Final summary of IPFS official Q&A!

Q: What is the current interopnet status of the two interoperable filecoin implementations, GO-filecoin and Lotus?

A: Go-filecoin and Lotus are currently interoperating on Testnet Phase 2. We find issues and patch them, but they are currently fully interoperable.

Q: When were the specific economic model parameters determined?

A: Detailed parameters will be released only after all core features are completed and thoroughly tested on testnet. The current parameters in testnet are temporary placeholders.

Q: Are there any incentives for testnet miners? Currently the required/recommended hardware (2TB, 8-core CPU, 128GiB ram) is cost prohibitive.

A: Yes, there will be a testnet miner competition mechanism, and miners will have the opportunity to earn a large amount of Filecoin. In order to participate, miners should be ready to store and retrieve real files for clients and perform industry lifecycle upgrades.

Q: Is customer data stored directly in the nodes or distributed after being sliced?

A: Client data is sealed directly into sectors stored by mining nodes. Clients may decide to split their data into multiple pieces and/or store multiple copies in different providers. We expect that Filecoin client software will emerge to automate redundant long-term storage.

Q: How long does it take to reseal?

A: Just seal. You can check https://filecoin-benchmarks.on.fleek.co/ (requires the latest benchmark) for a rough estimate of sealing times for different hardware combinations.

Q: If I have stored files on the Filecoin network and no longer wish to store them, how can I delete them?

A: There is currently no specific mechanism to request or require storage providers to delete data.

Q: If a client has 1GB of data to store, are the 32GB sectors generated independently to store it or are they pieced together with other users' data?

A: Both are possible, it is up to the storage miner to decide. There is no clear mechanism for the client to request dedicated sectors.

Q: When can we join the testnet? When will testnet/2 start?

A: You can join Testnet Phase 2 today! You can join at any time starting last Thursday.

Q: If user data is pieced together, metadata must be generated to mark the sectors and locations where the user data is located. So where does this metadata exist?

A: The sector storage module used by Lotus and Go-filecoin stores metadata about sectors in a key-value data store separate from the sectors themselves.

Q: Most of the hard disk orders purchased by Chinese miners were returned. Is this an official action? Will the hard disks continue to be sold in the future?

A: Please use the #fil-discover-support channel on Slack to ask/help with filecoin discover orders.

Q: How do you define valid data? If a third party is used to audit my data, it may be manipulated by humans. How do you solve the problem of cheating?

A: This is a question related to verified clients - we don't have a good explanation yet. First, please stay tuned for a talk I will be giving in the next few days, which will explain these mechanisms in more detail.

To answer this question simply, verification does not mean verifying the data itself, but verifying the client itself and handing over a certain amount of data to a trustworthy client. This will be a pragmatic solution for situations where it is difficult to distinguish between malicious fake transactions and real customer transactions, and will incentivize the correct operation of the network.

First, please note that data from unverified customers is still good and profitable, and the network can accept it. But it will most likely not be as profitable as mining data from verified customers... (see part 2) Please check out my talk once I give it, it will help clarify a lot of questions -Juan

Part 2:

Cheating is addressed through an incentive network:

(a) There must be many validators

(b) The verifier trusts the real client, based on the client’s application (not unlimited)

(c) Validators themselves are checked to do accurate work, and if they fail to do so they lose the ability to validate.

Q: When will we switch to NSE?

A: NSE is a work in progress and we are still evaluating it. We hope to have the evaluation done soon (before testnet2 incentives). We don't want anyone to waste any money, so please be careful and don't buy a lot of prototyping hardware by any means. --Juan

Q: Will ASIC chips be officially supported? If so, when? Will the current demand for GPUs disappear during the sealing process?

A: No one that we know of is developing ASICs. We specifically chose the PoRep algorithm to make it suitable for commodity hardware, not ASICs.

The only ASIC we are considering developing in the future is a VDF ASIC, which is related to Ethereum 2.

Note that PoRep and consensus algorithms may change and evolve after Mainnet, so the ASICs may no longer work and be wasted - we are targeting commodity hardware - Juan

Q: After the current SDR algorithm is changed to the NSE algorithm, will AMD server sealing still have an advantage? Will Intel servers work?

