Fetch.ai’s staking model explained

Jun 21, 2019

We’re delighted to be able to announce plans for staking in the run up to the launch of the Fetch.AI main network at the end of this year. In an earlier article, we provided the key information that our supporters will need to participate in the staking programme. The purpose of this article is to provide some background on the role of staking on the Fetch.AI platform, the rationale for the design of the staking model, and how staking will evolve as the Fetch.AI network grows in terms of user-participation and the sophistication of the ledger.

Staking is similar to being paid interest on a bank account, except that individuals are rewarded with FET tokens for agreeing to lock their stake for a certain amount of time. The operators of validator nodes (computers that run copies of the ledger) lock their stake to gain entry to the Proof-of-Stake consensus. This then earns them the right to earn rewards if they follow the protocol correctly.

In the following sections, we use a question-and-answer format to explain other aspects of the staking programme.

So what is Proof-of-Stake and why do we need it?
The idea behind any Proof-of-X (where X is “Work” or “Stake” or some other valuable resource) is that the nodes that maintain the blockchain need some mechanism to identify valid blocks. After all, it is very cheap and easy for an attacker to write a program that outputs plausible-looking lists of transactions and other data that could go into a block. Without some type of penalty, an attacker could then flood the network with these fake blocks and stop the blockchain from operating. This is avoided in Proof-of-X chains by requiring block producers to provide evidence that is easy to verify (a proof) of a significant amount of resources being expended in the block’s creation. This makes it expensive for an attacker to disrupt the blockchain with fake blocks. In Proof-of-Stake chains, the expense is provided by the node “locking-up’’ a significant amount of funds for a certain amount of time. The private key that is used to lock these funds can then be used to append digital signatures to blocks to prove that they were created by the same individual. Invalid signatures can be ignored, while valid signatures can be used to allocate rewards for good behaviour or levy penalties for misconduct via the withholding of stake.

Bitcoin uses Proof-of-Work. How does that differ from staking and why is staking better?
In Proof-of-Work, a difficult computational puzzle is used to check that a miner has expended significant resources in producing a block. This proves that a substantial sum of money has been spent (and energy burned), and is the cause of Bitcoin’s well-publicised environmental impact. The main benefits of Proof-of-Stake systems is that they are much more energy efficient and can support higher transaction throughput. Another important advantage is that staking is what economists would call a “homogeneous” cost, which means, in simple terms, that it is the same for everyone. The specialised hardware that is available for Bitcoin mining leads to different costs that depend on which hardware you are using. This is one of the reasons that mining has become centralised in China where the hardware is manufactured and energy is cheap. The homogeneous cost of staking means that it delivers greater decentralisation.

A final advantage of Proof-of-Stake is that it is likely to provide greater security. This arises from the stakes being a resource that is intrinsic to the blockchain, which enables misbehavior to be discouraged through the withholding of stake. In contrast, the hashing power used in Proof-of-Work chains is an external physical resource that is independent of the ledger, which means that miners cannot be restrained by other participants in the protocol. This is one of the main causes of the bitter disputes that have taken place after forking of the Bitcoin blockchain.

You have a sharded blockchain. What is that exactly?
One of the main weaknesses of existing blockchains is that they have limited throughput, which is caused by bottlenecks on, for example, how fast the nodes can execute transactions or send large messages to each other. It’s possible to avoid these bottlenecks by breaking these tasks up into different groups or shards that run in parallel with each other. If sharding is implemented correctly, then the throughput of the ledger will increase linearly with the number of shards thereby enabling potentially unlimited throughput.

So are the validators running a single shard or multiple shards?
At the launch of the main network, all nodes will be running all of the shards. In blockchain terminology these are known as “master” nodes. In later updates of the ledger, we’ll also introduce “light” nodes that run a subset of the shards and have a correspondingly lower operating cost. These nodes will also be able to participate in consensus and earn fees and rewards despite being unable to create blocks by themselves. Additionally, they will be able to host instances of the Open Economic Framework to support deployment of multi-agent systems and earn fees from providing services to agents.

What is the reason for having two hundred master nodes at the launch of your main network?
The design of our consensus protocol and our sharded blockchain make it technically feasible for the ledger to scale to accommodate a much larger number of nodes. However, it’s important to remember that running all of these nodes involves quite a significant operating cost, and that this cost must be paid by people using the platform. To put it another way, there is a trade-off between the number of nodes (and the protocol’s degree of decentralisation) and low transaction fees. A total of two hundred nodes is a major improvement over many other Proof-of-Stake chains, which reflects the high efficiency and performance of our ledger. A restriction on this number also allows fees to be maintained at an acceptable level in the future.

Will there always be two hundred validator nodes or are you planning on increasing the number at some point?
The two hundred nodes will maintain the ledger for a specific tenure, which will have a duration of around a month. At the end of their term, another auction will be concluded, the stakes will be returned or reallocated, and they will be replaced by a new and potentially different set of validators.

The ledger is a work in progress and will continue to develop after the main network is launched. Our roadmap lays out a development plan to improve performance, decentralisation and security of the ledger. These improvements will also enable us to increase the number of nodes that can participate in consensus. It is also likely that other types of nodes will be available for supporting deployment of autonomous economic agents that are not directly involved in consensus.

Why is there a reserve price of 200,000 FET for becoming a validator?
An important aspect of the incentive design is that it needs to provide cryptoeconomic security. If a single entity controlled a majority of the staked tokens (say, 51%) then they would have effective control of the blockchain. We need to set the reserve price at a sufficiently high level that this would be difficult for anyone to achieve. If the number of staked coins is high enough, then the increased demand for tokens will cause the price to rise significantly making the attack progressively more expensive. The reserve price is set in the expectation that at least 15% of the circulating supply will be staked in the first three years after the ledger is launched.

I have less than 200,000 FET but would still like to earn rewards for staking. How do I get involved?
We will be inviting our partners in industry and academia to operate ledger nodes at the launch of the main network. In exchange for locking your stake and delegating it to them, they will serve as validators on your behalf and pay interest on your contribution. Since this is likely to be a competitive process, we anticipate that the margins of these operators will be low and that they will offer very good interest rates to retail investors.

What is the interest rate that I can expect to earn from staking?
We’re using an auction to allocate staking rights to the two hundred validators. The auction will be implemented in a smart contract, and will entitle the winners to receive an initial annual reward of approximately 50,000 FET from acting as a validator. If all the stakes are sold at the reserve price then this will correspond to an interest rate of 20%. In practice, it is likely that the bidding will drive the price higher leading to a lower interest rate. If you are a retail investor this will allow you to delegate your stake for use by a validator at a specific rate of interest.

Are the rewards going to decrease similarly to Bitcoin or Ethereum halving?
For those of you not familiar with halving, an explainer can be found here. Unlike Bitcoin or Ethereum halving, Fetch.AI staking rewards will increase with token supply. This will reward our supporters for holding tokens as the supply and adoption of the token increase. During the first three years of operation (when the majority of the tokens are released), staking rewards will increase to maintain an allocation of at least 15% of the circulating supply to staking. After this period, we’ll be transitioning to a fee-based incentive model as FET is a non-inflationary token that will preserve its users’ value.

How do I find out more?
Additional aspects of the staking program are discussed in these FAQs. I’m also happy to answer questions on Twitter (@jonathan6620 or @Fetch_AI). Alternatively, you’re welcome to post questions on our Telegram channel.