Emin Gün Sirer is a computer science professor at Cornell University and co-director of The Initiative for Cryptocurrencies and Contracts (IC3). Sirer created a P2P proof-of-work digital currency (Karma) that preceded Bitcoin by six years, has played a significant role in the development of core consensus protocols, and has previously alerted the cryptocurrency community of previously-undetected security flaws in various protocols.
I spoke with Sirer in his office about increasing blockchain adoption in the general public and how groups within the crypto-space have changed their ideas around the immutability of code as these systems continue to evolve to solve real social needs and issues.
Sirer talked about his concerns about the low level of decentralization in Bitcoin and Ethereum, the enormously wasteful energy consumption of proof-of-work systems, and about a pseudonymous-published whitepaper about a new class of consensus mechanisms that could make blockchains much more efficient. He also spoke about hurdles to widespread public of blockchain tech and how he hopes blockchain projects will directly address the opaque business practices and existing entities that led to the 2008 financial crisis.
Aaron Fernando: To start, I want to get into your background — the stuff that got you interested in this field in the first place.
Emin Gün Sirer: My foray into cryptocurrencies started with peer-to-peer systems. Around the time when Napster came out, there was a big problem in how to construct healthy peer-to-peer systems. It appeared that most people would just join the swarm of people sharing things. But they would take, and they would not give.
That’s very common in human societies, that’s what people do. And so there were lots of different solutions to this problem, it’s called leeching. It’s a technical term, but you know amount the wider population it’s known as The Tragedy of the Commons: when you take things from society without contributing back in return.
So, BitTorrent was a response to this. In BitTorrent, there’s a barter economy. You give me a block and I’ll give you a block.
My solution was, look BitTorrent may or may not work. It’s much better if we give people a currency. So, whet don’t I invent a currency that can be created out of the thin air — it’s worthless — but you can use it for swapping blocks. So to get a block, you have to pay me in this currency and then to get the currency you have to provide blocks to other people. That’s what begat a system called Karma. It has proof-of work. It came six years before Satoshi did.
“What is being referred to as the blockchain revolution is really the process of figuring out how to make computers work in a setting where people distrust each other.”
So I was quite early in this space at that time and I’ve always been interested in self-organizing systems, peer-to-peer systems and so forth. Karma was very early and, of course, Bitcoin came along and when Bitcoin came along, my initial reaction was let’s take a look at this, let’s see what’s going on, let’s understand at its security properties. So we did this work with Ittay Eyal, [titled] Selfish Mining and I found that a lot of the things that people say about Bitcoin were not true.
It wasn’t secure under the conditions that they thought it was secure. And we characterized when it’s secure what had to be done to keep it secure and we also gave them patches against people who might do selfish mining. And then I worked on security and scalability and so forth. The rest is a ton of work that happened afterwards. So that’s my early foray into this.
What specifically got you involved and interested in the currency usage? Because that’s one of the many things you can do with a blockchain but it seems like your focus tends to trend toward the currency and financial services.
I guess it does. So here’s what we see: computers came onto the scene maybe thirty, forty years ago in mass. They’ve been around for longer than that, but [that’s when] they were accessible to a wider audience. And they made inroads into various different industries and sectors. They made inroads into science. All of the physicists were using computers before anybody else. But business did not really adopt them and Wall Street really did not adopt them. If you look at the history of banking, ok fine, they have databases. But they didn’t really know how to make connected machines work. So what is being referred to as [the] blockchain revolution — it’s really that process of figuring out how to make computers work in a setting where people distrust each other.
That’s an amazingly powerful idea in a very challenging space. I find myself drawn to it because it’s challenging, it’s difficult, it’s nuanced, lots of things can go wrong and there are lots of glutches and counterintuitive results. If you just design something by gut feeling, it will not work.
Also, we all lived through the 2008 crisis… That made everybody into, necessarily, and expert in the economy. You had to figure out, Oh my God, credit default swaps are something that is affecting me on a daily basis, and it’s changing the way I have to do things.
Bitcoin was a reaction to the craziness of those times, [with] the way it dealt with money supply, and so forth. For me, at a personal level, yeah it piqued my interest in, “What’s happening? Can we build financial systems that are more transparent, that are more easy to audit, and can we built a better finance structure that’s built with a different aesthetic?”
