# Gitcoin Hacking Sessions Agenda

###### tags: `gitcoin` `Notes`

:::info
Updated by March 2021
:::

*Authors: Danilo Lessa Bernardineli*

## WS #8 Gitcoin Under Attack :volcano: + Scientifically approaching a Conjecture

MoH: @danlessa

**This working session is going to exceptionally have a 3hr duration instead of 2hr**

### Agenda

- Experimentally test the maximum matching subsidy conjecture by :octopus: (45min)
- (10min): explanation of some concepts
- A / B / C test for the partial Rounds 9 data when comparing: (90min)
- QF + pairwise / vanilla QF / QF + pairwise + sybil score threshold
- Proposed steps:
1. Visualize the current data
- Sybil score per grant
- Conjuctured Optimality Gap
3. Identify or inject a attack vector
- Many to one
-
5. Define a sucess metric: Percentage of captured funding
6. Scenario 1: How it performs on QF + pairwise
7. Scenario 2: How it performs on Vanilla QF?
8. Scenario 3: How it performs on QF + pairwise + sybil score >= 0

### Notes

- Overview of the attack
- AirDrops -> bots giving for everybody
- What if they're skewing the intended recipients?
- Same grants receiving the same amounts regularly, mostly from new accounts
- a
- Some attackers: mix of suspicious and legit grants
- a

### References

- :octopus: outline for the conjecture / experiment: https://discordapp.com/channels/701149241107808327/799718483537297409/821600886987816981
- Partial Rounds 9 data: https://github.com/gitcoinco/gitcoin_cadcad_model/tree/main/data/2020-03-18
- :octopus: pre-hackaton: https://colab.research.google.com/drive/1F-EJTSu4mI__X9oYAxrv6xt7_75TadYT

## WS #7 Attack Vector Design and Testing

MoH: @Jiajia20

### Goals
- quickly explain how QF and pair-wise funding works
- implement a few premeditated collusion scenarios
- Visualize collusion pattern with NetworkX and plug them in cadCAD and see the result.

### Agenda
- quick revision of QF, and example of how to game it (10-15 min)
- Implementing simple example on gaming QF and visualization (15 min)
- Test simple example on cadCAD (15 min)
- Community design collusion pattern and test against cadCAD (30 min)

### Useful resources:
- [Master Doc for everything](https://www.notion.so/Gitcoin-Modelling-Co-Lab-TL-DR-Public-4e115525dbc0430289bb78d60ad79a04)
- [Presentation slides for QF and attack vector](https://drive.google.com/file/d/1eK6Id7i7TCn4EltOWIpHFw13Mr-pQ-pI/view?usp=sharing)
- [GitHub Repo with simple ABM and cadCAD simulation](https://github.com/jiajia20/GitCoin_attack)

Repo currently offline

## WS #6: Einstein Summation / Unit Testing / Profiling

MoH: @danlessa

### Goals

- Make sense of what the anonymous benefactor code is doing
- Create a quick way to benchmark QF implementations
- Create a quick way to test QF implementation to make sure that they work in the same way

### References

- Anonymous Benefactor Code: https://colab.research.google.com/drive/1-rqewX6NkX0TMEK-W2VZEzWa_rlT-Cfx#scrollTo=EQUUoJN0ylAP
- Pairwise proposal: https://ethresear.ch/t/pairwise-coordination-subsidies-a-new-quadratic-funding-design/5553/9
- Einsum explanation: https://stackoverflow.com/questions/26089893/understanding-numpys-einsum
- typing library https://docs.python.org/3/library/typing.html
- MyPy: static typing checker by Guido. http://mypy-lang.org/

## WS #5: Making Quadratic Funding Faster

MoH: @danlessa

### Goals

- Have a iterative version of QF
- Run the cadCAD model for all the time sequence.

