Constellation: Can blockchains ever move beyond economic models?
Ever notice how the cryptocurrency world just decided that money was a good proxy for trust?
Creating a cryptocurrency is easy. Anyone can create an Ethereum token in about 60 seconds flat. It probably won't be able to do anything, but it'll exist.
The hardest part is creating a cryptocurrency that can actually function and sustainably run on its own blockchain and its own network rather than just piggybacking on top of Ethereum or another platform.
The reason it's so much more difficult is because it means committing to a multi-dimensional balancing act with functionality, security and economics. This is a complete and utter pain because each of the three functions also depends on the other two.
Building a functional decentralised network and cryptocurrency is kind of like trying to build a house of cards by throwing a full deck into the air and hoping they all land perfectly balanced.
- Its functionality (immutable ledger, payment system) depends on the network being secure and bitcoin having value. The network is useless if it's not secure and you can't make payments with a worthless token.
- Bitcoin's value depends on the network being functional and secure. Without functionality and security the token itself has no value.
- Bitcoin's security depends on bitcoin having value. Without token value, there's no incentive for participants to maintain network security or functionality. Without transaction fees, there's little protection against spam attacks, and without economic elements like mining energy costs, there's little protection against Sybil attacks.
The intersection of these three components is human behaviour. It's all about kicking a lot of strangers into the spontaneous motion of securing and operating the network and being able to rely on them continuing to do so. And the most reliable motivator to date is money.
This is why incentive mechanisms are (one of) the heart(s) of decentralised networks.
The trouble with money
Economic incentive models aren't perfect. The main problem might be that it means trying to denominate every element of the user experience, and every kind of motivation of any network participant, purely in monetary terms.
Remember, the real goal is to shape people's behaviour. But as any economist will tell you, human behaviour is shaped by many different kinds of costs and rewards which can't necessarily be accounted for in dollars and cents.
And even within the realm of hard economics, there are effects which can't be neatly covered by ham-fisted payouts like block rewards and transaction fees. There are all kinds of shades of grey in opportunity costs, risk-taking behaviour, a willingness to pay premiums for predictability and different people's economic circumstances which mean everyone who uses a network will be motivated by the presented incentives in a different way.
Purely economic incentives might also have the unintended effects of psychologically stripping away any factors except profitability. The platform invites users to approach it as a cash cow to be milked, rather than as a useful tool, a community of like-minded people or a project under construction. It should be no surprise that it's then treated as such. This alone might help explain why cryptocurrency markets lean so persistently towards speculation and stay so volatile.
It can also undermine the actual functionality of projects in unexpected ways. Just look at what happened to EOS. It brands itself as a very human and community-centric platform, but also uses purely economic incentives. As a result, the network's decision-makers immediately gravitate towards the most personally profitable course of action in any situation, even if it hinders the actual functionality of the network and makes it worse at serving its intended purpose. For an example, take a look at the practice of hoarding and selling RAM at the cost of platform useability.
Purely economic models also introduce some problematic inefficiencies, and there are limits on how finely tuned they can be. As Ethereum founder Vitalik Buterin recently noted, heavily used blockchains tend to see wildly varying transaction fees since the system for "bidding" on block space (higher transaction fees are processed first) results in unpredictable transaction times and fees.
And in the long run, he notes, the stability of these systems is questionable. As block rewards drop over time, the economic pillar of a blockchain fundamentally changes, with corresponding weaknesses in the security and functionality pillars.
"There is at present no good mitigation for this," he said.
The trouble with not using money
Money isn't perfect, but it's not the only way to incentivise people and shape human behaviour, and several networks are exploring alternatives. So far, in the absence of any feasible way of putting Maslow's entire Hierarchy of Needs on the blockchain, reputation is the most popular stand in.
Different networks have developed different approaches.
- POA Network: This is essentially Ethereum, except every single block validator (like a miner) is a known and identifiable person. In this case, they're all US public notaries, which might be thought of as a kind of "real world" equivalent to an everyday trusted authority. You can meet them here. The idea is that transparency and being identifiable encourages people to act in the best interests of the network.