I’d like to go on a slight bit of a tangent, and this is important. There are multiple different ways of doing things in computer science. And these ways… the only way to describe it is ‘aesthetics’ is the right word. You can build really complex, clunky monolithic, closed software — and we have systems like this. The most well-known one (people who are over the age of 60 will know this ) iss an operating system called Multics. It’s clunky, it’s big, and it did everything. You want it to do something, you told it, it would do it for you. And then, there is this other aesthetic, exemplified by a system called Unix. That’s where everything is a small tool, they mesh together, you could see the output of every tool and you could check things and so forth.
“…they can’t self-govern. It’s not clear they know what they’re doing at all.”
So part of what’s going on is that that second aesthetic being brought to bear to Wall Street. These guys were building really complex, really opaque structures. Nobody could tell what the hell was going on behind the scenes, except on the occasions they fail. If you look back, on average every eight years these guys fail. There’s an actual crisis in the market every eight years, in some portion of it.
It’s actually more frequent, when you look at the smaller crises. It’s like multiple a year.
Undoubtedly. If you start looking across countries, absolutely. So it’s incredibly complex, they can’t self-govern. it’s not clear they know what they’re doing at all. They themselves will admit that they have these cycles — these boom and bust cycles — and it’s very difficult to keep tabs on what’s happening. Anything that we can do, as technologists, to make this process as opened-up to the common person — provide some accountability, provide some transparency — that’s a good thing to do.
Regarding, behavioral incentives: I pay a lot of attention to the crypto world, but also the sharing economy/ new economy and I’m seeing very, very different ideas about what motivates a person, in both.
I find that with one world, the view is very altruistic and that doesn’t tell the whole story. And in the other, it’s very profit-driven and that also doesn’t tell the whole story.
Exactly. The nice thing about the profit-driven one is that if you have an idea there that makes sense, then there’s somebody who’s going to jump in and try to implement it — so adoption is very, very quick there.
But as you can see, they don’t always lead to good societal outcomes. Tragedies of the Commons, you have people exit scamming, and lots of fraud in that space.
On the flipside, in the sharing economy — the new economy and the entire set of values exemplified by the local currency movement — they are supremely nice… in that they are trying to maximize societal welfare. That’s a good thing, but they don’t always get traction because, how do you actually get going? They have a bootstrapping problem. They find it difficult to raise money.
So those two worlds are kind of at opposites. Deep down, the underlying ethos is different. In one, you have these highly individualistic, highly profit-driven people and they want to make money and that’s the prime objective for them. And on the other, you have the exact opposite type of people. They want to make the world a better place and they don’t care about personal monetary compensation, typically.
But at the intersection of these two worlds are fantastic ideas. If you can come up with something that is easy to bootstrap, that does have some incentives built in for the people operating the schemes, and also makes the world a better place — then we’re talking. And we’re beginning to see such projects. There are some built on the Ethereum platform for microlending for example, or for identity management, or for a bunch of other things like this that are maybe a little to primitive, a little too early, or a little to basic a building block yet, as opposed to a complete product.
“In the early days of cryptocurrencies, we had a lot of people proselytizing about how these systems are immutable, unchangeable… As we saw with various different blockchain projects when humans decide that these things are not serving them, they change them.”
We take Augur and Gnosis — prediction markets. They could go the wrong way. They could go towards betting parlors, which I think would not be a good outcome. Anybody who’s seen a gambling addict knows that that’s bad. On the other hand, they could go exactly the right way and provide insurance against bad events that is easily priced; they give you a way to hedge portfolios, they give you an advanced class of derivatives that are not available on any other platform.
Identity on the blockchain is an interesting thing. Again, there you could end up losing control of your personal information. Or we could build systems that don’t reveal who you are except prove that you possess a property. So there’s nothing personally trackable about you, but a website can know that you’re over the age of 25 or whatever it is. A car insurance company can know exactly what bracket you’re in without knowing precisely who you are.
So at the intersection of these two worlds are very nice projects and I’m hoping that these two worlds collide in just the right way to produce things that are good for society, that are good examplars of blockchain technology, and provide some incentives for the operators to stay in business.
In its current state, what do you think needs to happen — and maybe isn’t happening — to provide the groundwork for adoption by non-technical people?
That’s a great question. Almost everybody else you’ll talk to will tell you, “Oh it has to be more useable. We have to have wallets that make it easier.” That’s all true. But let me tell you about two things that they will not tell you that I think are essential.
Number one: the architecture of these systems is incredibly complicated. Therefore, I don’t believe it’s a matter of just [putting] a thin veneer in front of the user. I don’t think the usability problem stems from the lack of bad wallet software — which we have. We have bad wallets. But I don’t think we can possibly have good wallets over the systems that exist today. They’re too complicated.
