Mitigating Governance Sybil Attacks

tomatosoup

Hi all,

Just wanted to start a discussion about how we can mitigate sybil attacks on our governance as we grow. As you may know, we currently use Quadratic Voting on our snapshot governance votes. This means that each wallet’s voting power is determined by the square root of their sLOBI balance instead of their sLOBI balance itself. This is meant to combat large holders calling all the shots while normal participants feel their vote doesn’t matter. In this mechanism, if most people agree on a certain option that option may win the vote even if they don’t control most of the sLOBI used to vote.

This is great. However, this can be gamed quite easily. For example, consider a hostile actor who holds x sLOBI. If they hold it in one wallet, they have voting power x^0.5. If they hold it in four wallets, they have voting power 4*(x/4)^0.5 = 2*x^0.5. With sixteen wallets they have 4*x^0.5 and so on. Since we’re on a blockchain, it can be hard to distinguish between sixteen legitimate holders and sixteen wallets controlled by a hostile actor.

In view of this I would like to discuss some potential mitigation strategies. Some that come to mind are:

Ditch quadratic voting for normal voting but this brings back those old problems.
If x% of the sLOBI is in favour of an option, it wins. (For example, take x = 90. Then if 90% of the sLOBI vote for Option 1 but many small accounts make Option 2 win in the Quadratic Voting, Option 1 will still be considered to have won.)
If the difference between normal voting and quadratic voting is too big, go back to normal voting.
etc.

What do you think?

mrthedude

Option 1 should not be considered in my opinion, the whale voting risk is too high and it will discourage boarder community participation at the same time. Option 2 seems like it is just option 1 with more steps unless “x” is set for an extremely high percentage (99% for example), if “x” is not set very high then whale voting power can just exceed that value. Are there any examples of quadratic voting being gamified in the way you described? I’d love to optimize the voting system but option 2 does not seem like an improvement to me

tomatosoup

mrthedude Not sure about examples but it is well known that this is an attack vector. As Cthulhu mentions, a similar problem is at play with airdrop farming.

Cthulhu

By quickly glancing at staked wallet size, and getting the impression it looks like a pareto, I think quadratic voting is essential to avoid significant centralization of lobis voting power.

Could we possibly look into methods of tracing connected wallets? How do other protocols do it when they want to avoid sybil airdrop farming?

For example, there aren’t that many staked lobi holders at the moment, so it should be relatively simple to see if I sell to eth, transfer to one of my other wallets, and then buy and stake lobi, right?

tomatosoup

Cthulhu Yes that is an option we could look into though it would not prevent a sophisticated attack.

tomatosoup

Some other things that could mitigate to some extent:

Voting power increased based on participation metrics
Increase voter participation somehow (I’ve advocated for a delegate system)

mrthedude

The delegation system might be a good alternative that still incorporates quadratic voting while encouraging community engagement through their free choice on candidates to pick based on social reputation. Was the delegate system mentioned above ever analyzed on how it might be implemented ?

tomatosoup

mrthedude Here’s a link to my original proposal. https://forum.lobis.finance/d/38-governance-of-lobis/23

plutusIIIIII

I seem to recall seeing a post recently about devising a method to award greater voting power for members who are demonstrably more invested in the protocol from the standpoint of time and/or engagement. For example, a wallet with just 10 LOBI, but that has been holding LOBI since launch, might be given the same voting power as a wallet holding 100 LOBI that’s only been active for a few days. Of course, the “devil is in the details” as they say, with how to actually implement such an idea, but it definitely strikes me as one with merit.