When people think about cryptocurrencies and decentralization, they almost always think of Blockchain with it. This data structure had such good PR that the idea of a decentralized platform that isn’t based on Blockchain is nearly impossible for most people to comprehend. As we all know, Blockchain, in its current form, suffers from some serious flaws including scalability and privacy which will probably prevent it from getting the mass adoption we are all waiting for.
Recently, we observed a new movement of projects which aim to create decentralized systems, but without Blockchain. One of the best-known ones is the Hedera Hashgraph project. They are developing a permissionless, decentralized public ledger which is built on the Hashgrpah data structure. The new data structure combined with its Virtual Voting and Gossip Algorithm will allow them to overcome the scalability issue.
We had the privilege of meeting with Tom Trowbridge, the president of Hedera Hashgraph, to talk about the emerging project and how they took an existing company which created centralized solutions for enterprises and converted it into a company that was capable of creating a real decentralized and public ledger.
Could you introduce yourself and the core team behind the Hadera Hashgraph?
The two co-founders are Mance Harmon and Dr. Nieman Barry. Manse has two degrees in computer science and led the War Game simulator at the missile defense agency in the US, and he’s the “non-technical co-founder”. The “technical co-founder” is Dr. Nieman Barry who has a PhD in computer science from Carnegie Mellon which he got in two years and nine months. We believe that this is the fastest PhD in Comp Science at Carnegie Mellon district. He has over 100 published papers in math and holds a lot of patents. They met in the Air Force in 25 years ago where they were basically working for the AI team of the Air Force, and they’ve done multiple startups since.
My background is more in financing, I started off doing telecom and Technology Investment Banking, and then I did telecom technology and media private equity, after that I was in a variety of firms such as Goldman Sachs for about four years and a variety of asset management companies. Ever since my private equity experience, I kept investing personally in technology companies, and so I found these guys and invested in Swirlds the parent company, which developed the Hashgraph consensus algorithm, and then as they started to advance the build out of the public ledger they brought me on board to lead that effort. And it’s interesting because we’re very complimentary even though our networks are entirely different, the education is different, our work experiences are different, but it’s worked really well given the unique abilities we bring to the table.
“Blockchain is designed to be slow, and if it’s not slow it doesn’t work.”
People love blockchains, but blockchain is designed to be slow, and if it’s not slow it doesn’t work. It has great attributes, and it has all the security. But people say okay it’s really secure we like it, now how do we make it fast. That doesn’t really work, and if it does, you have all these security trade-offs.
Dr. Baird decided that there was probably a way to do it without those trade-offs by starting from the ground up, reimagined it, and combined virtual voting with a gossip algorithm, together those two existing on a directed acyclic graph (DAG) resulted in an algorithm limited pretty much only by bandwidth in terms of speed.
And that is unique because it’s incredibly fast, it’s secure, and it’s fair. It involves nodes communicating via a very fast and resilient gossip protocol. You can take nodes out, they can be down, and it doesn’t matter because gossip is propagating randomly, simultaneously and regularly. Gossip protocols and voting algorithms are well-known, the unique piece was realizing that each node could build up a history of the transactions on the network and allow each node to know what all the other nodes know, so you could vote independently without sharing votes over the network.
Effectively you have each node coming to a consensus on its own about all the other nodes and because we all run the same algorithm and because they don’t know what everyone knows they come to a consensus at the same time. And that’s what makes it incredibly fast because you don’t have to wait for votes to be exchanged and you don’t need to have a mining concept in it.
You mentioned Swirlds who created a centralized solution based on centralized databases for enterprises. How was the process of taking this solution and make it public and decentralized?
Swirlds developed the Hashgraph consensus algorithm, which can be deployed in a private permission basis for one company like CULedger, any company or at Hedera where it’s deployed in a public ledger. It’s the same consensus algorithm, the difference between the two of them is that in the private permission basis it’s one node one vote. In a public ledger that doesn’t work because someone could stand up a whole bunch of virtual nodes and disrupt consensus. So what we added was token-weighted voting where it’s not one node one vote it’s one token one vote, and so that allows a public ledger to run the Hashgraph consensus algorithm with the vulnerability only being the accumulation of a third of the tokens by an attacker.
On Hedera (what we call the public ledger) it uses the same algorithm, it’s just token-weighted voting as opposed to node weighted voting. Governance is also different because in a private ledger governance doesn’t matter. It’s one company it’s one thing they do whatever they want. However, in a public ledger governance is a huge issue and we’ve created this Council of companies for governance. The council set pricing and govern the network, in a way that we think is the most diversified governance model out there, because it’s diversified by a company, by sector, by geography, and by time. That is something we’ve seen a lot of in governance models out there. We’ve seen many of the market challenges, some quite publicly and we think this is an enterprise-friendly governance model. We also think it will ultimately benefit users because it will result in a very stable network governed by companies which intend to use it extensively.
