First let's explain regular proof of workload. It's a way for the blockchain to implement *** knowledge. It's the way Bitcoin uses it, so proof of workload is commonly known as "mining". Bitcoin is a decentralized peer-to-peer transaction system, and in order to maintain a ***identical and identical ledger on different nodes to record all transactions, and to ensure that transactions are not duplicated, and that no money is spent more than once, a rule is needed to maintain the consistency of the ledger. The fact that everyone follows this rule together is ****sense. A common approach to blockchain is to divide this ledger into many pages, each page being a block. Each block is bookmarked by a node and then distributed to other nodes to replicate, so that the ledger is the same on all nodes. But which node keeps track of each block requires a rule that everyone can follow. The method used by Bitcoin is to have all the nodes do a simple math problem, which is very simple but computationally intensive, and it usually takes about 10 minutes to come up with the answer. Getting the answer is time-consuming, but it's easy to verify that the answer is correct. Then all the nodes do the problem at the same time, the first node to do it, you get the next block of bookkeeping rights. Because each block has only one node that made the question the earliest, the bookkeeping right of each block is unique and it is also easy to be verified by other nodes. Once a node verifies that another node has been given the block bookkeeping rights, it must copy the block and add it to the local blockchain, as well as start the competition for the next block bookkeeping rights. In this way, Bitcoin ensures that the blockchain is consistent across all nodes.
The process by which nodes compete for block credits through a large number of computations is called proof-of-work. A proof-of-workload system (or protocol, or function), therefore, is an economic countermeasure to denial-of-service attacks and other service abuses. It requires the initiator to perform a certain amount of computation, which means that it consumes a certain amount of computer time. The concept was first introduced by Cynthia Dwork and Moni Naor in 1993 in an academic paper. The term proof of workload (POW), on the other hand, was only really introduced in 1999 in a paper by Markus Jakobsson and Ari Juels.
There are many other ways to implement blockchain *** knowledge, such as POS, DPOS, POA, PBFT, and so on, but Proof of Workload is the only time-proven (11 years) blockchain *** knowledge mechanism that works on public chains.
What is one problem with proof-of-workload? Or let's use Bitcoin as an example. Bitcoin nodes do that math called hashing in order to get out of the block. It's a very large computation, and every miner involved in Bitcoin mining has to do it all the time, using a lot of power. This calculation is not like other calculations such as big data processing, which can generate some value, but its only purpose is to compete for a node to be the next block maker. Currently Bitcoin consumes about 2.55 billion watts of electricity per year, which is 0.5% of the world's electricity and is the amount of electricity consumed in Ireland in a year. Opponents of POW have accused mining of wasting power resources on a nebulous digital currency, and have called it liberal "slop".
But what people who think POW is a waste of electricity don't realize is that it's the energy and arithmetic power that makes Bitcoin secure and impenetrable.
A $100 piece of cash isn't just you or I thinking he's worth 100, it's an entire social group thinking he's worth 100, and value is what comes from ****ing knowledge. Bitcoin is community behavior, people from different countries gathered into communities, using the Internet to establish order, its meaning is also from the group **** knowledge, as long as everyone believes that bitcoin has value, only **** knowledge exists, then he has value, exactly the same as fiat currency. So produce value identity does not necessarily need to be driven by the state, bitcoin reformed a carrier and medium to transfer trust, thousands of years, human society through how much bloodshed wars to establish the regime and **** knowledge, and now bloodless, just cost some electricity to realize, is not more advanced.
In summary, to design a decentralized and secure digital currency, energy and arithmetic power are necessary costs. Proof-of-work is the only way to build a secure authentication system in a decentralized form. That's why people who think POW is a waste of electricity don't realize that it's energy and arithmetic that create Bitcoin's secure and impenetrable system. Now that Bitcoin's network-wide power has reached a very scary point, anyone who wants to launch a 51% power attack is already an impossible thing, and the POW algorithm makes the Bitcoin system unbreakable.
Consumption for value creation is not called waste.
But can so much power be used to create more value? In FileCoin's words, proof-of-work, is there any other use for it?
FileCoin is a star project in the distributed storage industry. His development team, Protocol Lab, is the same team that developed the IPFS protocol, so much so that many people can't tell the difference between FileCoin and IPFS. It can be said that it was FileCoin's 1CO in 2017 that pushed the industry to its peak and led to a series of projects of the same type. This article is not intended to praise or belittle this project, but only to combine my own experience in this industry to express some of my own views, and try to be objective and fair. I hope that people engaged in this industry have some inspiration.
