Upcoming EMC(EdgeMatrix Computing) SNS Decentralization Sale

What will be in place to prevent the following scenario?

during Decay Period(1), while df(1)=1 and p(1)=100%

Reward for computing power user: R(u) = M(tx) * P(n) * df(n) (n ≥ 1, n ∈ N)
Reward for computing power provider: R(p) = M(tx) * P(n) * (1-df(n)) (n ≥ 1, n ∈ N)

As we see here Users gets 100% transaction cost return. Nodes gets 0%. Same holds true for Decay Period(2).

Node A places excess computing power for sale on marketplace. Then User B(Node A alt ID) purchases this via contract. Once completed the contract is reposted and the process is repeated until 24 hrs have passed. After receiving the daily reward User B then repeats the process again effectively draining the reward pool on a daily basis limited only by the computing power of the node.

This reward pool accounts for 18% of total distribution or 378m EMC. Utilizing the example you provided of a 1000 EMC transaction in DP(4) this would be drained in the first 378k transactions if these transactions occurred within DP(1)or DP(2).


It’s a thread that every technician who wants money should read.
Making SNS a DAO for a truly great product team.

1 Like


As a part of creating the technical report on EMC, I am going through the links that you have mentioned in response to my query. I have an extensive background (20+ years) in software , hosting , virtualization etc. I understand that you have no way of verifying my experience. But you will get to see my experience play itself out in my questions and commentary as we go along. You can also see my posts on forum.

I have gone through the document links that you have provided. When the link discusses the sdks, it states “it provides a rich set of…”. Since the link talks about the present tense, can you please let me know where these sdks are and what their specs are? Conversely if you don’t have these sdks just yet, then can you please let me know as well?

I will have some other questions as well. I would rather have a participatory discussion with you and others in the team. However if you think that this might be too much time commitment, i understand. In either case, i will include any response or otherwise in my report.


Looking forward to it, and thank you for taking your time.

The only SDK on github GitHub - EMCprotocol/emc_java_sdk states " to work with the EdgeMatrix Decentralized Computing Network, without the additional overhead of having to write your own integration code for the platform." It is used with MAVEN (build tool for Java). So no idea what they mean with "(EMC is open-source; its design is public, nobody owns or controls EMC and everyone can take part.)!

When the source code of the node itself is planned to be released open-source? The link you’ve posted contains only Java SDK for interaction with node. I haven’t managed to find any info about it in whitepaper either. Sorry for duplicate, if this question was already answered

Hi there, so let’s just call it deep learning here, and for EMC protocol, I have two main questions here(for myself, about 5 year+ of deep learning experiences):

  1. about the data: If your team plan to perform model on-chain training action, how do you input the train data, and where is the data input from? off-chain or on-chain? What kind of data format is it? video or image or txt file? Is it stored on a canister or on centralized server?
  2. about the model: If you want to train the model on-chain, is there a plan to open a specific sub-net to run the train model epoch? how do you store the model weights? And to train the large model, there are needs for performing large scale of tensor computations, how do you deal with it?
    Above these questions, if you have something(whatever public code or document, please let us know)public, please post a public link here, but not the above java sdk. :grinning:

Any quick idea about locally deploying any AI app that provides a Restful API ? Running the edge-matrix node works as described (except the fact that one has no idea what exactly is running since no open source code) tried on Ubuntu 20.04 LTS and Win11 (flags everything just as it does with any mining software) but have little experience with any AI model, a Rest API using FLASK I can do, just waiting for someone to drop an easy example to continue testing. Thanks


The tokenomic of EMC white paper was designed 6 months ago, and now it is being adjusted according to the mechanism of SNS and suggestions from the community, reducing the voting power of the founding team and EMC Foundation, all for the purpose of being more decentralized, and soon EMC will announce the final adjustment results.


This is an interim status (with questions) on where I am with the evaluation of EMC.

I have been doing a deeper dive into EMC to create a “technical report” for the purposes of this community. This technical report is NOT intended to be financial or legal advice. It is meant for education/entertainment purpose only. With that disclaimer, here is some background about how I intended to do this evaluation.


  • I understand that certain amount of hype is necessary for the core functionality to be attractive to the community. However, to the extent possible, sticking to the facts makes it easier for an evaluator (like myself) to evaluate. Also I draw the credibility of a source to the extent that that source is credible in statements other than the core functionality.
  • I take the whitepaper/ other technical business articles to be authoritative; meaning that they are fair game for detailed scrutiny.
  • I understand that certain jurisdictions have some limitations in context of disclosing real identities and therefore may have difficulties in positively identifying their experience. The best guidance for such folks is to demonstrate their intent and experience through code commits/twitter posts/discord discussions/telegram channels/ Youtube Vlogs etc; all of which is at least partially anonymous
  • The Web3 experience means that the teams must be responsive multi channel on a timely basis. The crypto world is 24x7 and therefore the founding team must so responsive.
  • The questions are expected to be answered (if they are) in the literal sense. Please don’t interpret why I am asking the question. Anyone is free to discard the question as being irrelevant if there are logical arguments for discardation.
  • No bounty to be accepted from the interested team; so that there is NO perception of bias.

Where am I in my evaluation

  • I have gone over this forum topic, the technical whitepaper (From The Origin - WhitePaper of EMC), the website (https://edgematrix.pro/), github(EMCprotocol (EMC) · GitHub)
  • I am lurking on their telegram channel as well as the chatbox to figure out the kinds of interactions and responsiveness to queries
    ** For example, there was warning from someone called Sam to prepare adequately for SNS launch because he was aware about the Sonic thing
    ** Currently I am not seeing a lot of traction of chatbox. For example if I join DaVinci Diffusion channel, an offer for help is presented through a robot; however when I ask for help, none is offered.

Whitepaper Deep Dive Issues

While I am merely pointing out the issues in this update in order to get clarifications, there are also many more positive aspects.

  • Product Overview
    ** More details into how IC figure into the vision as described by the product overview. While there’s whole section around Internet Computer Protocol, it is not referenced in the product overview at a sufficient level of high level lingo.
    ** In an similar line of thought, more forward reference to the EMC Dao is required to explain the overall philosophy of the DAO

    ** I presume that this is still articulating the vision (as opposed to implementation) of the DAO.

  • Utility Token of Computing Power
    ** I presume that the wording, in this section, is authorative because it creates an expectation on already distributed EMC tokens through the presales etc. if my presumption is incorrect , please correct.
    ** The expectation that it creates in the present is “people who hold $EMC can use it to purchase computing resources, and computing nodes can accept $EMC as a payment method for transactions”. How can a $EMC holder currently purchase computing resources?

  • Tokenology of $EMC
    ** leaving this alone because of potential rewrite.

  • Roadmap
    Again I consider this section to be authoritative. Specifically I am interested in knowing about the open source EMC Protocol as well as implemented IC Smart Contracts and Client for Node Service. This should be in github with all previous commits (to give the sense of continuity.

Where I need help

EMC Team

@EMCprotocol @alxgoh @ninadou
Please review the commentary above (and previous question) and respond appropriately in a timely fashion .

Dfinity Team

@diegop @zire
According to EMC, Dfinity has done it’s DD on EMC. Please provide, to the extent possible, what the purpose of that DD was and what was the result of the DD.


I believe that you have been accepted into Paul’s incubator fund. Congratulations! Can you please let me know whether you tried to run the service from EMC or are going to try.


Please add any additional information (or DM me ; if you wish to stay anonymous)


This is all great and all, we thank you for doing it taking your time, as a sidenote did you end up running the edge-matrix node ? I am stuck at the local AI model as in no idea which one to use. Other then that everything seems to work as described also just my 2 cents it all runs on very low resources so not sure why the top of the notch hardware requirements in the whitepaper.

Thanks @ZackDS for the compliment.

I have NOT tried out the edge-matrix; precisely because i thought i would hit a dead-end (like you did). The only way to get around that is for additional instructions and_or open sourcing the code. Additionally i don’t like installing stuff from unknown origin without the source or reasonable assurance that it is free from malaware.

On the top notch requirements for the hardware, my interpretation in a white paper, is that is aspirational (i.e. not currently present). However i do note that somewhere EMC claims to have 150 nodes. So i presume, that if correct, then at some of those nodes should be high end. But again, i think that these are recommendations not requirements. IMO having the ability to consolidate multi-type nodes into a cohesive workload would be a game changer; especially for inference. I can see how @MCLYang is interested in it from a music generation standpoint

1 Like

Yes I stated in a previous post that Windows flags everything just as with any mining software and would be nice to see the code. I had a spare 500GB SSD so I ran it on bare metal both on Win11 and Ubuntu 20.04 just to avoid them saying it was not meant for virtual machines, that would have been way easier. Anyway what I found strange is that they argue they are working 24/7 and have no time to argue but then again with the EMC (ICP L2) AI computing Testnet Launching! Nodes whitelist recruitment overfunding collection! - #10 by stopak post they said “Any PC users can participate in the test network” and when I asked about hardware was referred to the whitepaper so I highly doubt that 100 people did that testing with that criteria. And also 17 days ago “The POC algorithm is doing final environment testing and we will commit to github soon.” still nothing to be found.

Sometimes i am guilty of that myself. When my PM asks, “are we done yet?”, i ignore for a whole day because i am in the middle of the code. I can imagine how it is to be on the opposite side. I think having a team that can insulate the core-devs is pretty important in Web3.

This seems like a great start to DD on EMC - have recently posted my SNS criteria for similar reasons. Looking forward to the results of your due diligence, thank you for your time in advance!

Understandable I just think that if the final ENV testing is running for 17 days now with a huge dev team raises an eyebrow. Time will tell we shall see what they come up with, also @cymqqqq has 2 very good questions I would like to see an attempt of answer to.


Yes. I have taken note of your SNS criteria and they, for the most part, will be included in the technical report.

As i am sure you are aware, the purpose of having an participatory discussion is to ensure that we put an interested team in as positive light as possible; while at the same time learning about their perspective and they of ours.

For example, in this particular team, there appears to be a legitimate fear of disclosing devs real identities due to potential reprisal by the state. In that case, we have to suggest mitigation strategies that can, if provided, accomplish the goal of verifying continuous progress by an accomplished team ; but at the same time accomdate the team.


@Icdev2dev I am thankful and appreciate to the effort spend on evaluating EMC. While our dev team continue making progress on the development side of the project, I would share more context on other aspect related to organization, operation and where are we heading on the decentralization journey. Our team members will continue to published and update of source code to Github for further technical evaluation.
Summarising EMC values and what EMC brings to the IC communitites:
EMC seeks to solve AI computational challenges and global shortage of GPU supply for AI via decentralized computing and blockchain technology. The EMC protocol consists of 2 layers - Operation layer provides GPUs computing network (by pooling GPUs from abandoned ETH miners and idle GPUs on FIL miners) built as an L2 network of IC. The EMC protocol handles the task scheduling and task distribution for decentralized GPU computing nodes, optimizing for training AI models and running AI inference. Control layer with Ledger, consensus and economic model built on IC canister. IC Canister records the status of the state machine of the EMC operation layer (by optimism rollup to canister every 15-minute interval) and processes all economic transactions using canister smart contracts.
EMC expands AI computation capability and flexibility to IC networks. EMC repurpose GPUs used on hash mining (ETH POW) to power the AI model training and inference will be an Eco-friendly Web3 project and adaptable to ESG. While the IC network supports decentralization of general computing with homogeneous computing nodes, EMC L2 network added thousands of heterogeneous GPUs nodes to fulfill AI on-chain computing needs. EMC network processes and stores large data in Canister (estimated 0.9gb/15 min roll-up interval) lead to significant cycles burned.
EMC supports IC ecosystem development (especially in APAC). EMC adopts a POW mechanism to reward computing node providers (miners) and a POS staking mechanism to reward validator nodes and routing nodes will increase TVL on IC, by converting ETH and FIL miners to EMC nodes provider. EMC will build a Computing marketplace on IC, for users, developers and node providers to trade decentralized computing power. Economic incentives will increase participation of node providers, bringing AI developers into IC, and migrating users from web2 to web3. The Computing Marketplace also provides AI training models + computing power as NFTs (preloaded clusters with TensorFlow, Stable diffusion, Ready to train AI models) and pre-trained AI model NFTs (ready-for-production AI models).
Tokenization and Decentralization of EMC through SNS. The EMC founding team are advocates of decentralization, with a vision to build a truly decentralized computing to power the future of AI. EMC built an entire control layer on canister with the economic value of EMC network integrated on IC, SNS is the de facto choice for EMC tokenization and IDO. EMC Foundation was in-corporated in Singapore as an operation body, and EMC SNS as decentralized governance for development and direction of EMC protocol. SNS provides the effective mechanism for EMC decentralized fund raise and establishment of DAO. EMC testnet is live and mainnet target to go live in July-Aug of 2023, which put EMC SNS launch before end of June.
EMC team is reworking on the tokenomic and distribution with aim of decentralized governance without centralize control of voting power. What EMC can disclosed, apart from grant from ICP, early individual seeding (1%) and presales of 0.5% token close on 22nd May (600K raised with max 1000 ICP/account), the project started as bootstrap from founding team, and has no VC raised as of this point in time.


@Icdev2dev In responds to your question on the whitepaper:

Product Development & Developers Tractions

  • Closed the 48hrs presale on 21st May for 0.5% of EMC tokens (coming from the early adopter pool). Presale Data: 12062817.7 EMC (114%) raised ~$600K
  • EMC minted 5% of DIP20 tokens for TestNET node staking and rewards (4.5% from ecosystem rewards pool). All DIP20 tokens will be 1:1 with ICRC-1 tokens after EMC SNS launch.
  • The Testnet is currently running smoothly and can be verified through POC; participating nodes will be rewarded, total 3M EMCs reward over 60 days.
  • Computing power nodes have implemented AI applications to validate the functionality and stability of the EMC network.
  • The IC contract has implemented three functions: node registration, reward distribution, and subscription.
  • 100+ computing nodes, 7 validation nodes, 10 routing nodes running on testnet.
  • EMC source code is open sourced in our Github repository.
  • 5 AI project teams are now building on EMC testnet, with two applications integrated into the architecture, scaling deployment of more computing nodes.

Thanks for the DD by Icdev2dev. This contributes to building trust between the EMC team and the ICP community.

Due to the time zone difference, most of the EMC team is still located in the Asian time zone, which may result in delayed responses. We also have contributors from North America, and we are actively building an international community through the IMN program (International Media Nodes) which we learnt from BanklessDAO. We are also in the process of establishing a North American team to become a truly international ecosystem and community as soon as possible.

Regarding the information gaps, some team members in a semi-public capacity due to certain policy reasons. However, this does not affect the vision, architecture, roadmap, ecosystem, business, or the DAOization of the EMC project. We will release the coupled technical architecture diagram of EMC and ICP, open-source all the code (before the formal SNS proposal), launch a visual AI demo deployed on EMC and ICP, and fully implement the first real AI commercial application. These efforts are actively underway, and we will eventually show the codes.

In terms of team background, we have a founding team with experience in Web2 publicly listed companies, Web3 multi-chain development, mining and node experience in BTC/ETH/Polkadot/Filecoin/Cosmos, professional investors in primary and secondary markets, executives from multinational corporations in the Asia-Pacific region, and architects with expertise in large-scale communication and billion-scale networks. Currently, more developers are joining the EMC ecosystem as full-time or part-time contributors (Openverse, AI applications). These details will gradually be showcased in the upcoming project implementation, developer interviews, and AMA sessions.

If you are interested, you can DD the Ethereum projects I have been involved in, such as Status (https://status.im/), as well as ethfans, DAOONE, and EnlightDAO. These accounts have had a certain level of influence in the WeChat Official Account and Chinese community, advocating for the crypto industry and the Ethereum community.

As long-time builders, developers, investors, and participants in the blockchain industry, we believe that the development of blockchain must be based on the vast external market rather than internal financial games. This is the foundation of our strong belief in the AI+Web3 track. We hope that more people can see this future and join us in exploring and realizing this vision, starting from turning idle Ethereum miners into AI computing power sources.

Lastly, please be patient for a few more days as we gradually reveal the relevant technical details, code, demos, and an updated tokenology that aligns with SNS.