Holo Fuel Economics 101
Alastair Ong, Art Brock, Mary Camacho, David Atkinson
What follows are the slides and speaking notes for a presentation done by Alastair Ong prepared in collaboration with Art Brock, Mary Camacho and David Atkinson.
Welcome. My name is Alastair Ong and I’m part of the Holo team. Today I’d like to give an introductory presentation on HoloFuel Economics.
In this presentation, I will talk primarily about the conceptual design, that is how we want the system to behave and the design choices we make to achieve this. In doing so, the behaviour we are most interested in defining are the economic signals produced under different conditions, and ensuring that these facilitate a healthy ecosystem.
What we care most about is being able to effectively facilitate and record transfers of value across the Holo network, since these transfers of value represent the process by which hosting is provisioned correctly across a decentralised network. To do this, we defined several requirements in the green paper.
From a technological perspective, we need to be able to handle extremely high levels of micro-transactions, on the order of millions of transactions per day, since that is the nature of of how distributed hosting will be provisioned. This requires cryptocurrency-like technologies, since existing payment networks are not able to handle micro-transactions economically.
From an economic perspective, we need price signals that are both stable AND accurately reflect underlying activity, so that users and hosts can make long term plans with confidence. We assert that this requires a currency that is backed by the hosting capacity of the network.
I would then further assert that making it hosting backed requires agent-centricity, and therefore a cryptocurrency built on Holochain, that is: HoloFuel
Let’s dig deeper into that assertion that we require agent centricity. For HoloFuel to be backed by hosting, hosts either individually or collectively have to always be willing to provide hosting services in return for fuel. To make hosting truly backed, HoloFuel should be created through allowing hosts to spend on credit and they must provide hosting services so as to earn the HoloFuel to pay back that credit line. It also means that new HoloFuel is generally created when the network can produce more hosting, and vice versa.
If we want to enable credit lines automatically for hosts, we must understand how credit-worthy they are and be able to enforce it. Doing that requires agent-centricity since it requires understanding the identity of nodes participating in hosting. At this point I want to point out that while this requires the removal of anonymity from hosts, that is needed anyway to comply with global AML/KYC regulations.
Now that we understand why we need an agent-centric HoloFuel, let’s talk about how it works in practice.
As we’ve already discussed, HoloFuel will be backed by hosting in that holders of HoloFuel can always spend it for hosting services on the Holo network, and the issuance of HoloFuel is decentralised and directed by individual hosts decisions related to their pricing of hosting services and and their activity and preferences for the redemption of HoloFuel. In addition, we make 2 other design choices.
Firstly, the flows between purchases of HoloFuel and redemption of HoloFuel, mediated through the aggregation of hosting prices set by hosts and configuration of redemption preferences at reserves, along with platform metrics work to limit the overall amount of fuel proportional to the hosting capacity of the network. This manages the overall risk exposure in the system and ensures that the value of each individual unit of HoloFuel cannot be inflated. Secondly, reserve accounts add structural resilience by amplifying certain economic signals from hosts and acting on hosts’ behalf to sell/redeem HoloFuel.
Reserves are accounts that can sell HoloFuel against a credit limit as long as certain conditions are met: the total amount of HoloFuel must be below a system-wide cap based on platform capacity metrics. Each reserve works with a paired currency and there must be demand for that currency from hosts. The price of HoloFuel will adjust at each Reserve based on the aggregated currency redemption preferences of hosts.
Anyone can buy HoloFuel at a reserve account, but only HoloFuel earned through hosting can be redeemed there. This creates a circular flow where new HoloFuel bought from a reserve enters the general circulating supply until it is spent on hosting. At that point a host could either sell it on an exchange or redeem it, depending on pricing. If they sell on an exchange they still keep their receipt of hosting so they can redeem HoloFuel at a reserve later.
Reserves will be created by Holo for practical reasons. Reserves need liquidity to function well, so we will ensure that critical mass is achieved before each additional reserve is created. There will be a regulatory complexity involved that must be managed, and reserves obviously have systemic impact by design, so we need to be careful with the roll out.
The HoloFuel price at a reserve is a critical management variable since it acts as a stabilisation price. Reserves do not have agency and act on behalf of hosts. Therefore the price on the reserve accounts will reflect the aggregate pricing and currency preferences of hosts. Specifically, hosts are able to set their personal floor price for HoloFuel in their management software. So for example a host could set it so that HoloFuel will not be redeemed below $0.1. Reserves will collect data on all hosts individual floor prices and then set a reserve price, for example top quartile pricing. This price would of course refresh to take account of new data. There may be additional feedback loops. For example if HoloFuel is consistently sold out on reserves the price could automatically increase.