A: It is very difficult to benchmark an algorithm until it is fully developed and final parameters are determined. We will make an announcement in Slack once NSE is ready for benchmarking. That said, the current construction does involve SHA hashing at some stages, so CPUs with SHA extensions will benefit during those stages.

Q: In the current testnet, we can see that there is still some code work to be done. When will the testnet be reset? Will the mainnet be launched as scheduled?

A: Yes, there is still some work to be done before the protocol is frozen for mainnet. We are moving on to performance and UX issues, as well as the last sprint of protocol work.

You can check out our latest roadmap update for more information on release estimates. Note: These are estimates and are not confirmed yet. When we have a final launch date we will announce it on our blog:

https://filecoin.io/blog/roadmap-update-april-2020/

Q: How many full-time R&D staff are there in the lotus and go-filecoin projects?

A: The Filecoin project has a core team of about 50 engineers and researchers working on various components of the implementation, not including those working on supporting protocols such as libp2p and IPFS. In addition to the core team, there are dozens of people in external teams working on various parts of the protocol - from building developer tooling protocols, security audits, performance improvements, etc. Finally, there are hundreds of external contributors in the open source community who have made meaningful contributions to the project.

Q: In the second phase, the testnet often occupies memory. Is this related to the algorithm?

A: I'm not sure if you mean the memory jump for basic node operations, or the memory jump for mining (sealing). I don't believe there hasn't been a noticeable increase since phase 1.

Q: What kind of sectors does the mainnet support and when will the hardware standards be released?

A: The sector size and hardware standards will be determined soon after the final decision of NSE and SDR as it depends on that.

Q: Will Filecoin design data redundancy and encryption in the future?

A: Data encryption and redundancy are left to the client's discretion, and there are no current plans to build them into the core protocol. We expect standards and conventions to emerge in the storage market and collaborate with customer software designers and implementers.

Q: Is it confirmed that the main network will be launched in July or August, or will it be postponed again to ensure the security and stability of the network?

A: The security of the Filecoin protocol, implementation, and network is our top priority. We will not launch Filecoin unless we are confident it is secure.

You can check out our latest roadmap update for more information on release estimates. Note: These are estimates, the launch is not scheduled yet. When we have a final launch date we will announce it on our blog.

https://filecoin.io/blog/roadmap-update-april-2020/

Q: If a miner's hard drive previously stored data generated by himself, should the original data be overwritten after receiving a transaction from a verified user?

A: The storage market distinguishes between "committed capacity" sectors (with no client transactions) and sectors storing client data. Committed committed capacity sectors will be upgradeable to replace zero data with client data (this mechanism is not yet fully implemented). However, sectors with any transactions (regardless of whether the client is the miner itself) cannot be replaced by "verified transactions" - only empty committed capacity sectors can be upgraded.

Q: When will Filecoin be able to store hot data?

A: Filecoin can already store hot data through Filecoin payment channels and development tools such as Powergate. This hot data can be served quickly, and we hope this will be helpful for Web3 applications.

- If NSE is eventually realized, it will make the storage market data more hot, although it is still not hot enough at present.

-We expect that future PoRep construction will be faster and faster, and filecoin researchers have already made suggestions. It just may not catch up with the mainnet launch.

Q: How will the 1 million FIL bonus for Test Phase 2 be allocated?

A: We will be announcing the final structure of the testnet incentive mechanism soon! If you are interested in participating, we strongly recommend that you set up a mining operation so that you can compete with real users for storage and retrieval transactions.

Q: What is the biggest technical difficulty at present?

A: The Filecoin protocol is relatively complex, with many interacting parts. The biggest difficulty is ensuring that multiple parts work together to achieve a secure and useful protocol.

Q: How do I handle data corruption on one or more nodes?

A: Miners should keep backups if they are concerned. If you lose data from an active transaction, or miss a window POST challenge, you will be penalized.

There's no getting around this, since proving you've stored the data is the whole point of FILECOIN, and there's no way to tell the difference between "my data is corrupted" and "I never stored the data like I said I did".

Q: If valid user data stored in the hard disk has expired (the storage period in the storage order has expired), can miners delete the data?

