Price of Identity Forgery Protocol
Original idea →How Upala works
Grosso modo, the idea is to use market forces to accurately price identity forgeries. In other words, how much does it cost to create a fake for a given identity claim. The goal is to measure sybil protection, human verification mechanisms in real value.
With smart contracts, we can achieve credible neutrality. The market rules are mostly to protect forgers and must be fair to minimize market distortion so that they actually use this market and don’t retreat to their current black market. Prices are computed by the market dynamics (Forgers VS identity providers).
A group controls admittance and set stake amount required for entry, fees and also a score that is strictly higher than the stake plus fees. To assess a user reputation, a verifier has to check membership status and sum the scores of groups they trust that a user is a part of.
Once a member, users can either, stay in the group, exit with their stake or liquidate their entire identity and get the cumulative score value of all their memberships.
The stake, fee and score must be adjusted continually by groups to react to market dynamics.
The dream is to have digital identities worth more than state issued ones.
Glossary
User: Entity who is a member of one or multiple groups.
Verifier: Entity who calculate the score of a user.
Group: A set of users.
Stake: The part of the entree cost that is kept by the group to pay for future liquidations.
Score: The value of the potential liquidation and by proxy the value of the membership.
Fee: The part of the entree cost that is directly given to the group.
Group Liquidation Ratio: The percentage of members that can liquidate before insolvency. GLR% = (total stake / (score * member count)) * 100
Group Liquidation Profit Margin: The percentage difference between the stake plus the fee and the score. GLPM% = ((score / (stake + fee)) - 1) * 100
Actions
User
Apply for group membership: Must pay a fee. If accepted stake must be deposited. If stake deposited, membership and corresponding score acquired.
Exit a group: Stake is recovered, membership and score loss.
Liquidate: TOTAL score amount is recovered, ALL memberships and score loss.
Group
Evaluate User: Should be private information. For a fee, evaluate a user membership request then either accept or reject it. Should be timed, after times up fee is refunded.
Adjust Parameters: Must be public information and should be time gated. Must hold Score > Stake + Fee.
Deposit or Withdraw Stake: Must be public information and should be time gated. Changes by proxy the group liquidation ratio!
Change Liquidation Method: Must be public information and should be time gated. Choose between; ratio-based liquidation, first come first served, etc…
Incentives
Users
stake, score, group liquidation ratio, fee High stake to fee ratio allow users to exit without losing too much but adds friction to entry. High score makes identities more valuable. High GLR means users can safely exit or liquidate. Higher fee, higher barrier to enter groups.
Groups
fee, score, stake, group liquidation ratio High fee, more group revenue. High score attract users but liquidations cost more. High stake prevent low score and big stake pool can be borrowed against or used elsewhere but is a barrier of entry for users. Low liquidation ratio means liquidations potential loss are limited and stake can be used for other purposes but users will not trust the group if the ratio is too low.
Privacy
Groups may sell membership information or access to it. This can’t be prevented depending on the information gathered during verification and if the group is pro-privacy or not.
Memberships must be private or at least not tied to users. A ZK proof could be generated for verifiers.
Metadata leaks is a concern since liquidation must be for all memberships. By using a single ID, who join what group and when could leak even if IDs cannot be tied to users directly.
Examples
Trusted circles
This example is the simplest but maybe most powerful. The group is administered by one person and only people they know and trust are admitted. The admittance test is repeated real world interaction and trust building over a period of time. The fee can be set to zero and the stake high since we can assume that both persons would enter each others group. The score can also be high since the admittance is very strict. The only down side is the limited utility. If I meet someone with zero group membership in common then I can’t assess their score. The famous “I know a guy that knows a guy” can now be verified and quantified 
Identity Businesses
This group sole purpose is to make money via the protocol. Cheap, fast and sound verification methods are the focus. Since users are already verified anyway and are also impacted by liquidations they become investors of the company via stake dividends. Stake pools are borrowed against to raise fund for research, new verification pipelines or new groups to address different market segments. Bad company go bankrupt while good ones flourish. 
Phenomenons
Bank Run
When a group looses members via liquidation their stake pool shrink. As the pool shrink user are more tempted to either liquidate or exit leading to a snow ball effect.
Group of groups
Automatic entree into a group for being a member of a specific set of other groups could be viable economically.
Cartels
Multiple groups and users all controlled by one entity with the goal of either stealing fee and stake or fooling verifirs.
Misc.
-
Group Expulsions
-
Can groups front-run members identity liquidation?
-
If technically impossible then expulsion should be allowed.
-
If it can’t be prevented, it would distort the price since a user would never be guaranteed to get the money when liquidating.
-
-
-
Partial Identity Liquidation.
-
By not allowing this, it makes the market less dynamic. A user can only make a binary choice to have the identity or the money. A forger’s strategy would be to sell “turn-key” identities.
-
With it, a forger can more efficiently target vulnerable groups. Pre-existing score can be used to fool any future admittance. A forger’s strategy would be to “farm” vulnerable groups until they adapt or bankrupt. A user would just liquidate instead of exiting if there’s no down side…
-
-
Individual user stake, fee, score in groups
-
Flat values for all members simplify administration.
-
Can always start a new group with other parameters.
-
-
Parameter Dynamicity
- What is the effect of members gaining entree with a set of parameters but those parameters changing over time? If score, stake or fee go up, users benefit but group may ask for additional stake or expel. If stake go down, users might want to exit then reapply with the lower stake or ask for refunds. If score decrease then users might exit.
-
Addresses as identifiers can’t be revoked or delegated and are not useful in other systems.
-
Use VLADs instead.
-
Use ENS domains instead.
-
-
Membership should be private
-
A ZK proof could be generated that aggregate every memberships. It can then be reused.
-
How can a destroyed identity be removed system wide if membership are private?
- RLN commitments should be taken as inspiration. One identifier not tied to any identity if desired. P-logs could contain the commitment, encrypted or not.
-
-
Malicious groups aimed at stealing stake and fee from honest users
-
How can the malicious admin distinguish honest users from other forgers?
-
Over time, groups would develop a reputation.
-