# New Bots or New Users? Toward a crowd-sourcing mechanism design of optimality gap.

###### tags: `Gitcoin`

## New Users or Low-laying Sybils
In GR11, we observe many contribubtors merge with brand new Github account but and seemingly legit contribution patterns. Some of them only contribute to highly reputable grants such as Gitcoin official dev fund and sybil resistence tech like POAP, others mix contribution to white list grants with less well-known and sometimes suspicious grants. Other users contribute to reputable grants in the first few rounds, and then turn towards suspicious greants.

While those users are automatically flagged and sent to human stewards to review, it's hard for human stewards to judge whether they are new users exploring the Gitcoin world, or low-laying bots waiting to strike in future rounds. Yet, if they are bots, there are clever and ill-intentioned humans behind them trying to reverse engineer the ML sybil detection algorithim to game the Gitcoin Quadratic Funding matching system for their own benefit.

The semi-supervised ML sybil-flaging process is intentionally nebulous. After all, if the detail of the algorithm is open to all, the sybil attackers will have no difficulties finding ways to game the system. Yet, such nebulousity also encouraged the sybil attackers to evolve with the detection algorithm. **As the algorithm gets better at detecting sybils based on their past behavior, the sybils come up with new strategies to hide themselves among the forests**. In other words, the sybil and Gitcoin strewards are playing a ever escalting evolution games -- only the smartest sybil survives, as a result, they gets smarter and smarter.

Yet the goal of quadratic funding is not to generate smarter and smater sybil and sybil detector, but to promote more public good funding, and more invovement in public good funding. If the game continues, new, inexperience real users can be crowded out by ever increasing screening and measures, while the reasoned sybil attacker stays and vampires on the system. **What we need is a mechanism to encourage lasting, authentic contribution, while filter out short-term thinking, self-dealing attackers.**

## A new Optimality Gap Design
I propose a new mechanism design to improve transparency of the matching fund result and encourage users to build long-term reputation through their contribution pattern.

The Design is Simple. Each user have two major index, their own money spend on projects and the money their contribution attracts. The New Optimality Gap measures the ratio of money spend (a cost of the user) and matching fund attacts (a cost of Gitcoin ecosystem and its grand match makers).

On a user's page, there are two index:
- “Life time optimality Gap”: the ratio of personal money vs matching fund money of this user’s contribution of a lifetime.
- Formulated alternatively as $\frac{\sum c}{\sum m}$, where $\sum c$ is the sum over all contributions for a given user, and $\sum m$ is the sum over all contribution-generated CLR match.
- “Project specific Optimality Gap”: as list of project of which the ratio of personal money v matching fund money attracted of a specific grant this user is attracted to

On the grant page, there are also two index:
- “Life time optimality Gap”: the ratio of money raised from individual users v the money raised from matching fund

- “User Specific Optimality Gap”: a list of users of which the ratio of money spend by user v matching money attracted by the user

With simple color-coded visualization. It will be obvious to any Gitcoin participants whetehr a project is attracting significantly more money than they spend, and which user is causing such pattern.

## Sybil Catching for Everyone

This idesign convert the incenive into “infiniate game” of donor reputation building from a “finite game” of extract as much money as possible from Gitcoin’s matching grands and ecosystems in one or multiple rounds.

Moreover, the optimality gap design present easy-to-understand matrix to every participant of Gitcoin eco-system without extenuous technical background in graph theory or machine learning. **Transparency is meaningless without ease of understanding.** Instead of presenting the public with a one-off, abstract measures that flag users, the "who gets how much money and why" shall be intuitive.

Such transparent KPI shall be publically displayed on the user’s personal page for everyone who have access to Gitcoin’s website and platform to view. Thus the extended finite game of self-dealing can easily present itself to a huge crowd of participants. While experts and stewards can still help the process, the respinsbility no longer falls solely on them as the information can be open without the risk of sybil attacker's mata-gaming.

## GitCoin as the reputation layer of Etherum Ecosystem

The new optimality gap's mechaism design encourge long-term reputation buiding. This coupled with aggregated effort in more traditional sybil resistence effort such as POAP and BrightID, enables Gitcoin becaming a new reputation/identity layer of Etherum Ecosystem.

This enables a new layer of the infinite game, contribution to public good projects are a reputation gain. Behavior economics theory sugested that people who contribute to charities are motivated by different inventives: some seeks the warm feeling of they helped others, others wants to be seen as moral. Early and meaningful contribution can be a strong signal of one's familarity of the ecosystems and taste of various projects and thus a signal of legitamacy. This new design enables Gitcoin as the repuation layer of the incentive systms can further mobilize the second crowd and also help resolve the problem of sustained long-term funding to public good proejected raised by Vitalik in [Gitcoin round 6 retro](https://vitalik.ca/general/2020/07/22/round6.html) post.

- Since all the matrix and donation pattern is completeley open and relatively easy to build (The KPI can be updated after the round ends, which also give the team relaxing time) and change the game dynamic, colluders of grants and sybils will be exposed