactions, so it is not feasible for a would-be fraudster to alter all of them. A plain-language description of Bitcoin, done a few years ago, is here.
It is clear that Bitcoin, and presumably its emulators, can function to make payments. Serious businesses are willing to accept Bitcoin in payments – even some law offices such as this one. How they establish a value equivalent to the dollars used for accounting and tax purposes, given the fluctuations in value of Bitcoin, is a separate issue. Early adopters have made considerable money. My ISP has earned more from Bitcoin appreciation than from operating profit in recent years.
Many people now say that Bitcoin is only a special use case of the blockchain, which can be used for far more transactions that need to be secure but that now rely on trusted intermediary institutions whose role causes delay, expense and even uncertainty. This week’s newspaper reports that the Canadian Stock Exchange (an alternative stock exchange) wants to record stock purchases and sales via a blockchain. The blockchain is also apparently an “opportunity for lawyers.”
The US National Institute of Science and Technology (NIST) recently published an accessible but thorough description of the blockchain. A legally-oriented description is here.
At present there do not seem to be many actual implementations of blockchains for purposes other than cryptocurrencies. Many thoughts have been generated, many patents issued (most to financial institutions, whose traditional role of financial intermediaries is said to be threatened by blockchain applications.) Not much on the ground – or in the ether, no pun intended – yet.
As a result, it is not quite clear how these applications are going to work in practice. Whether it is a good thing to have an open field of innovation fenced off by multiple speculative patents at this stage is a separate question too.
It has been suggested by an expert observer that any blockchain application is going to have to operate in a well-defined legal framework, including laws of general application, system rules and implementation specifications. It is not just the technology, in other words. The technology has to “be” somewhere.
I raise for discussion the question whether blockchain applications are going to be prohibitively expensive.
Cost considerations – Bitcoin
Start by looking at the primal application, Bitcoin. One obtains bitcoin by “mining” it, which involves solving complex mathematical problems. The problems are so complex at this stage (the complexity increases over time) that they take a lot of computing power to solve. Despite amazing advances in technology and techniques (said to be a hundred million times faster than a decade ago), the amount of electricity needed to drive the banks of computers devoted to mining Bitcoin is also huge.
A common measurement is that running Bitcoins in the world today uses as much electricity as all of Denmark. (One sees Ireland and Israel mentioned in such comparisons too.)
It should be said that some authorities challenge this measurement, or say that improving technology will reduce the demands, or that it is not as expensive as it seems if you consider that it aims to replace entire governmental monetary systems. (Please stand by…)
But as transactions go, Bitcoin is said to be some 5000 times more expensive than a Visa purchase. And Bitcoin can handle under 4 transactions a second, while Visa does nearly two thousand. (Ethereum by contrast may be able to do 20 a second in an ideal state, and PayPal something over 100.)
Moreover, to get one’s transaction added to the Bitcoin blockchain requires payment of a fee, whose amount depends on how busy the system is and how quickly one wants to get the transaction recognized (“to get your transaction processed quickly you will have to outbid other users”) – and thus made irrevocable. There can be backlogs. What this can feel like in person is described in this Maclean’s article from January 2018.
Those fees have been described as “skyrocketing” to something over $20 per transaction. At some point the traditional systems start looking economical. The fees seem to have declined somewhat in the past few weeks, but they are volatile, unlike credit card fees. In any event, payment with Bitcoin is not just a matter of generating a practically free electronic message.
It may be argued that overall, the overhead for Bitcoin is far less than for fiat currency, when you consider the cost of all the intermediaries. The electrical cost of mining Bitcoin comes from the proof-of-work requirement. It requires massive processing times. The processing difficulty, matched to scale in terms of volume, is what generates the security of the blockchain.
The paradox is that when you attempt to reduce expense to make generating tokens more economical, you correspondingly impair the robustness of the blockchain. One reads of so-called “permissioned” blockchains, which depend on trusted “nodes” that govern who gets in and on what terms. The nodes already trust each other, though such a system may lose the trustworthiness that a distributed network of strangers is alleged to create. Sometimes – often? – creating a permissioned blockchain may be a lot of effort – or the application of a trendy label – for a system no more secure or tamper-resistant than traditional centralized databases.
On the plus side, it’s possible that a shared private blockchain (say for securities trading among existing intermediaries) could eliminate some need for disparate back office systems and eliminate reconciliation requirements that cause delay and expense. But these largely remain to be invented.
Cost consideration – blockchains
One important question in determining how a blockchain application will cost is how users gain access to the system. How do they acquire what they have to contribute to the ledger, whether that is a vote as a shareholder, an interest in a contract, or shares in a public company? Does somebody have to set up mathematical puzzles that need to be “mined”? Or is there some other “proof of work” to participate? The latter is most likely – but rather obscure in practice. What work, how done, and proved to whom?
