A letter about the design defects of the Filecoin protocol

Greeting! @molly @protocol lab
I am a public-chain economy mode researcher, developer, and filecoin fans. My team and I hope that filecoin will achieve further human/commercial history.

I studied the filecoin protocol carefully and participated in Space Race deeply, I found many defects in the design of the public-chain protocol. These defects will lead to problems such as unfairness, centralization, selfish mining, permission, poor user experience, etc. of the public chain.

Next, I will explain the design defects one by one, and hope to have in-depth discussion with you on the solution.

EC Protocol

The probability of a miner being elected as a block leader is positively correlated with the historically accumulated storage power. This leads to miners who enter the network first, and the historically accumulated storage computing power will form a “moat” for miners who enter the network later. This is a bit like “the coin age attack”——popular in old POS protocols that are eliminated. For a permissionless public chain, this is obviously unfair.

In addition, considering miners generally “selfish mining”——miners only pack their own messages, a miner have to pack as much his owner messages as possible once he get the right to produce a block, so that he can improve his storage power. For small miners, the probability of obtaining block rights is too low, resulting in small miners have no chance to increase their stake. In all other popular POS chains, any small miner can easily increase his stake (for example, pledge more tokens or get more votes). This will inevitably lead to increasingly centralized filecoin.

When calculating the block weight, the main chain recognizes blocks produced by miners with high computing power. This is very unfair. Big miners are likely to fill up their own messages selfishly into the blocks they produced, which will make it easier for big miners to increase their storage power. This will further lead to centralization.

Gas fee

The intention of basefee is to adjust the busyness of the network, that is, more transactions → higher basefee → fewer transactions → lower basefee, but I am afraid it does not work.

Even when the network is very busy, the selfish miners who only pack their own messages will actively control the gaslimit of the messages in the block to keep the basefee at a low level in order to save the gas fee for their sending messages.

For miners who are eager to increase their power-growth rate, when it is their turn to generate blocks, they will ignore the rise in basefee and pack as much information as possible into the block.

Fundamentally speaking, miners are both producers and consumers, and their dual identities will cause market price adjustment mechanisms to fail.

rewards

The baseline increases exponentially at a multiple of 3. If miners want to obtain baseline-rewards, they must increase hardware costs exponentially. Small miners have no money to exponentially increase hardware costs, so their starage power will be dropped exponentially. In the end, block rewards are hardly available for small miner.

Compared with small miners, Big miners have block rewards to balance the cost, so big miners can provide lower storage prices. Small miners have no price advantage and are mostly likely no deal. This status will get better until all block reward tokens are issued.

To sum up, small miners have neither block-rewards nor storage deal income. They have to leave the network and make the network more centralized.

slash

Miners must continue to submit POST messages to prove that their sectors are intact. However, if the miners are selfish and never packed the competitor’s POST messages, this will trigger the system slash and cause the competitor’s adjusted power to decline. It has happened in the race. This is a very terrifying, extreme and bad situation.

I know the difficulties and challenges of public chain design, but we are with you. looking forward to your reply.

Hi, thank you for the note and shared vision. Let me shed some light on these issues. Some are valid concerns that are being addressed and others could be misunderstanding.

  1. Consensus is based on proven storage so the more storage a miner has proven, the higher the likelihood that they mine a block. Miners may prioritize their own messages first but that doesnt mean that there is no incentive for them to include other miner’s messages or these block producing miners can include their own messages for free (base fee is burned for all miners). It’s just that when a block producing miner produces a block, they can pay themselves a close-to-zero gas premium while it will be higher for other miners. Not including other miners’ messages also leave money on the table since some other miners will earn their gas premium. All miners are incentivized to help the network get closer to the baseline to unlock greater rewards.

  2. It is false that the current block producing miner can influence the BaseFee that they are paying in the current epoch with how they pack their own messages. It is a variable set by the protocol. Nonetheless, they can influence the BaseFee for the next epoch. Some observations are due to how the SpaceRace faucet worked and BaseFee has become more stabilized with recent improvement. At the end of the day, it’s a resource allocation problem. Message senders are bidding for space in the next blocks. Would love to hear your alternative proposal or further analysis.

  3. There are many points here and many are misunderstanding. The presence of the baseline makes rewards more proportional to storage added than just time or sealing hardware, as rewards scales to storage on the network. It is also not encouraged for large miners to exponentially increase their hardware cost if it doesnt make sense for them. There are other PoReps that could make sealing more efficient as the protocol improves and evolves. Baseline helps smaller miners and the situation will be even worse in the absence of a baseline. To compete, smaller miners still need to provide reliable services to storage clients. They can differentiate by geographic location and product offering to attract verified clients.

  4. Apology for the mining experience, mechanisms are being proposed and evaluated to recognize operational risks in mining. There are multiple constraints that the protocol is dealing with but hope there will be good news soon.