You definitely need to understand the concept of mempool if you’re going to be dealing with bitcoin. You can’t just pay with bitcoin. It might look like you did, but at the end of the day, somebody will check to see if your transaction was committed into the blockchain or whether it got stuck in the mempool — that’s a phrase, and every bitcoin user knows what that phrase means because they have to. It’s part of the architecture. There’s no using bitcoin without knowing it.
So that’s problem number one: the architecture has to get simplified. Problem number two is the social frameworks — the social architecture around these systems has to change. That current social architecture right now, is based on hodling. It’s based on holding your coin and riding its inevitable rise to the moon and finding other people to buy into your coin and getting rich by selling to them and so forth. That’s a terrible way to be and the sooner we get out of that mode the better.
You had mentioned something in an interview that when you talk about the real value of a currency, that the value really comes from the community that uses it. What might happen when there are many communities using many different things and they might not agree on what’s valuable?
Sure, I think that all of these systems that people build are human-serving systems.
Especially in the early days of cryptocurrencies, we had a lot of people proselytizing about how these systems are immutable, unchangeable. That their algorithms are baked into them and cannot change — and therefore you have some assurance.
None of those narratives are actually true. As we saw with various different blockchain projects that when humans decide that these things are not serving them, they change them. The 21 million limit in Bitcoin got blown through when people started making forks of Bitcoin. So there are many more than 21 million sum total tokens going by the name Bitcoin right now.
What, really, then is the constant that you can rely on? What is that you have assurance in? That assurance comes not from any software but from the community itself. If there’s a community out there with your shared beliefs about how this money should be governed, those shared beliefs are what you can count on. Those and those alone, nothing else. It is not the software that protects you. It’s the community. The security of the coin is really something that emerges out of the community itself.
So that, I think, is the most critical thing to realize about these things: communities are essential. They are what give rise to various incentive structures. A community that’s healthy will have a number of striated tasks and lots of capable people following out those tasks — developers, proselytizers, merchants, various different service providers, and so forth — as with any currency system. There is a lot of infrastructure out there for you to have the green dollar in your back pocket. Same with cryptocurrency. For you to be able to pay at a website, a lot of things have to be there in the background.
There have to be educational materials for people who do development. There have to be a healthy dev community. There have to be multiple dev communities so you’re not at the mercy of a single dev community controlling everything you do, and so on and so forth.
Those are essential, and very few cryptocurrencies actually possess those properties right now. It’s less than three.
In your blog, you mention that sometimes you may have to call out mistakes that may be made in zeal, especially around adoption and things like that. Is there any trend you find worrisome in any of the ecosystems today?
There are many, and I could pick on quite a few. Let’s pick on, as of today what I’ve been thinking about — and quite a bit for the last few weeks — is the energy consumption in proof-of-work systems.
I think one of the reasons the local currency movement has shied away from coins like bitcoin and other proof-of-work based coins is that they are not sustainable. They end up consuming a lot of energy. How much energy? Well, Bitcoin’s consumption alone is the same as all of Denmark’s. If you want to visualize exactly what that is, it’s about two giant nuclear reactor’s worth.
And it’s not the case that the energy used for Bitcoin comes from sustainable sources, exclusively, anyway. So some people have said that this is a way for China to export coal over the internet. They burn the coal in China, they pollute the air there, they mine/ manufacture these Bitcoins that they sell to westerners who then use it for hodling or whatever else it is that they’re doing.
Yeah exactly, it’s where the electricity is subsidized, usually.
Right, some of it comes from surplus hydroelectric power that they have in the Sichuan Province and that’s fine. But undoubtedly, some of it comes from other sources — dirty sources of electricity that they just have and don’t know what to do with. So these are not… this is not a good way to run any system and these miners are not really providing a high degree of decentralization.
We did a recent study that was published in the last nine months or so that examined decentralization in Bitcoin and Ethereum. It found that Bitcoin has, essentially — roughly speaking — about nineteen mining entities. Pools and solo miners combined. So, nineteen of them is almost nothing. It’s not very hard to take those nineteen people and — the task that any one of these folks is carrying out is nothing that different from what you would be carrying out without any of the hardware.
We can build systems for nineteen people to cooperate and they will sequence transactions and validate them just as well as they do with mining hardware. So all that hardware is some kind of a race among the miners themselves, but the sum total decentralization they bring to bear is only nineteen. And three of them make up the majority of the hash power.