In contrast, HoloFuel is redeemed based on price layers. This can be understood as the most recent price at the specific reserve account at which HoloFuel was last purchased. This ensures hosts receive a price that is close to market/spot. It also means that all fuel from a reserve is fully backed and that we are not artificially propping up prices.
Finally, reserves do not replace exchanges, since only fuel earned from hosting can be redeemed.
Here’s a simple example of how the price layers work, using numbers that are easy to deal with.
We start with an empty reserve account that currently sells HoloFuel at 75 cents each.
Now someone decides to buy 100 HoloFuel for $75
There is now $75 in the account and the redemption price is 75 cents.
Let’s say that the market changes and the price of HoloFuel on the reserve goes up to $1, and now someone else buys 200 HoloFuel at that price.
The redemption price is now $1
If we assume no further HoloFuel is sold by this reserve, hosts collectively can redeem up to 300 HoloFuel. The first 200 units will be redeemed at a price of $1. At that point the price will fall back down to $0.75 and there will be 100 HoloFuel available. If that 100 HoloFuel is then redeemed, the reserve account is empty and no further fuel can be redeemed at this account.
The reason why we implemented these reserves is to add structural resilience and ensure that HoloFuel is an effective tool for facilitating hosting. In particular, it addresses several major needs.
As shown in the previous example, hosts could choose not to draw against their credit limits. In the absence of reserve accounts this creates a credit freeze effect that prevents the supply from being dynamic. This increases volatility and may also make it difficult for users to get hold of HoloFuel.
It also helps us ensure that HoloFuel is driven by hosting dynamics. Reserves allow anyone to buy HoloFuel for any need, including non-hosting needs. This allows the supply of HoloFuel to adjust to non-hosting use. However, since reserve pricing is always linked to hosts, any supply change is still based on those host dynamics.
Finally, it reduces volatility of HoloFuel prices, which makes it easier for our ecosystem to make long term plans.
The result of these design choices is that the supply of HoloFuel is affected by the maximum potential supply or credit extended, and the actual usage of that credit.
The max supply of HoloFuel is controlled algorithmically by the network’s code. This is analogous to how Bitcoin or Ethereum’s supply is based on a block reward algorithm but clearly more complex since HoloFuel itself is a more complex, dynamic system. The management algorithm will set a “credit factor” that extends credit and allows hosts to spend to a negative balance, based on their track record. It will also use network statistics to set the HoloFuel price on reserves and the amount of HoloFuel available. We will cover how prices are set later, but as I said in the last slide it is a reflection of hosts preferences.
Within this maximum, the actual amount of circulating HoloFuel at any point of time is driven by 3 things:
First, the Holo organisation has a credit limit that can be drawn down to fund network development and is backed by the network transaction fees that we collect. Secondly, hosts themselves can spend on credit, which they may do to finance new hosting, if they think that prices are too high, or many other reasons.
Finally, reserve accounts can sell HoloFuel on credit, in exchange for funds (e.g. HOT, USD, BTC) which will later be paid to hosts who have provided hosting and want to redeem their HoloFuel.
HoloFuel use that does not result in changes to borrowing, including trading on exchanges, does not change the supply.
In this next section we will talk about some of the factors that influence fuel prices. I want to reiterate that nothing here constitutes price predictions, guidance, or investment advice. All prices and price movements discussed are chosen purely to help communicate HoloFuel economic concepts effectively.
In addition, we also discuss an abstract “unit of hosting”. In reality there will be no such thing. Rather, there will be prices for different hosting-related services such as prices for CPU-hours, network bandwidth, and storage, and that there will not be a single price, but different prices for different tranches based on hardware specs, geography, certification, reliability, and so on.
When thinking about HoloFuel and prices, there are a few different things to consider:
First is the actual hosting price offered to publishers or applications on the network. This price is denominated in HoloFuel but hosts will want to know that this price covers the costs of their hardware, electricity and bandwidth/connectivity typically in the local national currency.
The price of hosting is driven by typical supply-demand dynamics. What will publishers pay and what will hosts accept based on availability and demand.
But this model is also dependent on the HoloFuel price because after a hosts charges in HoloFuel, they will be able to redeem that fuel for the national currency at a Reserve or eventually at a 3rd party exchange.
The HoloFuel price is the price to buy/sell a unit of HoloFuel on either exchanges or reserves.