“Economics matter tremendously and plays a huge part in most of these decentralized solutions, in ours it doesn’t”
Some will say that Hedera Hashgraph is not exactly de-centralized due to the fact that there are 39 governing members who actually control the network, what is your response for this claim?
What I would say in terms of centralization is that Hedera is one company, but Hedera consists of 39 companies diversified by geography, by sector, and by time and a third of them turning over every year. And if people say that it has concentrated, I would say compared to what? One foundation seems a lot more concentrated to me than that, block producers elected, potentially forever, seems a lot more concentrated than that. So I guess I would look for what is a more decentralized model that actually works, because what you end up having even if you allow anyone in the community to vote you then end up having a very small number of actual active participants who do the voting, so you end up having a concentration just based on participation.
I would say give me a model that is more diversified than what we have set up, and I’m happy to have a discussion on it because we’ve tried so hard to figure out what would be more diversified and it’s not a simple thing, because if you were to have anyone who runs a DAPP can vote, anyone who owns a token could vote. Then what that ends up happening is people who own the most tokens, get the most votes, which leads to concentration. How is that better? That’s economically driven, and it doesn’t sound like how the model should work at all. Our model is completely non-economic, it’s just about these governing members who continually turn over.
The key thing is people don’t think about is economics. Economics matter tremendously and plays a huge part in most of these decentralized solutions, in ours it doesn’t. These governing members don’t have to own any tokens, we’re getting them because they are the best-known companies, but they’re not selected because of an economic interest they have in it. With mining, there’s increasing benefits of scale and that leads to concentration. We don’t have that with our system at all, we actually cap our rewards at certain points to encourage more nodes in the network, so we try to actually flip that economic concentration drive that happens approve of work.
What is your current state in terms of funding?
The private sale and crowd sale are both closed. At some point next year we will release our token, but we are not doing an actual ICO. The token will trade, and we’ll release it to investors and management, and we expect that some exchanges may list it too and they’ll be able to trade.
Could you explain the purpose of the token in the Hedera Hashgraph platform?
I think it’s a unique differentiator, so two pieces of it are relevant. First, it’s token-weighted voting, and the token is how the network is secure because if we didn’t have a token, the network couldn’t function as anyone could stand with a bunch of nodes, have over a third of the network, and disrupt consensus. It’s critical for security, and I think it’s very unusual for tokens. There may be one other one like it. Secondly, the token is used to pay for API calls. That means that every DAPP needs tokens to use the network. What happens then is that token goes in the pays for the DAPP, pays for the API call and then it goes out to compensate the nodes for participating in consensus.
The nodes are paid on the tokens they hold, the more tokens the node holds, the more reward they get. There’s an incentive to hold the token and a node because you get paid for holding the token, and there’s a need to spend the token to use the network, and so you have both of those opposing dimensions at work, and we expect that tension to be a positive thing. Because if you just have a reason to hold it then it’s purely speculation and if you’ve just a reason to spend it there’s no reason to hold it, and the utility isn’t as clear.
You should be able to pay somebody a hundredth of a penny for something. That’s never been possible before.
So how do you see the Hedera Hashgraph three years from now, what would be a successful framework?
The algorithm is very fast because it uses very little data. We believe it uses the minimum data necessary to achieve consensus, which means it’s also very cheap. While a cryptocurrency transaction on Bitcoin is 50 cents or a dollar, Ethereum is between 30 cents and up to 40 cents – who knows? For us, we think the cost is around 6 millionths of a penny maybe that price ends up being a thousandth of a penny or 10 thousands of the penny we are not sure yet. But it’s going to be a tiny fraction, and so for me, initial success is seeing the deployment of significant use cases that take advantage of the micropayment capability that we have, which frankly has never been possible before.
On our network, you should be able to pay somebody a hundredth of a penny for something. That’s never been possible before, and it can be used in IoT, it could be used in media, you could buy electricity by the minute, you could buy songs by the minute, you could do anything that could be metered in an incredibly atomic small way. For me, success is seeing that deployed effectively in ways that we can’t imagine right now.
The reason I say that is because as an example on the Internet in around 1989 or 1990 the first browser was launched.
On average the top 20 websites launched in 2002 10 or 11 years later. Broadband is really what started that in 1999. In 95, 96, 97 we couldn’t imagine Uber or Google Maps. So, there is that potential for business models that we haven’t thought of yet because we haven’t had enough time to see the capability of the technology.
Can you tell us a little bit about your next milestone and if you have any announcement to make to your audience or our audience?
I don’t know if we have announced it but, our SDK has been downloadable for over a year, we got thousands of downloads. We’re doing API reviews now with the DAPPs who have expressed interest in building on us. We’re are slowly releasing the API’s to particular DAPPs and we’re going to be testing them and testing the network and hardening it over the next six months. And so I expect to see more and more use happening over that period of time before we fully release the API in Q1 of next year.
In addition, we just launched Hashgraph.org, a free resource to the community, moderated by Hashgraph Ambassadors and select community members.