So you’re really buying into what in Bitcoin parlance is called three out of nineteen multisig. If those three people accept your transaction, then you’re in. And if they don’t, then you’re kinda out. (Sorry, I’m mistaken, I should’ve said four out of nineteen multisig for Bitcoin. It’s three out of eleven for Ethereum.) So the numbers are pitifully small, and we don’t get much decentralization, and the impact on the environment is huge so we very much need technologies that can make this entire process green.
So there are a lot of consensus mechanisms out there that are in the works…
No, there are not. There are many different whitepapers floating around but all of it comes down to two different approaches, plus a new one that was unveiled last week and I had a hand in the unveiling, so let me tell you a little bit about this.
“It describes three protocols: Snowflake, Snowball, Avalanche and they have a very curious property. They combine the best of Nakamoto consensus with the best of classical.”
There are only two classes of consensus protocols out there. One of them is called classical consensus, and it was developed by people like Leslie Lamport and Barbara Liskov, both of whom have Turing Awards. In classical consensus, the core protocols work by defining who’s in the system — who’s a validator, who’s making decisions — and then agreeing when they’ve made a decision. They have a bunch of protocols deciding for exactly when a decision is committed.
They’re very fast, but as I said they require that we know who’s participating. So they’re not usable for open currencies like Bitcoin, but they are used for permissioned blockchains. All of these company-oriented sector-based efforts — for example, when people talk about putting a supply chain on a blockchain — they are typically talking about using a permissioned blockchain, using one of these protocols. So that’s one approach. There are many different names that it goes under — HyperLedger, Tendermint, DPoS, Delegated Proof of Stake, Proof of Stake — all of these different approaches fall under that category.
And on the other side, you have Nakamoto consensus, invented by Satoshi himself. Satoshi does not have a Turing Award, but he has about $10 billion worth of coins so he’s well-compensated if he should ever go and claim his coins. So those are the only two, or those were the only two until last week [May 16, 2018] when a new paper was released and it came out with a third way of doing things. The paper is called “From Snowflake to Avalanche” and it describes three protocols: Snowflake, Snowball, Avalanche and they have a very curious property. They combine the best of Nakamoto consensus with the best of classical.
In Nakamoto Consensus, of course, it’s all open. Anybody can join. You don’t need to get permission because I don’t need to know who’s participating in the system. Nobody needs to know. You just kind of solve a crypto puzzle and you end up owning some coins. So that’s lovely about Nakamoto Consensus, except it’s not green, it’s not sustainable.
There are also inherent performance limitations. These blocks need to come only so often — say, every ten minutes — which means that your transaction is not in the ledger for ten minutes. And there’s a particular throughput, for Bitcoin it’s only three transactions per second. So this new protocol family — from Snowflake to Avalanche — upends all of this. It says, “Look, we can combine the best of the two. We can give you quick finality, high performance, without having to know who all is in the system, but we approximately agree that’s great. We don’t need to precisely agree.”
We can make it work without any of this non-sustainable proof-of-work mining. We don’t have to burn energy if there’s no decision to be made. It’s a quiescent protocol. So this, I think, is groundbreaking and it’s completely going to change the scene in two years to come.
So the thing I was leading up to with that question was also that… so let’s say there is this technical ability to make the consensus protocol really great — so if Bitcoin is still the most hyped-up currency and then people push up the demand for it, won’t it just continue with proof-of-work and continue to burn through?
Yeah, so essentially the question here is, what benefits are there to be had from… three sources. One, being the first mover. Second having the mine share that Bitcoin has collected over the eight years of its existence — and we all contributed to that. We all worked really hard. I mean, I translated Bitcoin wallet software to my own native language, Turkish. We went and gave talks. I briefed politicians, many different ones on different occasions. So we worked hard to get this thing accepted and into the public’s mind. So we pushed for Bitcoin and Bitcoin is enjoying the fruits of all of that labor.
And so you might ask, “Is this going to continue forever?” And of course the third thing is, “What are the economies of scale?” If everybody is using Bitcoin, aren’t you better off if you use Bitcoin yourself?
“…there is nothing about Bitcoin that makes it a good store of value. That line of narrative is incredibly dangerous.”
There’s this thing called Metcalf’s Law, which says that the value of a system like this is n-squared the number of participants. So you have n people using the system, the value that can get generated is n-squared. This is called a law and Metcalf postulated it. It’s not a law. There’s no proof of it — it’s just a conjecture.