Ethereum is said to offer, or be about to offer, a “proof of stake” system that will address overhead concerns. Permissioned blockchains can make entry cheaper, though the cost of authentication may still be high for large-scale systems.
Will participants in any commercial blockchain need to expend computing and thus electrical resources to get into the system? Will they all have to maintain the records of the whole ledger in their hard drives, as part of the distributed security process?
Will such computing demands be put on participants for every blockchain application they use? So if a person owns shares in, say, twenty-five companies – not an unduly heavy portfolio – must they mine or work their way into every company’s ledger?
If a company has a hundred million shares outstanding – again not an unduly large number – how can all the shareholders be provided with blockchain tokens to vote by proxy at meetings. Perhaps such votes would not be considered by proxy – an intermediary, after all, which the blockchain promises to eliminate – but a direct vote. In any event, how do they acquire a demonstrable right to vote, and authenticate themselves to the system when they vote? Proof of work? If a trusted organizer has to govern how they come and go, is the fancy computing of a blockchain really cheaper or justifiably better than current centrally-managed systems? There does not appear to be any actual proxy voting blockchain app on the horizon.
And all this must be combined, presumably, with a digital signature system to encrypt one’s block in adding it to the chain. Digital signatures rely on public key cryptography, a process that is extremely difficult to apply to a large body. The mathematics can support it, but key management is a challenge, and tends to be centralized rather than distributed.
Would the operation that proposes a blockchain application provide a user-friendly package of signing/encryption software – on a token? Delivered online – securely, to the right people every time? That starts to sound like a system that needs intermediaries, or a great deal of sophistication on the part of users. There are no examples of publicly available, generally usable digital signatures, though the dream has been active for decades.
Will the blockchain voting system, to continue that example, have transactional limits like Bitcoin’s, so only a few voting instructions could be processed per second? Recording votes of 100,000,000 shares for several votes per meeting might take a while.
How much electricity would all that take, including the qualification – proof of work in some way – and the processing both at the voters’ end and the corporate end?
Multiply these numbers by the thousands of applications promised for blockchains – thousands of public companies dealing with their shareholders, but then countless other promised uses as well. Smart contracts, anyone?
Some people – including the founder of Ethereum – contemplate that in less economically developed parts of the world that do not have the intermediaries that are familiar to us, the blockchain can avoid the need to create them, the way that mobile phones have taken the place of banks in payment systems in some parts of the world. Will the demands on power delivery infrastructure in such places cause problems for this ambition?
Inconclusive Conclusion
A large number of very smart people are excited by the potential of the blockchain and are working to develop applications. Perhaps all the questions raised above will be readily, or eventually, resolved. But if a new system offering theoretical advantages is a lot more expensive than a less ideal alternative, the move to the new offering may be slow and its success far from guaranteed.
We have lived through the stock market’s technology bubble at the turn of the century. Some technology companies survived its bursting and thrive today. Perhaps the blockchain is a conceptual bubble that needs to blow up in a lot of faces before its really solid – and affordable – applications become clear.
To add to your sources, john, Patrick Greenfield, in an article entitled ‘Thousands of websites hit by malware mining virtual currency‘ and published in The Guardian, Monday 12 February 2018 indicated a futher are of concern.
Stephen Mason
Great questions.
I won’t try to answer them here because there are so many of them. But I’d like to make an analogy. I hope it will point to the overall theme that answers to these questions will probably follow.
I remember telephony before the Internet. It was so expensive that companies disabled long-distance access on their phones to stop people from making personal calls. Telephony worked through dedicated connections through switchboards. Few big companies or governments controlled those connections and switchboards.
Then someone invented tcp/ip followed by http. Those protocols rewrote communication technology from scratch, without precedent. Eventually the web was born when people started serving html/js files over http.
YEARS later, when the physical layer of the internet was sufficiently upgraded, internet telephony was born.
NOW, who watches the clock when making international calls? Most of them are free anyway. A lot of the time, people make video calls, also for free. When I was a kid it was a sci-fi dream.
When tcp/ip and http were published, they must have been impenetrable. Also, it must have seemed ridiculously expensive to send so much data (especially so much text with http) over copper. Early Internet was expensive indeed. People who said that one day these protocols will support HD video calls must have been laughed out of the room over the squeaking noise of modems.
Yet, tcp/ip and http have hardly changed over the decades. It’s a multitude of layer 2 and 3 applications plus the investment in the physical infrastructure that exploded with the diversity of content or use.
It’s the same thing with blockchain. Bitcoin creators wrote a financial network from scratch. Conventional financial networks took millennia to evolve. Bitcoin is ten years old. It will likely be a base settlement layer facilitating financial applications. Work on layer 2 has already begun: see the Lightning network.