FileCoin's whitepaper proposes to implement a useful proof of workload, which actually recognizes that to build a secure and unbreakable blockchain, you have to consume workload. However, they don't want the computations made for that workload to be completely wasted, so they want to put that workload to use. So, the way they came up with was to add storage space utilization to the proof of workload. In this way, all the nodes will have to provide storage space to store files in order to form the **** knowledge. This storage space can then hold user data that is just useful.
So let's look at how FileCoin implements this useful proof-of-work *** knowledge.
The *** knowledge mechanism used by Filecoin is not a simple proof of workload, but a mechanism called Expected Consensus (EC). The goal of this mechanism is the same as that of other mainstream ****recognition mechanisms, allowing miners to compete for a highly unique block right and get rewarded. The miner who gets the block right is called the Leader, and in each round of block contention, there is a unique leader to keep score in order to ensure the reliability of the ledger.
In other words, the core of *** knowledge is to choose who will be the Leader, and there are generally two ways to choose the Leader, interactive or non-interactive. There are two ways to choose a leader, interactive or non-interactive. Interactive is where miners vote for each other. For example, PBFT is interactive, several participants send messages to each other, and the one who gets a majority (more than 2/3) of the votes is the leader.The expected *** knowledge adopts a non-interactive way to elect the leader, the participating parties do not send messages to each other at all, and each node conducts calculations independently and privately. At the end, a node says, "I won the election," and then provides a proof that the others can easily verify that he did win the election. This verification method is a zero-knowledge proof.
The Expected*** Knowledge mechanism presets an expected value of outgoing blocks for the blockchain network. For example, every 1 epoch (epoch) generates 1 block, but there is also a possibility that an epoch may have an empty block or multiple blocks. So in Filecoin, each epoch is not a block, but a set of blocks called a TipSet, and this TipSet may contain multiple blocks. So in effect Filecoin is a TipSet chain. Expected*** knowledge can not guarantee that only one leader will be elected in each round, so there is a possibility that there are more than one leader in a round, so the chain structure becomes a mesh structure of DAG. So FileCoin also assigns weights to blocks to achieve effective convergence.
The EC *** knowledge used by FileCoin has an advantage. One of the major problems with the traditional POS *** knowledge mechanism is the inability to control forks. That is, due to the low cost of mining, participants can mine multiple chains simultaneously for profit. The expected *** knowledge has designed for this, that is, through the weighting and collateral mechanism to prompt miners to choose a best chain, and penalize miners who mine multiple chains at the same time, which can promote convergence very quickly. This suggests that POW and POS ***used together would be a good way to go.
Each miner's likelihood of getting a block out is positively correlated with its current effective storage as a percentage of the total network-wide storage. This expectation *** knowledge mechanism is actually more like a POS proof of stake, except that it replaces the POS stake (Staking) with the effective storage share. But where does the miner's effective storage come from? It is through the storage of user data. How to prove that the miner stored the user's data, FileCoin has created a new proof mechanism called POST Proof of Temporal Replication. This POST is FileCoin's workload. By replacing the power-consuming arithmetic power with storage space to store useful data, the meaningless arms race has turned into a competition in the storage service market. This is indeed an improvement of FileCoin. Only, in order to successfully block out, after a miner is selected as a blocking node through the expected **** knowledge, he has to do a POST proof in a block of time (now 45 seconds) and successfully submit it, in order to block out. Otherwise the opportunity is lost. So, in order to make sure that miners can get their blocks out within the specified time, the official decision was made to use a GPU, which doesn't work all the time like a proof-of-work, but there are still energy-consuming calculations throughout the implementation of the ****-knowledge that go against the grain of the useful proof-of-work idea.
Also, when it comes to expected *** knowledge, the notion of epochs is important when we talk about how every epoch out of a block is not a block, but a set of blocks. How do you control epochs? Each miner in the election before, need to Mr. into a Ticket, this Ticket is actually a random number, he needs to go through a VDF and VRF process, the VDF full name Verifiable Delay Function, verifiable delay function. His calculation process is serial, it takes a certain amount of time, and this time can not be reduced through multi-core parallelism. This guarantees that the time each miner has to spend generating Tickets will not be accelerated by optimizing the hardware. This sounds like a perfect function, but the VDF doesn't even exist yet! The FileCoin test network now directly uses a wait function sleep, which is a UDF, Unverifiable Delay Function, the closest VDF solution that consumes a lot of computing resources. To put it bluntly, it still consumes power and is still not environmentally friendly.