A: Yes, a sector may be terminated when all transactions in it have expired and have reached the end of their declared lifetime, and the storage used to seal a new sector.

Q: How does the Filecoin network handle content removal requests? For example (child pornography) If I am a storage provider, what are my responsibilities here?

A: Filecoin is a protocol that helps coordinate file storage between a decentralized network of storage providers and clients. The network cannot delete files on participants' computers. However, we are developing a powerful set of tools to help users control the type of data they are willing to store and serve from their own machines.

Q: Will you design and develop official mining software?

A: The filecoin project has developed the https://github.com/filecoin-project/sector-storage module, which is used by both Lotus and Go filecoin. Miner operators are free to customize or re-implement it as needed.

Q: For the retrieval market, like at launch. What prevents miners from charging exorbitant retrieval fees and effectively holding the data hostage?

A: Retrieval should be economically sound, and both miners and clients can replicate different miners to minimize this risk. The free market works!

Q: Where can I see the talks that Jbenet keeps mentioning?

A: The discussion about verified clients will be posted in the future - Jbenet is referring to a discussion he will give and link to later.

Q: There will be Fil rewards in the later stage of the second phase, but it is not recommended to buy hardware in the early stage. How should we choose?

A: When we announce the incentive plan, it will be accompanied by a description of the intended algorithm we will use for mainnet launch. We will ensure miners have a few weeks between announcing any incentives and starting to compete (as long as we can provide it).

Q: Data verification design can motivate miners to store valid data. For example, miners with orders can get a coefficient reward of 1.1, which is a very meaningful weak dominant strategy. However, using the hard disk sales method obviously cannot guide miners from the perspective of technical improvement, and it is easy to lead to short-term behavior. Have you considered how to deal with the possible negative effects of the 10x coefficient, such as: contrary to decentralization, too high a coefficient leads to the concentration of block rewards, so that it will eventually become a playground for capital rather than a paradise for technological progress?

A: The Discover project only accounts for a very small part of the total, and there will be larger and more decentralized validators and many verified users in the future.

Q: How do I understand the exact meaning of the committed capacity sector? How can I avoid committing a wrong committed capacity for return?

A: I’m glad you asked! When there is no storage demand on the network, miners can choose to prove committed capacity sectors, which can then be upgraded to sectors with actual transactions to earn transaction rewards (no lock-up). In the absence of storage demand, it should be more reasonable for miners to seal committed capacity sectors than self-transactions.

Q: Will this test be restarted next?

A: We will restart when we are ready to start more functional testing. We are not sure when that will happen yet.

Q: Has PL considered the problem of malicious short selling in the financial market? If someone in the market shorts Filecoin, how will PL respond? How should miners respond?

A: Please refer to the Code of Conduct - Please do not ask questions about the market, token sales, etc.

Q: Has the epidemic in the United States affected our progress in time? When will you come to Shanghai, China?

A: The pandemic has slowed us down a little bit, but not a lot. We are prepared for this and are working hard to cope with it. We hope to go to China when it is safe to travel, but unfortunately it is unlikely in the next few months. I hope to be able to go in 2021 or earlier.

Q: Network baseline setting is critical. What kind of benchmarks are you considering? For example: How long do you expect it will take for the network to reach a certain goal?

A: We have done a lot of work and will finalize the network parameters later. We are excited to see the baseline running in the testnet and block rewards have increased 7x since the testnet launch! We hope you are encouraged as well.

Q: Regarding the data hard disk, how much data is there currently? How much did the first batch sell for? How much more sales are planned?

A: Filecoin Discover has been very well received - the orders we have received so far represent data that would take years to publish directly over the internet. We are excited that it will support the preservation of so much important data.

It is important to emphasize that Discover is just one of many verification clients. In the coming days we will be announcing some small changes to Discover to reinforce the point that participation is not mandatory and should not be the main focus of a mining operation.

Q: When will the 5PB large miner test begin?

A: We will announce our next steps soon.

Q: How does filecoin storage utilize the ipfs system?

A: Filecoin storage does not "leverage" IPFS, but it is easy to store IPFS data on Filecoin, and it is also easy to put Filecoin data on IPFS.