FIP-0080: Motivation for a Scheduled Reduction in the Quality Multiplier of Fil+ Deals and Example Timelines #774
Replies: 57 comments 301 replies
-
Thank you for getting this well-thought discussion open, @dcasem! I will have more 🤔 to share later, as I mentioned to you during the review, one thing I think needs to be discussed is "Should the cutoff multiplier be 1x?". I personally think it could be one day, but not now or soon at least:
If we were moving towards this direction (scheduled phased out), I think having a schedule that checks the "data storage %" on the network is worth considering.
I'd admit this is not the simplest approach. Thus I am looking forward to seeing other proposals that can also address the concern mentioned above! |
Beta Was this translation helpful? Give feedback.
-
Firstly, support this proposal to end Fil+, but it has to be said that reducing DC sector’s computing power is a huge challenge,Fil+ = Pandora's box, closing it may cost 10 times. and hope everything goes smoothly. |
Beta Was this translation helpful? Give feedback.
-
Thank @jennijuju for mentioning my proposal it is indeed simpler to implement and completely permissionless. |
Beta Was this translation helpful? Give feedback.
-
Support this proposal. |
Beta Was this translation helpful? Give feedback.
-
Great proposal! Explaining the problems with Fil+ very well. |
Beta Was this translation helpful? Give feedback.
-
Cancelling FIL+ is easy, but what are the next steps? Change to another way to encourage SP to perform sealing sector? |
Beta Was this translation helpful? Give feedback.
-
FIL+ can be regarded as a network subsidy. In the case of bad market conditions, network subsidies are very necessary. If there is no DC, most sp will not continue to invest in FIL, because the income cannot cover electricity costs, cabinet costs, machine costs, and network costs at all. Rather than canceling 10X, I think that 10X can be opened up so that all SPs can easily get 10X, just like getting CC. |
Beta Was this translation helpful? Give feedback.
-
I strongly object to Option 2 of the Fil+ phase-out: Breach of Trust: The 10x quality-adjusted power (QAP) was presented as a given for 540 days. Many SPs, including myself, made strategic decisions and contractual obligations with investors based on this promise. Gradually decreasing the multiplier within this period would be a breach of trust, undermining the credibility of the Filecoin network and potentially discouraging SPs from making further contributions. Financial Risk: Significant financial commitments have been made by SPs based on the 10x QAP. The proposed reduction would effectively reduce the ROI for these investments, possibly leading to financial losses. SPs that have entered contractual agreements with third parties based on the initial assurances may face even more severe repercussions, potentially even legal action. |
Beta Was this translation helpful? Give feedback.
-
support The power is increased by 10 times, and the pledged currency also needs to be 10 times. Originally, 10 machines + 6500FIL can do 1P power, but now they can do 10P, but it costs 65000FIL. If it costs 25$ each, it will cost 12 million yuan. how can I afford it? Therefore, the currency price must be lowered to switch to FIL+, so the final result is that within one year, the currency price has gradually dropped by 10 times, and the power of FIL+ has increased by 10 times, and a new balance has been reached. The problem now exposed is: after everyone rolls FIL+ together, they find that the final income has not increased. As time goes by, the marginal benefit decreases. 10 times for everyone = no 10 times for everyone, which instead increases the acquisition of DC quota. A series of centralized "authorized" operations such as validator audits have been protested by the community, demanding that FIL+ be phased out within one year and return to a state where everyone is equal. If FIL+ is canceled tomorrow, there will be two situations:
中文版讨论在这: |
Beta Was this translation helpful? Give feedback.
-
贸然取消fil+可能将fil带入万劫不复,但是公证人制度是肯定不能继续存在了,或者说所有人都可以是公证人,当前的公证人制度非常不合理,还不如彻底放开fil+,所有的非CC扇区都可以是十倍算力,可以考虑加入检索相关的考核,例如扇区通过一次检索后才能获得十倍算力,检索失败的扇区将暂时失去算力等 Canceling fil+ rashly may bring fil to the end of the world, but the notary system must not continue to exist, or everyone can be a notary. The current notary system is very unreasonable. It is better to completely release fil+. CC sectors can have ten times the computing power, and you can consider adding search-related assessments. For example, a sector can only get ten times the computing power after passing one search, and sectors that fail to search will temporarily lose computing power, etc. |
Beta Was this translation helpful? Give feedback.
-
Oppose high-table meetings and establish community councils.If FIL+ is good, why are so many people in the community opposing it? Clearly, the project team does not want to make any positive response, and they are indeed trying to turn a blind eye to certain community feedback until the community completely loses trust in the project. |
Beta Was this translation helpful? Give feedback.
-
cancel fil+ is like the eth difficulty bomb, with the filecoin difficulty bomb. The filecoin community needs to find a solution in a deadline, let's say 1 year. |
Beta Was this translation helpful? Give feedback.
-
I am with a group seriously considering becoming a reasonably large scale SP. The ROI without Fil+ does not justify investment. While we all like the idea of Filecoin, the reliance on SP's to source the paying customers is a large dis-incentive to participate. This biz dev requires a different skill set that comes at a cost. Perhaps I am more commenting on the fundamental assumptions which this project is relying upon for it's growth, but, unless I were running a shrinking Datacenter business already paying a sales team, Filecoin without Fil+, or a functional consumer facing gateway, I fail to see how this proposal will increase network size or increase FIL value which would incentivize SP growth. |
Beta Was this translation helpful? Give feedback.
-
In fact, when all CC expire, they will all become DC,et go of the threshold of DC, so that everyone can get DC. I think it's a more neutral solution. This is softer than canceling FIL+ directly, and it also solves the problem of DC trading and abuse in the current market. |
Beta Was this translation helpful? Give feedback.
-
The white paper was written by humans not gods. Because there is no god among us, we could expect no one making absolutely right decision. It’s all a matter of consensus. AFAIK, focusing on the one and ony important thing, let go of the threshhold and complicated restrictions is a optimal way to get more consensus. I feel uncomfortable when people keep saying "useful" storage, it's a biased word, it's distorting the protocol. More people will advocate Filecoin if Filecoin is only a storage network. Less people will like Filecoin if Filecoin only applies to the data that a small group of people think is "useful". Afterall, no one cares what you think is useful, but people indeed care about storage. Focus on "STORAGE", not on "USEFUL" please. |
Beta Was this translation helpful? Give feedback.
-
Even with a mere 1x difference, dc-2, cc-1, it will motivate the majority of miners to strive towards dc, unless your goal is to lock up a larger proportion of fil through fil+, which appears to be an overdraft on the future. The fairness, efficiency, and trust issues of the notary mechanism are the most critical problems, and the main reasons for the community's opposition to dc. Finding a more fair and effective method is the key to solving the problem. Unfortunately, I can't think of what that method might be at the moment. |
Beta Was this translation helpful? Give feedback.
-
One single proof that Fil+ has "negative" storage value (and that's enough IMO): SPs are PAYING to store clients' data! Previously, for CC/unverified sectors, SPs are not paying for storage itself, meaning the data stored is garbage (zero value). With Fil+, SPs are literally paying to offer service to clients (instead of getting paid), which fully proves that storing the data onboarded with Fil+ has negative value. The economy sees everything. |
Beta Was this translation helpful? Give feedback.
-
The existing Fil+ (including 10x Power Multiplier etc.) has, to some extent, affected the health of Filecoin by making it harder for CC SPs to obtain fair incentives. But reducing the multiplier for the DC sector could result in instability in the global network's power and affect the enthusiasm of SPs to package real data. The current source of DC rewards is key to the existing problem. Initially, DC rewards were supposed to come from the Clients but due to a lack of customer payments in the early stages, DC rewards needed an additional source (unrelated to power), so as not to disrupt the previous consensus in Filecoin. Hence, we propose the following solutions:
If you have any suggestions, feel free to comment below. @raghavrmadya @dcasem @steven004 @jennijuju 中文版: |
Beta Was this translation helpful? Give feedback.
-
Overall, I personally do not think it is super urgent to (gradually) abandon FIL+, instead, we need to improve the governing process (I mean a very HUGE CHANGE to the current status). In terms of the problems listed the this proposal: 1&2: the consensus overload and complexity are already there, they will have very limited harm if we keep them this way. 3: I don't think this is a significant issue. 10x power, 10x collateral, if you have a lot of Fil, pledge more, if you don't pledge less, you get what you pledge, fair and simple. 4&7&8&9&10&11: This is exactly what concerns me most. We need way much better governing on Fil+ than current. I filed this amateur complaint here also (filecoin-project/notary-governance#965 (comment)). 5: This could be a problem, but canceling Fil+ would hardly help now IMO. Without Fil+, we could be seeing even less RBP increase due to the high cost and low FIL price. 6: I am not 100% understand the logic behind Fil+ and market price of Fil, so I would not comment on this. 12: I think this is acceptable, if the KYC and DD are done in proper ways. |
Beta Was this translation helpful? Give feedback.
-
The debate is consuming. We often tend to rationalize our actions after the fact. What's more, for the FIL+ staff and authors of FIL+, after three years of work, it is tough for them to deny their actions and work. This inspired me, it was because tokens and power were highly concentrated, it took community members 3 years to gradually realize that the community and PL’s disagreement in terms of vision and values has reached the point where the entire network is endangered. Why can't the debate today happen earlier? What if we decentralize power and let the huge amount of vesting filecoin be used strictly and transparently under the community's supervision? I think it can make us aware of the problem earlier when we make other wrong decisions in the future, and advance the debate we're having now. We don't intend to underestimate anyone's work. It's actually Filecoin core protocol doesn't need FIL+ and multiplier as subsidy from early beginning, but we community realize it too late. So it's more practical we skip the obsesses of details and numbers that are discussed now, start correcting the mistake(I personally insist Option1) and pushing a decentralizing and transparent evolution to the Filecoin community. If we can make filecoin a more transparent and open framework, we can stop losses and correct deviations in time, and truly let the market test the demand for products and eventually skip the next consensus crisis. The second-best time is now. We don't want to make the same mistake twice. If we do it right, we can call it an upgrade, not a fork. |
Beta Was this translation helpful? Give feedback.
-
I am quite surprised that the mechanisms proposed in various comments here that allow all sectors to take advantage of multipliers are not being taken more seriously by the authors. E.g. from @nicelove666 (#774 (comment)) @stcouldlisa (#774 (comment)), @kkarranscu (#774 (comment)) @herony-fil (#774 (comment)). Let’s call this class of proposals “free multipliers”: the SP has a choice of multiplier (1-10x), free from any governance or process restrictions. These proposals strike me as highly aligned against the primary problems stated by the original proposal. There are definitely variants available and details to be figured out, but they strike me as:
I see two categories of objection to the original proposal:
The free multipliers proposal immediately dispenses with the second objection, leaving the debate to be about network goals and whether subsidies etc should exist at all. No core protocol designers have identified any fundamental problems with the idea (yet). CEL have published initial analysis which says it’s basically fine. The worst thing anyone can say about it is that it achieves the stated goals. It’s a gift! Immediate cutoff of multipliers is simple, but doesn’t meet the network stability objective stated in the OP. The authors in comments seem to admit this, but express a view that it’s a necessary cost of solving the identified problems. It isn’t necessary. Because the design space is large, and comments here are necessarily brief rather than fully worked out, there may be some misunderstandings. The biggest one seems to be:
The proposals aren’t to require 10X collateral, but make the multiplier a free choice (between 1 and 10) for the SP, regardless of data in the sector. Please stop attacking this straw man. With free choice of multiplier, cost of sealing any RBP will initially remain the same, and then decrease as some SPs opt in to higher multiples, presumably slightly faster than under the FIL+ system (all those who would have gone through FIL+ still get multiplier, plus it’s open to others). The minimum initial pledge per sector will decrease as the network pledge:RBP ratio increases. The more pledge there is on the network, the less new onboarding needs to put up, since the total pledge is targeting a 30% of supply, regardless of power. Limiting multipliers to 1x would actually increase the pledge that today’s 1x sectors need to put up in the future. As older multipliers and their pledges expire, the circulating supply will increase and pledge requirements will too. And that’s after the initial shock of uniformly unprofitable onboarding removing the circulating→pledge flow that keeps pledge requirements stable today, which would give a sharp supply and pledge increase. Another misunderstandings phrased as objection:
Limiting multipliers to 1X doesn’t either. In fact it is likely to be much worse. By making all onboarding initially unprofitable, RBP is likely to drop much faster, contributing to the reduced block reward (further reducing profitability etc). On the other hand, with free multipliers, CC sectors (with opt-in multiplier) can become profitable. And deal-makers can onboard faster, with fewer barriers. There is good chance that the net RPB onboarding rate would increase since it is easier to onboard sectors and SPs can chose a multiplier that works for their economics. There are some mentions of altering the baseline function, e.g. to switch to QAP. Some changes here might be required anyway, they’re certainly part of the design space and worth exploring. Another objection that’s slightly off-target:
This is a totally valid question to ask, we should ask it and discuss how to improve the basis for Filecoin consensus security. It’s generally poorly understood at the moment, and some colleagues are working on better documentation. However, it’s mostly beside the point here because the proposals end up in basically the same place. Whether multipliers 1x or 10x, the network pledge function targets 30% of the circulating supply, which is independent of QAP or RBP. The total pledge contribution to network security is independent of any multipliers. If all multipliers settled at 10x, then the distribution of rewards to RBP would be exactly the same as if they were all 1x, as would the sector pledge, and FIL-on-FIL and fiat returns. I think this convergence is quite likely. At that point, the concept of multipliers could be removed entirely. Until that point, though, multipliers represent an SP choice over hardware vs financial-capital intensity:
Rational choices would depend on each SP’s relative cost of obtaining hardware and pledge tokens, and revenue mix between block rewards and storage payments. An SP with good paying clients (like @dcasem?) could continually chose min pledge, and have their unit economics improved by the fact that some others pledge higher amounts to improve per-TiB rewards. Their high pledge is the service provider’s advantage. Ironically, I think we might see higher pledges for empty sectors rather than full ones, reversing today’s forced pattern. Nothing is perfect. Would things be simpler with no multipliers? Yes. Although maybe the flexibility introduced by free choice of multiplier is an unexpected boon to more varied SP business models. The free multipliers “looks bad” because the concept of QAP hangs around, though it won’t do much. We can get rid of it later if/when multipliers converge. Incorrect inferences like "all pledge becomes 10x" are easy to make. But it directly addresses the primary motivating problems of the original proposal, appears to have no detrimental effects on network macro economics or security, and is possibly even advantageous to SPs focussed on client-paid deals, because they’ll enjoy lower pledge than if they’re forced to equalize. Please accept the implicit offers here to find the best version of a change that could achieve your key goals, so that we can lift the discussion to focus only on whether those are the right goals. That’s a discussion worth having! |
Beta Was this translation helpful? Give feedback.
-
@anorth didn't comment under your original thread as this's a superficially nice but dangerously consequential proposal.
CC sectors should not use the same Power Multiplier range as real data sectors, because the difference in operating costs between these two is so large that it will directly lead to a large number of power driven SPs directly favoring the simpler CC sectors. Filecoin would move further and further away from its original vision because of reaching the small probability requirement of @dcasem.
Three clear and bad results would be:
Thank you! |
Beta Was this translation helpful? Give feedback.
-
I have observed that no one has discussed it for a long time, so this proposal is not passed? It didn't even enter the community voting session? |
Beta Was this translation helpful? Give feedback.
-
In the blockchain world, what has happened cannot be changed. If it is changed, there will be a risk of fork. Therefore, regarding option 2, I think it should not be supported. Regarding option 1, we can discuss more, but we need to discuss more than just canceling DC. If there is no better solution, how to encourage SP's initiative after canceling DC? |
Beta Was this translation helpful? Give feedback.
-
#803 points here for discussion, and I was asked how the unrestricted datacap proposal might interact with direct data onboarding #730. I don't see any interactions. The changes seem orthogonal and compatible. When using Fil+, the direct onboarding workflow requires a client to directly create a verified data allocation on chain by spending datacap. The only difference this proposal makes is that the client can mint their own datacap before doing so, rather than waiting for the Fil+ governance process to give it to them (or refuse to). Unrelated to direct onboarding, but the very simple mechanism proposed here does have a side-effect that when an SP only wants to commit capacity, but do so with a high multiplier, the verified allocation they claim must specify a data CID, which must of course match the committed data. (Before direct onboarding, they would do a full-on deal with the built-in market; after direct onboarding they can just directly commit the data). This would have the effect of giving the sector verified deal weight, and making it not appear to be CC, and therefore not eligible for subsequent snap deal with client data. I think we can easily work around this by detecting explicit commitments to the "Zero" unsealed CID, which is the same unsealed CID that CC sectors use. Even though this unsealed CID is not stored on chain, we can set a flag in sector on-chain info to indicate that the sector data is zero, despite having verified deal weight, and that it's therefore a valid target for snap deals. |
Beta Was this translation helpful? Give feedback.
-
The attitude of PL staff is very disappointing. You behave like the religious supporters who denied heliocentrism back in the day. Filecoin community members should all be ashamed of this proposal not being taken seriously. Supporters of Fil+ have repeatedly replaced science with stance, and this unscientific and reckless atmosphere has permeated the filecoin ecosystem. I don’t need to identify who is encouraging this atmosphere. To be clear, "cc is useless" is absolutely a stance not a science. The biggest challenge to Filecoin currently is the distortion of the crypto economic model brought about by Fil+. It's the top priority. Please pay more attention to priority management. Until this vital problem is solved, I don't think any network upgrade is necessary. |
Beta Was this translation helpful? Give feedback.
-
I have created a new discussion topic #844 for the proposal currently open in PR #803 "Removing Restriction to Access to Multiplier and Allowing Unrestricted Minting of Datacap". Please take future discussion of that proposal to that new topic. This discussion topic remains the discussion reference for the proposal in PR #788, "Draft of Phasing Out Fil+ and Restoring Deal Quality Multiplier to 1x" |
Beta Was this translation helpful? Give feedback.
-
interesting |
Beta Was this translation helpful? Give feedback.
-
PR #788 was merged as a first draft of FIP-0080 (I've updated the discussion title to include that). Moving a post from @Fatman13 here for discussion:
|
Beta Was this translation helpful? Give feedback.
-
o yehh!! |
Beta Was this translation helpful? Give feedback.
-
中文版 | Chinese Version
authors: @The-Wayvy, @Fatman13, @ArthurWang1255, @flyworker, @stuberman, @Eliovp, @dcasem
type: Technical
category: Core
created: 2023-08-02
Foreword
This FIP Discussion aims to explain why we want to sunset Fil+. We will discuss the drawbacks of Fil+, general design principles that should be respected, and our proposal for a reduction in the quality multiplier of Fil+ deals.
Background
Filecoin Plus programs introduced a permissioned layer to the Filecoin network in an attempt to maximize the useful storage that Filecoin could support.
A Deal Quality Multiplier was created to subsidize storage providers for participating in the Fil+ program, giving Storage Providers (SPs) who stored these verified data a 10X multiplier to their storage power (a 9X increase over non-Fil+ deals) and increasing their share of block rewards. The multiplier was intended to help subsidize the acquisition of data clients and the building of tools and processes for data onboarding.
A group of human notaries was selected to distribute the DataCap. The program evolved the following categories:
Problem Statement
Fil+ has attracted significant verified client and data to the network and has accelerated the development of tools and processes for onboarding data. However, the impacts of Fil+ (both the power multiplier and the off-chain processes for distributing Datacap) have also endangered Filecoin’s growth trajectory:
Starting in Q1’2022, Fil+ accumulated significant mining power in Filecoin, accounting for the vast majority of new power added to the network.
The hardcoded superiority of Fil+ deals has killed off regular sectors and deals. Doubts about the usefulness of data onboarded led to the creation of a new Fil+ roadmap, which requires greater control of Filecoin by the notaries and governance team. The direction proposed by the Fil+ team contradicts the industry’s core value proposition: decentralization and efforts to entrench permissioned mechanisms via the Fil+ and notary bureaucracy have been met with skepticism, dissatisfaction and determined resistance.[1] [2] [3]
We all want useful storage and a meaningful number of paying users on Filecoin, but it’s crucial to reassess whether or not Fil+ will help us accomplish that. The substantial loss of Raw Byte Power, lack of paid storage services, and endless conflict in our community highlight the urgency of resolving this issue. Filecoin’s future is at stake. Now is the time to act.
Design goals
Filecoin’s mission is to provide useful storage. The protocol design should reflect this mission and ensure that Filecoin becomes more useful (with more real data) over time. However, for the use of a blockchain to make sense, the features uniquely enabled by blockchain technology must not be discarded. Our proposal and position on future proposals are guided by the following principles, which are almost taken for granted in Bitcoin, Ethereum, and the wider world of blockchains:
Proposal
Keeping the above principles in mind, we come to our proposal.
As calls for removing human influence from the distribution of block rewards grow louder, the network becomes more useful following the release of the FVM, and the data onboarding tools and processes that Fil+ subsidized approach maturity, we believe the time has come to phase out Fil+.
We are opening this discussion to align on a strategy for gradually diminishing the influence of Fil+ programs on the Filecoin Network, aiming to unburden Filecoin’s permissionless segment, restore objectivity and neutrality to our consensus algorithm, and incentivize the development of sustainable storage services for users.
If a transparent and objective mechanism for evaluating the ‘usefulness’ of data is shipped in the future, the power multiplier can be reconsidered.
We expect the community to develop many proposals on the optimal schedule of cuts to the multiplier. To get the conversation started, we propose two examples below. This discussion aims to get consensus on the need for a commitment to a definite schedule for getting to 1x, and hopefully, the details of how we get there will come up concretely from this discussion.
Option 1:
Stop new Datacap allocations immediately. Eliminate 10X multiplier upon FIP finalization in an upgrade. All new sectors onboarded after the upgrade will have raw byte storage power, effectively removing the deal multiplier. For existing DC sectors, keep the current QAP until these deals expire (will need some technical solutions). Upon sector extension, the 10x multiplier will no longer be applicable after the upgrade.
Option 2:
Decrement the power multiplier on Fil+ deals from 10x to 1x over a scheduled period. For example, decrease the multiplier to 5X for the first 6 months. For the following 6 months, further decrease the multiplier to 2.5x, and so on. For existing DC sectors, keep the current QAP until these deals expire.
In either case, a commitment to a definite schedule for this transition will provide much-needed certainty to network participants about the future of Filecoin. It will remove the primary source of division in our community and enhance perceptions of Filecoin as an open and fair economy.
Both approaches still encourage the onboarding of useful data as the underlying technology for creating alternative revenue streams for SPs improves.
By way of example, let’s assume there are two types of SPs:
Type 1: A CC-only SP which is self-dealing (simply providing security to the network)
Type 2: An SP making paid, unverified storage deals with clients
Over a long enough time period, SP Type 2 will win out over SP Type 1, because SP Type 2 will be making more money with paid storage deals, and will be able to subsidize costs, including pledge collateral.
Call to Action/Open Questions
Changelog:
Beta Was this translation helpful? Give feedback.
All reactions