porobov commented 5 years ago

Request for Nest membership and funding (#60)

Team name: Upala

Research whitepaper: whitepaper | articles

Burn rate: $4,5k/month

Legal structure: NA

Team

Roadmap

Proposal

Upala is a digital identity system. Its purpose is to distinguish people from bots and clones (people with multiple IDs). Upala proves that a person has only one ID while preserving anonymity.

There is a similar proposal for a grant by BrightID.

The main difference of Upala is in the sybil attack protection mechanism. BrightID is doing sophisticated social graph analysis (doing a great job). Upala utilizes face-recognition and location proof, catching sybils when entering the system.

Upala sybil attack protection mechanism

Components: smart-phone app, face recognition server, FOAM’s proof of location service.

In order to confirm that two people are twins we need to see them together — same time, same place, same photo. But a more realistic approach is to witness them at the same time, but in different places (with proof of location). Here is how registering a new user looks like in short:

Take a picture of a registering user using Upala app.
Face-recognition algorithm detects twins among existing users.
If twins are detected they need to take selfies within a certain time interval in different places of the world (with proof of location).
Random validators confirm that the new selfies depict the same people as their ID photos.

A failure to do this procedure would mean that there is a sibyl attack.

This is a draft solution. It just shows how face-recognition and location proof can work together. Research on initiatives will shape the infrastructure and other details.

Upala benefits

The benefits we expect:

An ability to create independent clusters while preserving trust No need to have a chain of direct social connections ending with founders. New clusters may be created around the world independently.

Easy and fast onboarding The number of connections for an individual to receive a high score is much lower than that needed for social graph analysis. The amount of work is lower for users. Most users receive high scores almost immediately and with little effort (if not suspected as a duplicate).

A wider range of possible incentives We want to separate customers and node runners. We don't expect customers to run nodes or any computation. Customers buy information from network nodes (validators). Nodes are incentivized to do validation and store data. We believe this approach provides room to experiment with initiatives.

In general, our design approach is: it does matter who you are - face-recognition, where you are - proof of location, and who you are friends with - reputation. We will try to find the best balance between accuracy, usability, and incentives.

Bolder ideas

Depending on how good the face and location proof mechanism will perform we'll try to squeeze out some more benefits. These are some bolder opportunities we would like to explore:

Hidden social graph - better anonymity An open social graph may be prone to deanonymization. We'll try to abandon social graph analysis. This will allow us to encrypt the graph while using it for some simpler reputation mechanisms.

Recover account with face-recognition, location proof, and friends The boldest. Hesitated to include it here, but we do want to put some effort into this too. We'll try to replace login-password pair with face-recognition, proof of location, person's name and friends approval. This may boost adoption significantly - that's why we believe it is worth trying.

Adoption

We are focusing our research on adoption.

An identity system is useless without users. And there is no reason for a user to join a system which doesn't work. This vicious cycle is hard to break.

Adoption is the key to building a worldwide identification system. And we believe there is a way to make the system useful before it is widely spread. We will try to find this way.

Deliverables

We will deliver whitepaper (early draft here) describing the best solution we find. While fulfilling the task we will continue to publish our thought proceess on Medium and Reddit for the community to discuss.

CLAassistant commented 5 years ago

All committers have signed the CLA.

stefanobernardi commented 5 years ago

Thanks a lot for this proposal, very cool to see.

Unfortunately, Nest is not really for pure research products and most definitely is more suited to tools that have currently already started with some development and are very close to Aragon's software stack.