So, a useful proof of workload is still just a good wish, the ideal is very rich, but the reality is very bony. FIL, which has been hailed as the next Bitcoin, will have to continue to work hard to realize this disruptive **** knowledge.
To summarize the flow of FileCoin storage miners to obtain incentives: users store data, pay FIL fees -> miners store data -> generate replication proof -> complete the proof of time and space -> after the EC **** knowledge, elected out of the block Leader -> get the right to pack -> miners to obtain the FIL reward
.On this flowchart, you can see that there are two places where miners can get rewards. One is when storing a user's file to get the user's FIL reward. One is to get the FIL after getting the right to pack the block, and one of the prerequisites for getting the right to pack the block is to store enough user data. So, in the case of insufficient storage demand, miners will charge very low fees from users. If there are not enough users, the miners will even subsidize their own money to pay FIL to store the data, just to be able to store enough data to be selected as the Leader in the EC ****knowledge to get the packing reward. The effect is that FileCoin is very user-friendly and the storage cost is very low. Therefore, it will attract a lot of applications to do development on this platform. But the disadvantage is also very obvious, if the storage capacity is not large enough, miners can not compete with other people for the block right, so they do not get rewarded. In the end, the whole platform will develop in the direction of big miners and big mining pools, which is contrary to FileCoin's original intention of utilizing all idle servers to realize distributed storage. Or rather, we must wait until the industry has a certain scale and technology is more mature, before there is a chance for small miners to mine.
Let's start by briefly talking about the respective pros and cons of centralized and decentralized storage. Centralized storage devices unified management, good reliability, high performance, decentralized storage data naturally dispersed, easy to circulate, good disaster recovery, but low reliability. From an economic point of view, centralized storage is a heavy asset investment with high costs. Decentralized storage through the blockchain incentive layer, users join on their own, light assets, can reduce the total cost of storage. In the future, the storage and processing of application data will still be dominated by centralized storage, while decentralized storage can be used mainly for popular data traffic distribution because it is a distributed network. At the same time, because there is no centralized ownership, it can be the first choice for decentralized applications.
There is an argument in the market that decentralized networks are suitable for cold data backup, which is actually not an advantage of decentralized storage, really because putting hot data on decentralized networks is too unreliable, and the processing performance cannot keep up. Therefore, if decentralized storage can achieve a certain scale effect, greatly reducing the cost of storage, cold data backup as a core business, and targeting today's because the cost is too high not to be stored by the enterprise cold data, it will be a good direction of development.
In this way, technically speaking, decentralized storage does not necessarily have an advantage over centralized storage. If a new model can be implemented that combines the economic incentives of decentralized and centralized storage, it can absorb the strengths of both. Truly useful workloads are realized. there may be more of a market for the big mining model of data centers that FileCoin may enable in the future.
Now, 11 years later, Bitcoin hasn't realized its original purpose of being a peer-to-peer electronic payment currency, but that hasn't stopped humans from going out and buying it and owning it. Similarly, I believe that FileCoin has the support of a large enough community of miners and developers that even if it doesn't lead to a full-scale distributed storage application in the foreseeable future (which was probably never FileCoin's goal), I still believe that there are a lot of people who will buy it and hold it because of its ****ty knowledge. Going up to the philosophical level, humans are buying into the truth.
So what is useful in practice, or are we using storage for *** knowledge or *** knowledge for storage?FileCoin is the former.FileCoin wants to decentralize *** knowledge based on the amount of storage work achieved, which is theoretically perfect, and in the pursuit of perfection, humanity pays the price. That's why we've waited so long on this project. But once realized, it may bring great value to humanity and an infinite call to the market.
Except that decentralization is not a panacea for everything. One of the biggest advantages of centralization is that it is very efficient. Weakly centralized like dPOS or alliance chain *** knowledge to take into account the advantages of both, can more quickly to the market application, early start distributed storage industry, promote distributed storage application landing. Therefore, we are pursuing both the use of storage for **** knowledge and the use of **** knowledge for storage, and making our choices based on actual needs. In this process, I believe that the blockchain will also be further developed, gradually optimized, and become more and more useful.