- Parity Ethereum PoA configuration: This is also essentially Ethereum, but it is intended for private networks for business purposes, where it can be set up much like a network administrator sets up a work network and assigns others administrator privileges.
- VeChain: This is a hybrid system that uses both money and reputation. It motivates and punishes participants with the gain and loss of both. It might be thought of as China's social credit system as a consensus mechanism. Participants get a social score, and there are definite financial impacts for failing to comply with the system's authorities.
All of these systems have troubles of their own. In many cases, it's not all that different from direct financial incentives. POA Network, for example, is essentially just Ethereum and all the economic problems that entails, except with more carefully selected nodes. And VeChain is a bit on the dystopian side.
One of the sticking problems is separating reputation-based from monetary incentives. As soon as reputation is worth anything in a given system, it starts carrying monetary value and you're basically back to square one. Plus, the actual network resources required to maintain a system in the form of electricity and computing cycles have a monetary value in their own right.
This makes it seemingly impossible to find a way of balancing those three pillars – functionality, security and economics – in a system without monetary incentives. The security and functionality pillars just fall right over.
A project called Constellation, however, is assembling those pillars in a new way.
Constellation is a DAG system rather than a traditional blockchain, aiming to create a platform for decentralised applications with extremely high scalability and zero fees. Several other projects are going for the same thing, but Constellation is the only one aiming to do it with a reputation-based system.
Constellation VP of engineering, Ryle Goehausen, sees this as a necessity and a natural next step in the development of decentralised networks.
The goal of all decentralised networks, he points out, is to create trust of a sort, in the form of trustlessness.
"Fundamentally, the goal of decentralisation is to provide securitisation of a network by spreading out the risk among enough participants with auditing such that we can trust the results of the network operations," he explains. But "neither [proof of work or proof of stake] really addresses the requirements of a truly decentralised network."
The real goal is to instil certain behaviours in network participants, he explains. At its base, trust is what makes a system tick, and money is just one way of encouraging and quantifying that trust.
"Proof of stake obviously has some benefits [over proof of work] in the sense of not wasting resources and allowing more scalability – and in some sense might be considered easier to decentralise, but the only reason for that is that it uses a mechanism similar to a trust-based network, with the gross oversimplification of defining money as equal to trust," Goehausen says.
"If you were asking a layman to evaluate the trustworthiness of someone based on how much money they have, how do you think they would respond? It would be considered absurd. I'm not going to trust someone I've never met before simply because they're rich. In fact, the reverse would most likely be true. Those with great resources are generally viewed with far more suspicion, especially considering the history of malfeasance in the financial industry at large."
And in the long run, there might be no way for any economically-driven network to go except centralisation in the hands of the wealthy. Most PoS systems already carry fairly steep admission fees for stakers, but this element of centralisation is widely ignored for the sake of convenience.
"Ethereum is essentially proposing the network to be controlled explicitly by wealthy interests – which is completely antithetical to the idea of decentralisation," Goehausen says. "Conventional mechanisms already exist for establishing trust between people. We want to create a way to digitally capture and quantify the existing trust in the world and use it to secure the network."
How to trust
When you're trying to carefully balance those pillars in a sustainable way, without using money to prop up the economic pillar, a good starting point might be the problem of network resources. You need people to put tangible resources, in the form of time, electricity and computing cycles, into the network, and they're probably not going to do it for free.
Constellation gets around this, in a fashion, by biting the bullet and giving people monetary rewards in the form of system reputation and the native constellation token. But it's not intended to be a permanent fix, and it's directly proportional to expenditure.
"It would be silly to expect people to use electricity and compute resources to run our code without some form of payment," Goehausen says. "That is the entire motivation of the reward system – it serves no ulterior motive. The rewards are directly tied to participation in the form of resource expenditures, along with a small reasonable incentive on top of the gross cost of running the node. Our validation program is designed to last 10 years, and doesn't require that you 'get in early' to see a reward. The reward mechanism is only designed to cover the costs of computation; it doesn't funnel money towards the 'top' of trust, but rather fairly and equally pays operators."
The incentive program is set at that 10-year period to avoid locking the token into an endless cycle of inflation, while still giving the network time to grow. Those who participate and start providing network resources to the system will not only be compensated financially at first, but will also get to build their reputation in the system. The goal is to eventually flip from the economic to the purely reputational system.
How much is reputation "worth"?
Reputation is everywhere. It's just not usually quantified.
When you read a positive review of something, its reputation grows, and when you read a negative review, its reputation shrinks. When you're on a forum, the users who have been there the longest will often command a little more respect. When you're trying to judge whether someone on Twitter or Reddit is a robot, you might look at their followers or how long they've had the account.
All of these carry monetary value. People buy reviews, people buy highly regarded social media accounts and more.
The hard part of building a reputation-driven network isn't ensuring that reputation is valued. It's compiling it into a form that can be functionally used to maintain the network without the monetary value itself eclipsing the reputation's function as... reputation.
"The idea that reputation in general 'isn't worth it' is silly when you consider things like Yelp, or the Apple AppStore/Google Play Store," Goehausen says. "Reputation drives massive profits to developers who produce high quality products, and it's a completely natural system that is already widely used."
The goal is for the Constellation network to grow over time and eventually become sufficiently widely used that users can profit from their reputation. In this way, reputation can still carry the requisite weight to maintain a system, without all the downsides of hooking economic models directly to the network. Rather than being paid directly in monetary value, users can be paid accordingly and in different ways with adjustments to their reputation score.
As another advantage, the actual definition of reputation, and the signals which qualify as reputable or disreputable in different situations, can be programmed into a system. This might functionally be enormously valuable, relative to a more monolithic wealth-driven network.
On Constellation, someone who wants low latency for their dapp might get the network resources from someone with a proven reputation for low latency, or someone who wants real-world cachet might look for a suitable provider.
"We're following a model (along with several other recent trust/reputation based coins) which is designed to capture as much real-world information/label data as possible in terms of defining what trust across a network is. We explicitly want to allow people to assign trust to people (nodes) that they know (either secretly or publicly), and use machine learning to help fill in the gaps of missing data," Goehausen explains.
"In a perfect world, we would know everyone, who they trust, and how much they trust them. In reality, we will only know a subset of that. The deterministic reward criteria which pays out reputation based on publicly observed node behavior/validation participation is only a part of a larger model for determining trust – and it is designed primarily to help respond to sophisticated DDoS/network attacks. It is not our sole criteria in running the network."
The possibilities are endless, and allow for the creation of networks within networks within networks – or solar systems, if you will. This will let people gravitate towards the same apps or providers that their friends are using or that other people whose judgment they trust are using.
"The idea of allowing people to form reinforcing layers of trust through their social connections is incredibly important as well (i.e. you would trust a service more if your friends knew the creators of it and you trusted them transitively), and will factor into service reputation scores as well. Our reputation is not just 'one' thing, there are many sub-definitions and applications of it."
Will Constellation's system work?
But it is worth considering the limitations of strictly economic models, and the potential for creating useful systems and adjusting behaviour in different ways. And as Goehausen points out, the idea of boiling everything down to money might be a whopping step backwards given the potential of decentralised systems.
"Relying on money solely as your source of trust in the world would be a terrible idea if it was applied to other areas of life," he said. "People just seem to view this in blockchain as some sort of magical improvement because it doesn't require a PoW – a pretty low bar."
With that in mind, people absolutely do rely on money as a source of trust in the real world, and luxury brands and signs of wealth are thought to explicitly function as signals of trustworthiness. Humans have always been about the money as long as money has been around.
Whether that's a good reason to stick with strictly economic incentive models in decentralised systems or a really good reason to try to shed them as quickly as possible the way Constellation is might be down to personal preferences.
Disclosure: At the time of writing the author holds ETH, IOTA, ICX, VET, XLM, BTC, ADA
- Bitcoin Cash angles for adoption with HTC Exodus, LivenPay partnerships
- Inside the blockchain coup at the heart of the travel industry
- Ethereum, tZero advisor and CasperLabs, Alchemist co-founder Steven Nerayoff arrested for alleged extortion
- Hedera Hashgraph goes live with 500+ companies using it
- Bitcoin experiences existential crisis at Baltic Honeybadger Bitcoin conference