As for non-monetary applications, ICOs are essentially a blockchain-based method for raising capital from anyone in the world. Hundreds of millions have been raised. It’s still early so some ICO broke securities laws or turned out to be scams. But the proof of technology concept is there.
Great questions, John. This post takes a good dive into blockchain technology, and does so deeper than what we usually see on Slaw.
I like how you go beyond payment solutions (e.g. bitcoin/cryptocurrencies) and concentrate on some of the other more important questions:
– how do you scale blockchain tech
– how do you do that securely
– how do you maintain decentralization
– how can higher transaction volumes be accomplished
– how can it be cheap enough when transacting on it involves fees
I agree with Pulat’s “theme” of considering how these questions may be answered. The internet’s ability to support telephony or video was very limited for many years until the protocols and infrastructure evolved.
Bitcoin is first generation protocol, and while it has the network effect and leading position of the most popular/famous blockchain protocol, it is a very basic payment utility — you can’t do much more than pass the magic internet money (bitcoins) betweeen accounts. As you note, it can also get quite costly to handle transactions because of how this first generation technology was built (computationally heavy challenges require immense electrical power). It was never going to handle Visa-level volumes based on its design.
So things evolve.
Ethereum is second generation blockchain tech, and introduces extra functionality — a computing environment that can be programmed to make smart contracts and also other cryptographic tokens (which kind of make Ethereum the daddy of other “coins” which we see propagating like field mice).
You can program contracts using Ethereum with rules and conditions, and these rules are executed on the Ethereum network (fancy name “Ethereum Virtual Machine”). But to get the “machine” to run your rules and take your inputs, there’s an incentive model to deal with. The people who run nodes want to be paid in “Ether” (the name of the magic internet money that Ethereum uses, in contrast to bitcoins) to execute the functions you want to run.
Ether, or ETH, is what folks speculate on and trade, much like they do with bitcoins. But ETH/Ether is also essential as the gas that drives the machine.
And Ethereum is now facing the problems that you outline. Although its protocols do cooler things than Bitcoin can, Ethereum is also a pay to play system, with speed bottlenecks and scalability issues that drive up costliness to transact even simple, routine commands.
So this brings the third generation of blockchain technologies, projects like Cardano, NEO and EOS which are still in development. The third generation will dispense with some of the computationally heavy challenges that consume all that crazy electricity, and also be faster and more commercially viable. They will also have more features (i.e. they will be platforms and programming environments) for developing programs on. They will allow less mercenary “pay to play” environments, let developers control what functions have to be paid for, and just as you don’t directly pay every website you visit to run its webservers and send digital signals to your browser, not every application running on a third generation blockchain platform will nickel and dime users for every little operation.
This will be where the “what can we do with blockchain?” question will be answered. Imagine if Wikipedia was just a giant decentralized application run on the blockchain? That’s literally what Everipedia is doing on EOS.
We often hear of how slow blockchain is because of Bitcoin and Ethereum’s limited transactions per second. Visa, as you note boasts 2,000 per second. Consider that EOS says it will start at 10,000 per second, and move up to millions once. So the problem is not that “blockchain” is so slow, or that it is only useful for making ICO scam coins or speculating on magic internet money, it’s that the protocols for doing everything else that we do using the internet, are just now being built into the blockchain.
another sceptic, with better credentials: https://www.wired.com/story/when-the-blockchain-skeptic-walked-into-the-lions-den?mbid=nl_051518_daily_list3_p3&CNDID=48124250
And yet the Chief Financial Officer of the largest bank in Australia just joined the company (Block.One) behind EOS as its COO. Better credentials still?
The security of the blockchain depends on the wide distribution of the ledger, i.e. that the records are held in too many separate computers for their records to be compromised.
But in a ‘permissioned’ blockchain, there may be fewer computers, because the time it takes to register each entry on each computer can take a long time (thus the smallish number of transactions that blockchain-based systems can handle). As noted in the original post, a permissioning system is intended to avoid the need to “mine” one’s way into the right to participate, but it can also reduce the number of nodes needed to control the system.
It turns out that if the rewards are great enough, and the chain has few enough participants, the bad guys can manage to control 51% of the member computers and thus ‘vote’ for results that line their pockets. This article describes how what used to be thought highly improbable, if not impossible, is becoming “regular”.
Here is an excellent article on the various meanings the world is giving to the blockchain, pointing out why some of them give rise to exaggerated claims – particularly that any distributed ledger implementation has all the virtues that accrue only to a few.
And here is another very good critique of U.S. state legislation that purports to enable or validate the use of blockchain technology in commerce – legislation that is unneccesary and often very stupid.