Author: Gerry Wang
In the core mechanism of #Arweave, there is a very important concept and component, which is the storage fund #Endowment. I have already described it in a previous article "Is Arweave's $AR deflationary?" This article will introduce it in more detail.
In Arweave's incentive model, data uploaders are required to pay a transaction deployment fee for permanent storage and provide $AR to the network's storage fund (Storage Endowment) for prepaid fees for perpetual storage. This fund acts as a payment faucet, and over time, miners will receive payments through this faucet when providing proof of replication of data sets. As storage costs continue to decline, the amount of payment required to maintain a piece of data from the fund will automatically decrease accordingly.
Storage Price
The fee that users pay in advance for storing 20 copies for 200 years is calculated based on current costs. The advantage of the Arweave protocol is that it provides a very scientific trustless mechanism to determine the price of obtaining storage space from miners. During the period of a single block B with a specific difficulty d_B, the estimated number of partitions in the network can be calculated using the formula in the previous article:
Multiplying this expression by the partition size, we can calculate the current total storage of the Arweave network at the time of block B:
The amount of $AR issued to miners as rewards and the difficulty of the block can be used to estimate the storage acquisition cost - the fee paid for 1 GB of service for 1 minute at the time of block B:
src="https://img.jinse.cn/7206029_image3.png">
Where:
P_m^(B) = Estimated cost of storing 1 GB of data for 1 minute at block B
r_B = Total rewards for block B
Formula Notes:The total amount of data stored on the entire network is calculated in TB, so it needs to be multiplied by 1024 to convert it to GB; the "2" in the denominator means 2 minutes per block, so it needs to be divided by 2 to convert it to the estimated cost per minute.
Using a single block period to estimate storage prices will have high instability due to the difference between the collected transaction deployment fees and the difficulty adjustment algorithm. Therefore, in practice, the network keeps track of the difficulty and rewards released on a large number of blocks. These records are used by the network to accurately calculate the storage acquisition cost from miners in the 6 weeks before the block:
Where:
P_m(B) = the average storage acquisition cost of 1 GB for 1 minute calculated over a 6-week period.
h_B = the height of block B.
n = the number of blocks in 6 weeks (30247*6 = 30,240).
Using these calculations, the network can accurately estimate the acquisition cost of 1 GB of storage within a block cycle (about 2 minutes):
Based on this formula, the protocol calculates the current price of 20 copies of any data D in 200 years as follows:
This is the price charged to users as an upfront contribution to the storage fund. Over time, miners will receive payments from the fund after proving that they store the network dataset, calculated as follows:
Where:
r_e(B) = extraction from the fund at block B
r_i(B) = inflation reward released in block B
r_f(B) = transaction deployment fee for transactions accepted in block B
P_B(B) = estimated cost of storing 1 GB of data for one block period at the time of block B
W = At the moment of block B, all data sets stored on Arweave
This formula means that when the total storage cost is greater than the block reward obtained, fees will be extracted from the fund to subsidize miners. But from the previous data, the result of this formula is negative, so not only will no fees be extracted from the storage fund, but part of the fees will be deposited into the fund.
Deflation and Fund Value
Whenever data is uploaded, the Arweave network will move the corresponding number of tokens in circulation to the fund (Endowment) used to pay for the data storage fees accumulated over time. The storage purchasing power of the fund is elastic and changes with the amount of data submitted, the cost of data storage, and the value of the token.
One of the main drivers of changes in the value of the fund is that the reduction in storage costs leads to a corresponding increase in storage purchasing power, resulting in a reduction in the number of tokens that need to be released from the fund in the future. We call the rate at which the total cost of actually storing a unit of data decreases over a period of time kryder+ . This rate includes changes in hardware prices, electricity costs, and operational costs associated with data storage.
Figure 1: The health of the fund is affected by the difference between the protocolized kryder+ rate (0.5%) and the actual kryder+ rate, as well as changes in token prices.
Users pay for 200 years of replicated storage at current prices, so a kryder+ rate of only 0.5% would be enough to sustain the fund indefinitely without token price changes. However, over the past 50 years, the actual rate of decline in storage costs has been an average of about 38.5% per year. Given the significant incentives and room for improvement, this trend is likely to continue. In addition, the difference between the actual Kyder+ rate and the protocolized 0.5% kryder+ rate can be adjusted to provide a wide safety margin for token price fluctuations, which will also lead to deflation of the token supply over time.
Decentralized Content Policy
The Arweave network uses a decentralized hierarchical content policy system without centralized control and audit. The basic principle of this system is voluntariness: each participant is free to choose the data they want to store and provide, and the protocol does not impose any mandatory requirements. This system allows each participant in the network to create and operate their own content policy without the consensus of others. The result of this approach is a diverse choice of content policies, which are specifically reflected in three different levels:
Miners: Because miners store data and make their content public, they will be subject to any national or regional laws and regulations. So miners in the network can run arbitrary computations on the data they store (including various forms of analysis of text, images, videos, etc.) to filter out content they deem illegal or inappropriate. This allows them to avoid storing illegal content that does not comply with local regulations.
Gateways: Users typically access content on Arweave through gateways. Gateways act as portals that allow users and developers to access data in the Arweave network without running their own nodes. Just as miners can choose their own content policies for stored data, gateways can also independently decide what content they index and serve. In addition, the interoperability of gateways in the network allows users to choose gateways that align with their personal beliefs and values.
Applications: The last layer of content moderation that may affect Arweave users is at the application level. Each application built on Arweave data may use additional filters on the content provided by its interface, depending on the developer's code implementation. These application-level content policies can be embedded in the source code of the application itself and stored immutably on Arweave — allowing users to permanently trust how the application will conduct content moderation.