Every indication is that the conjecture is false, that you do not have n-squared interactions in a universe of n people. You are not as likely to interact with everybody else in the system. You will typically go through certain channels and those channels become choke points. There are power laws in effect where some people are people that everybody interacts [with], everybody buys from Amazon these days, right? So people go to those kinds of merchants and you don’t necessarily go and exchange money with your next-door-neighbor.
I don’t believe that the third one is an issue — that is, I don’t believe that it’s n-squared. We have seen Bitcoin and get bigger as a result of the division, so that tells us that that’s not true — that n-squared is not true. If that were true, then by dividing up into two small systems, you would have a net loss, but instead we had an increase in the market cap of Bitcoin after division.
So, that says Metcalf’s Law is completely wrong and so then there are two sources — technical reasons for why the first two issues (the mine share and so forth) might alter and will send people going to other points. Bitcoin itself is technically very hard to use right now. It’s incredibly limited in its scale and its performance and when the network is congested, the fees go sky-high.
People have paid hundreds of dollars to do a transaction in a timely fashion. That’s not a useable system. So we need to make these things incredibly efficient, not like this.
I just want to play Devil’s Advocate for a second, because with the community currency stuff, there’s this distinction between a store of value currency and a means of exchange currency.
Not in my mind.
Well, functionally, there has been historically.
Ok, I’m all ears.
For instance you can… there have been a few instances in history when you could equip a currency with something called demurrage. You might know negative interest but there are ways to do it — even non-technical ways like putting a physical stamp on it. That speeds up circulation.
But you could have this situation in which the demand continues to rise and the valuation continues to rise and then it won’t be the transactional velocity that keeps it continuation. It could just be this idea that, “Well, it’s going to continue to rise,” and then that could be made true, simply because of that belief.
Ok, so let me very clearly try to delineate why the store of value narrative is incredibly dangerous when applied to cryptocurrencies. Everything you said about past forms of money and past forms of money and past stores of value is true. There are things that are universally valued, that you know there exist a constant level of demand, so you can treat them as a source of value.
It’s like if you have grain tucked away some place in safe storage, you know how much demand is going to be there for it, year after year. So that can provide you that safety net — the value you can trade that for is pretty much set.
When it comes to these online currencies, they’re really all a bunch of funny numbers. There is nothing inherently valuable about them. Nobody will want any of these funny numbers and all of the these funny numbers are fungible — they are about the same. So if you have Bitcoin vs. Bitcoin Cash vs. whatever else, those things are not all that different from each other in the function they serve.
So there is nothing about Bitcoin that makes it a good store of value. That line of narrative is incredibly dangerous. It has been used to sell Bitcoin to the masses who don’t understand where that value is going to come from.
That value can only come from other people buying into the system. And so what you’re really doing is you’re betting on somebody else coming in. You’re betting on riding the adoption curve upwards. At some point, there will be no more people to jump into this system and it’s very, very similar to an inadvertent pyramid scheme — without a single beneficiary. So essentially what’s going on is in a distributed, decentralized fashion, these people are trying to jump into the coin early to benefit from people who jump in late.
That’s no good way to make money. It’s not societally good, it doesn’t produce actually value in a tangible form, it doesn’t make anybody’s life better. These coins are useful when you can send them across the internet, when they serve as a payment mechanism. The coin that you hold and don’t spend — can’t spend (or can’t spend without paying exorbitant fees) is a worthless coin.
That’s what we saw with Bitcoin over the course of the last year. When the fees went up, as we predicted they would, people stopped using Bitcoin. Bitcoin’s dominance dropped because people did not want to pay those extreme fees. Its folly to invest in these things because they are seen as valuable. There’s nothing that makes them inherently valued.
What do you see as the non-financial use-cases that are most primed to take hold, and what ones are you most interested in and excited about?
I mostly work at the infrastructure layer, so I’m application agnostic. So when I do research, most of my research is something that is applicable to all blockchains across the board.
But having said that, I am most excited about blockchain applications in insurance, because insurance companies are opaque, monolithic beings and they tend to go under. You can’t really see what’s going on inside them. The regulation around them is cantankerous to say the least. So that is ready to be disrupted. Those monoliths will find themselves facing an existential crisis unless they adapt.
I’m also very excited about this new idea that we are calling autonomous blockchains, where two people can engage in secure communications and secure data sharing without recourse to a public blockchain — without having to make any of the information public and without a private blockchain, either, without having to designate third-parties to hold their data. So you and I, or me and anybody else who is part of the system can start running nodes that we ourselves cannot tamper with. That provides the necessary basis for trust. That provides auditability and transparency that we seek in blockchains.