The purchasing power of HoloFuel is primarily impacted by internal competition and supply dynamics. For example, price competition on the network is in terms of hosting price in HoloFuel. Therefore if there is intense competition between hosts this price will fall.
Let’s delve deeper into the hosting price dynamics. Since the price of hosting is driven by supply and demand, it is ultimately driven by adoption and innovation. Let’s look at a few examples.
The chart shown here shows prices on the Y axis and quantity of hosting on the X axis. We have a demand curve D1, where the quantity of hosting demanded increases as price falls, and a supply curve S, where quantity of hosting supplied increases as price rises. The equilibrium situation happens where the two points cross, and Q1 amount of hosting is supplied and used, with a price of P1.
If holochain adoption grows then more hosting will be demanded at the same price, shifting the curve from D1 to D2. A similar dynamic would happen if Holochain-based projects are more profitable, since they can then pay a higher price for the same amount of hosting. When the curve shifts from D1 to D2 the new equilibrium price is P2. As prices increase to P2, new hosting capacity will join the network since it will now be profitable for them.
Conversely, if network efficiency improves then hosting hardware can provide more hosting. For example a single HoloPort might be able to host twice as many apps. If this happens then the supply curve S would shift to the right, the new equilibrium price could be lower, and this would attract more demand.
However, this increase or decrease in hosting price does not automatically translate into fuel price movements. This depends on network-specific dynamics.
Ok, so at the end of the last slide we looked at how an efficiency increase would decrease the price of hosting, all else being equal, since there is more competition. However, there’s another dynamic going on which is the internal competition between hosts on the network.
Imagine if efficiency increases doubled the effective capacity, and as a result let’s say the new equilibrium price of hosting is 25% lower and there was 50% more demand. Internally hosts would have 100% more capacity chasing 50% more demand, so competition would be really aggressive. In this scenario hosts would keep dropping their internal price, say from 10 fuel to 9 fuel, to 8 fuel, and so on, in order to get more utilisation. This is because they individually don’t have significant control over the price they get for Holo fuel.
What’s the floor for the internal price? The floor is the breakeven price. So in this case where efficiency means that hosts can provide twice as much hosting for the same hardware the breakeven price would be half the previous one, as long as HoloFuel prices stay static.
So if total hosting price was 25% lower but the internal price can fall by half, then HoloFuel price would go up by 50%. Of course, this creates a second order effect where because HoloFuel prices are now higher, hosts can push the internal price even lower, which again pushes HoloFuel prices up, and so on, until it reaches a new equilibrium.
Now these numbers are all illustrative, but the takeaway is that a key driver is how much the internal price changes vs. the price of hosting.
The other dynamic is the price signal from the reserve price. Since anyone can buy HoloFuel at the reserve price, and hosts can redeem at that price, the reserve price acts as a stabiliser that prevents the Holo fuel price from moving too fast. So going back to our example, the internal price may initially fall by 25%, but aggressive hosts may push it down further, while setting minimum HoloFuel prices (and taking the risk of having to hold). This will nudge HoloFuel prices up, and the cycle continues.
The other factor involved is the supply dynamics of HoloFuel. We have mechanisms that create a dynamic supply in response and counter to economic signals. Generally speaking, we want supply of fuel to expand when prices are going up and to contract when prices are going down, so that price changes are in response to underlying needs rather than artificial scarcity or overabundance.
Generally speaking, we expect market prices to move faster than reserve prices, so when prices increase, they will be higher at exchanges vs reserves, and vice versa.
As a result, there are several passive mechanisms in place that automatically incentivise counter-cyclical supply.
Another factor is velocity, which we don’t consider in detail in this document. Generally, we would expect velocity may amplify price movements but eventually come to an equilibrium. As prices increase it might create deflationary pressure and incentivises users to hold HoloFuel longer, effectively taking it out of general circulation
We think HoloFuel represents a sea-change in digital asset design by enabling a distributed, dynamic system that tends towards stability while being driven by its own internal economic needs. This is achieved through several key innovations that we think can be leveraged by others.
First, Holochain’s high transaction throughput and agent-centric approach allows more responsive, accurate, and targeted supply management from a technical perspective. For example being able to understand each node’s earnings history and redemption preferences.
Second, the reserve design enables the system to “listen’ to hosting needs specifically, while also being responsive to network statistics and over time, various components will be tuned to take into account and be responsive to other signals and activity.
Finally, since there is no PoW, PoS, or interest payments on HoloFuel, the HoloFuel system does not favour centralisation.
We’ve now reached the end of this presentation. Thank you for listening (or reading), and I hope it’s been helpful.
If you have more questions visit:
