SCALE-MAMBA and BLOCKCHAIN integration

[Julen Bernabé]

Hi,

I'm currently trying to integrate SCALE-MAMBA and Blockchain (in this case Ethereum) with the main goal of giving more flexibility to the way the SCALE-MAMBA network is deployed. However, I wonder if some research work is done before in this context. I've been searching, but literature says there is no research done so far... ¿Is that true?

Thanks beforehand,

Julen

[Nigel Smart]

It depends what you mean.  Some people have looked at MPC
in the context of blockchain, but the specific use case is always
a bit weird. Using MPC directly on the blockchain where the
validators are "MPC-nodes" to me looks like it not suitable for
permissionless systems such as Ethereum. Of course
MPC can be used by specific nodes/wallets etc do perform
transactions [this is well documented/deployed].

So really depends what you mean

Nigel

--
You received this message because you are subscribed to the Google Groups "SPDZ/SCALE-MAMBA Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spdz+uns...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/spdz/CAAvRsseze8si%3DaeJThbNyik-Qznfs3n5d2xra%2B%2BfJW3R-FQpCw%40mail.gmail.com.

[Xiang Xie]

We are currently working with Ethereum Foundation to implement the MPC version of proof of custody for Eth2.0 (about BLS signature and Legendre PRF)

The first basic version is based on SCALE-MAMBA (actually only the scale protocol, not using the compiler).

Check the following repo, if you are interested.

https://github.com/PlatONnetwork/proof_of_custody

Best,

Xiang

--

You received this message because you are subscribed to the Google Groups "SPDZ/SCALE-MAMBA Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spdz+uns...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/spdz/CAAvRsseze8si%3DaeJThbNyik-Qznfs3n5d2xra%2B%2BfJW3R-FQpCw%40mail.gmail.com.

--

Xiang

[Julen Bernabé]

Answering Nigel's questions:

1. ¿Why a blockchain? The idea is to build a data marketplace. There will be users that might want to do some data analysis (or even train their AI algorithms) with confidential data or with a database they do not have. On the other hand, there could be users that actually have that data and may be interested on giving it (always thinking about a reward). Hence, blockchain gives us some crucial advantages:
• No frontiers: users from all over the world could be able to obtain some reward for giving their data. New data communities can be created this way (of course, those communities once created can be fixed and might not need blockchain again, but what if someone else wants to enter the community?)
• Data sovereignty: for me, it doesn't mind who is giving the information, but the information per se. That is what the reward will be for: we are talking about users that can help each other but do not know about the existence of each other.
• Algorithm identity: if I'm doing data analysis, I probably would want to protect the algorithm I'm trying to train. The data givers, on the other hand, could be interested on receiving some tokens for giving their data, that afterwards could be exchanged for using the algorithm when it is trained.
• No deniability: everyone could interact with the marketplace.
• ...
2. ¿What data are they processing and why? Well the data that is being processed on the blockchain is the one asked when setting up the SCALE-MAMBA network:
• The players' data: common names, certificates...
• The prime
• The number of players
• ... (this is only an idea, currently not implemented at all)

Recall that I do not want players to join and leave the MPC computations, but to give them some place to meet each other and make communities grow there. Moreover, it could happen for MPC computations to be unique (e.g. a user wants to know the mean of salaries of all hospitals in the world for a research) or they could even have a higher aim that can be treated from the blockchain (such as training an algorithm and the possibility of using that algorithm in the future).

I suppose the idea is more clear now, so is there some research work on this? That was my first question :-)

Thanks for Xiang too, but I think our use cases are quite different (I'll check the repo, though).

Julen

[Nigel Smart]

Looks orthogonal to MPC to me. This is more like a brokerage to
find people to do MPC with.

Suggest take off list, as this is not directly SCALE related

Nigel

On 3/16/2021 8:43 AM, Julen Bernabé wrote:
> but to give them some place to meet each other and make communities
> grow there. Moreover, it could happen for MPC computations to be
> unique (e.g. a user wants to know the mean of salaries of all
> hospitals in the world for a research) or they could even have a
> higher aim that can be treated from the blockchain (such as training
> an algorithm and the possibility of using that algorithm in the future).

--
Nigel Smart

[Alex Su]

Hi,  Julen

I am a developer from project ARPA, which is also trying to build a permissionless blockchain-based MPC computation network and data marketplace. We started to research and design our architecture in 2018 and met several problems during the implementation. If you are interested, please find the details in these places.

Our whitepaper: https://docsend.com/view/t69gzij

ARPA website: https://arpachain.io/

repo: https://gitlab.com/arpachain.io

In short, ARPA tried to build a layer 2 network listening to privacy computation requests on Ethereum. If someone publishes a request, the ARPA network can automatically distribute the task to several ARPA nodes and do the MPC. Considering SPDZ is a full threshold protocol, the involved ARPA nodes can prove the correctness of their computation results. This proof will be sent back to Etherum to achieve the consensus. Based on this layer 2 secure computation network, we will build a data marketplace that allows people to put their confidential data for profit like a DVD rental. They can describe their dataset and give some samples for renters to refer.

But there are several problems here:

1. How to prove the dataset quality if they are encrypted 

2. Algorithm encryption or say obfuscation is another hard problem, which cannot be done by SPDZ

3. The computation power and availability of data providers cannot be guaranteed

4. Full threshold version of SCALE (SPDZ) is too complicated and heavy. If we use the semi-honest or honest majority version of SCALE, it is hard to protect the dataset from corruption.

5. anonymity of Ethereum will be trouble in building this marketplace.

Anyway, this is a promising topic. We are still working on it and willing to discuss it with you in detail.

Alex