Wenjia Ha - Data Analysis of Contributions for Round 9
Background
When round 9 attack happened she did some of the data analysis on contributions only using contribution graphs.
She looked at: Repeating, cancelled, amounts, patterns, & relations
Consider
A few people made extremely large amounts of contributions including 1 user with 1,000 donations. Half of these went to Etherdrops.
Takeaways
Contribution data is good for community gov because it is on site publicly anyway! We should make it easier for the community to get this data.
Half of bad contributer donations went to gitcoin grant because they were trying to establish legitimacy.
How much do they earn and how much do they lose? We need to understand which are the profitable patterns.
If you catch a suspicious grant & contributor, then you can see which other grants are probably bad. Same with contributors. This is due to the nature of the graphs.
When we catch them, they need to contribute to more legit grants thus raising the amount they need to give to legit grants.
Cost of gas + Amount donated to legit grants MUST BE HIGHER THAN the amount of matching attracted to bad grants.
The colluders are getting more sofisticated
Using rollups wasn't strategy in last time
We are always responding to last round. They start new behaviors, then we see what they are and respond. When we come up with sofisticatedsolutions, they will create more sofisticated attacks
Airdrop money to github users to get extra account
The cost efficiency strategy is not a detection strategy, it is an incentive engineering for
Better control if the token was:
Compare to the graph, incentivize people to catch colluders - and maybe slash colluder funds
Wouldn't want
Need to optimize for active ejgaged informed participation.
We should check out meta-gov,
Background is social computing Ph.D.
Action Items