### Agenda

-

### References

- Rough math spec: https://hackmd.io/7HYPPp7zQcShQLrKQ2KOxQ

## WS #4: Transitioning between finished and upcoming research scopes

MoH: @danlessa

### Goals

- TO update on the status of the BSci <> Gitcoin research
- The plan for Optimality Gap investigation has been delivered
- We have a upcoming plan for making QF really fast
- Collect suggestions and proposals for the WS dynamics
- To map the current research landscape in terms of opportunities and relationships through through Kumu

### Agenda

- Intro (10min)
- Results of the Optimality Gap research iteration (15min)
- What's upcoming: making QF really fast (10min)
- Reviewing the Working Session (15min)
- Introducing Kumu (15min)
- Mapping the current research landscape (50min)

### References

- Research Landscape: https://kumu.io/danlessa/gitcoin-research-landscape
- Kumu: https://kumu.io/
- Results of the 2nd iteration execution: https://hackmd.io/st_-z9tlTwWsUyh94bg5Ag
- Research plan of the 2nd iteration: https://hackmd.io/QCCJWZE0Ru27X_GRk6UKjQ?vie
- Scaffold of the 3rd plan: https://hackmd.io/@bsci-gitcoin/ry9fxz3Z_
- G-Drive for the TE RGs: https://drive.google.com/drive/folders/1mrwTN4qiJoeGspA2l0699NkdJYUy77hF

## WS #3: Impact of Attack Vectors on Optimality Gap through A/B testing

MoH: @danlessa

### Goals

- We know how much the optimality gap of known attack vectors are different from other grants as a whole.
- We know how much the attack vectors matching compares to the other grants as whole

### Agenda

- Intro (10min)
- Updates from last week (10min)
- Injecting a arbitrary subgraph into the contributions graph (10min)
- Optimizer improvements
- Optimality Gap
- Injecting the 1st vector (10min)
- Optimality Gap for the 1st attack vector (10min)
- Match extraction for the 1st attack vector (20min)
- Injecting the 2nd vector (5min)
- Optimality Gap for the 2nd attack vector (5min)
- Match extraction for the 2nd attack vector (5min)

### References

- Jiaja's notebook laying-off examples of attack vectors: https://github.com/jiajia20/GitCoin_attack/blob/main/attack_vector.ipynb

## WS #2: Implementing Edge Rewiring Optimizer (04feb2021)

MoH: @danlessa

### Goals

- Get the optimality gap for each grant.
- Build edge rewiring optimizers on top of NetworkX and do Pull Requests for them.
- Compare the results when using Monte Carlo optimization vs simulated annealing

### Agenda

- Intro (5m)
- We need optimizers for the edge rewiring problem (15m)
- Forking NetworkX repo (10m)
- Implementing the Monte Carlo optimizer (15m)
- Implementing the Simulated Annealing optimizer (15m)
- Submit PR (10m)
- Get results for the optimality gap through MC (15m)
- Get results for the optimality gap through SA (5m)
- Compare results (15m)
- Discussions (15m)

### Links

- 2nd BSci-Gitcoin research iteration plan: https://hackmd.io/QCCJWZE0Ru27X_GRk6UKjQ?view
- Notion: https://www.notion.so/blockscience/Gitcoin-Modelling-Co-Lab-TL-DR-4e115525dbc0430289bb78d60ad79a04
- Reference algorithm: page 32 of https://dr.library.brocku.ca/handle/10464/14776
- Gitcoin model repo (debug branch): https://github.com/BlockScience/gitcoin/tree/debug
- NetworkX fork: https://github.com/danlessa/networkx/tree/sa_optimizer

### References

A Hybrid Approach to Network Robustness Optimization using Edge Rewiring and Edge Addition: https://dr.library.brocku.ca/handle/10464/14776

A novel edge rewiring strategy for tuning structural properties in networks:
https://www.sciencedirect.com/science/article/pii/S0950705119301698

Optimizing network robustness by edge rewiring: a general framework: https://link.springer.com/article/10.1007/s10618-015-0447-5
(DOI: 10.1007/s10618-015-0447-5)

NetworkX - Python package for the creation, manipulation, and study of the structure, dynamics, and functions of complex networks
https://networkx.org/

## WS #1: Implementing the Research Plan Definitions

MoH: @danlessa