Proof-of-Deposit – initial proposal and request for feedback

I think you have a pretty good grasp of it (although it is not a “vote” that has weight in governance, but rather a “vote” by depositors deciding which validator to associate their money with). This is what makes it a market mechanism.

Which exact part of the block rewards should be shared is really just a minor detail. At its core, PoD is a mechanism which allows a market to dynamically adjust how block rewards are distributed to strike an efficient balance between incentive to adopt cStables and incentive to stake CELO to contribute to security.

That is to say, we see no issues with the 65k per validator rewards being excluded from PoD’s redistribution (although the 65k should eventually be phased out).

All your other points are circling around the same concern of centralisation, so I will address them in one go.

What does it mean for one system to be more decentralised than another? According to Frederick Hayek’s The use of knowledge in society which originally introduced the term, decentralisation is a degree to which participants can find a competitive advantage with their “knowledge of time and place”.

PoD will increase Celo’s chances of staying decentralised in the long-term as PoD makes a validator’s bottom line dependent on them attracting cStables (i.e money, rather than equity-like volatile assets). This is key because understanding the flows of money in communities and how to offer services to attract it is a skilled art, requiring specialised knowledge that changes from community to community and across time.

Does this mean that validators will not merely be infrastructure providers in the long-term, and will need to differentiate themselves by being more integrated into communities? Yes. And that is a great thing. Because unlike the current system where your influence is basically set in stone by the size of your wallet and how early you bought your CELO, with PoD you have a much better chance of changing the status quo as you can be earning a higher rate of rewards than even the biggest VCs with your specialised knowledge of local communities.

1 Like

What’s to prevent this from happening:

  1. I make 20 validator groups
  2. I give away 100% of rewards to stakers and eat the infra costs myself
  3. I control the entire network by virtue of owning all validators
  4. I implement systematic sandwich attacks and MEV
  5. I rebate some of the MEV to stakers to further monopolize the network, so stakers get >100% of theoretical rewards without MEV
1 Like

Because if you give 100% to CELO stakers and 0% to cStables, no cStable depositors will associate their deposit with you.

As your rewards is determined by min(%CELO,%cUSD,…), even though you are advertising a 100% rate to stakers, the actual amount of rewards you have to share with them is zero (and they will go vote for someone else)

1 Like

So I give a higher than market return to both celo and cUSD stakers.

My point still stands that if validators control the reward rate, there will be a competition to drive up the rates and someone will start extracting and rebating MEV

1 Like

You don’t want to make validator selection a contest of interest rates.

Once you make this into a interest rate game, the smartest player will win and dominate the space.

This is what happens when it’s a “competitive market”, the most competitive player wins the pie, and there’s very few rules in place to prevent them from MEV

1 Like

A validator won’t be able to offer higher than market rate on both, only one (rates have to add up to 100%).

The only way they can offer above market rate on both is using other methods outside of PoD. This is no different from the current system?

If you set up 20 validator groups now, and bribe stakers sufficiently, you will be executing the exact same attack?

Your post didn’t say it has to sum to 100%, just that it was using min() of rates.

Ok sure, thanks for the feedback. I will state that explicitly in the formal version.

I agree your concern is an issue if they can set their rates to sum over 100%.

1 Like

Please do, thanks.

As for why this doesn’t happen today with side channel payments, many validators have their votes from the foundation and it’s in bad taste to MEV while holding onto foundation votes.

The whale validator groups from the VCs are also sitting on so much gold that there’s no need to get creative.

On other networks such as solana that have order of magnitude more validators, the cost of such an attack is too high.

Ultimately having 10x more validators in the active validator set will decentralize and solve for such concentration problem.

2 Likes

Actually capping it to 100% of block reward doesn’t solve the problem because validator performing MEV can arbitrarily boost block reward for their own blocks via dummy tx with insane gas.

This will still allow validators to rebate back >100% of normal block rewards.

1 Like

Why do the offered PoD rates have to sum to 100%?

This seems somewhat arbitrary especially in the future if there many, many more cStables. Your average interest rate will have to drop each time a new cStable comes online if you want to keep a balanced offering.

If I want to offer 40% interest on 7 different assets for a promotional period to bootstrap a new validator group, why can’t I? (Capped to the amount I actually receive in rewards, of course).

What of the side-effect that basically makes the concept of validator groups null and void? No one needs to run multiple validators anymore, since you don’t get any financial benefit from it - you only benefit by capturing a larger portion of locked / deposited CELO and cStables.

Or is the plan to keep the validator group structure as is, so that multiple validators are required to keep your vote capacity high so you can indeed capture the balanced CELO/cStables “votes” you need?

You can delete posts if you make mistakes etc.

I think they are adding the 100% constraint in response to my hypothetical of a group of validators performing MEV and juicing the rewards to above 100%.

Once you can go above 100%, you can subsidize the interest rate to gain more votes and controls more of the network.

I also point out that even with the 100% of block reward cap in place, a validator can juice up their own block with dummy tx at insane gas price to rebate back to their voters.

Either way, POD creates a new game for validators to play and compete in, and allows aggressive players to take over the network with above market rates

@Thylacine I apologise for straying into confusing terminology, this should clarify things for you.

The parameter being set is the proportion of block rewards being shared with cUSD depositors (an additional parameter is required with each additional cStable). You are not directly setting the interest rate on cStables.

The proportions have to add up to 100%. We did not explicitly state this because we felt it was implied.

The reasoning is straightforward and not limited to PoD: there are always external incentives that can potentially be greater than the internal incentives of your system (e.g. some billionaire or government can afford to continuously bribe all stakers substantial $$ for years). What you don’t want is a way for external incentives to directly become internal, as that makes it all too easy for a hostile takeover to occur. Rather you want to force the external incentives to face a huge coordination problem by having to find, contact, and convince/bribe the majority of stakers (which should hopefully be as decentralised/diverse as possible).

Hardly. I was being polite by not saying that the implied 100% should have been obvious.

Providing feedback is fine, but if you’re going to make huge claims like this, please back it up with actual detail.

I already laid out the recipe a few posts ago.

  1. Artificially pump the block rewards for your own blocks to increase the returns to above market.
  2. Subsidize it by playing MEV on your blocks
  3. Get more and more validators elected by artificially inflating your interest rate
  4. Take over 1/3 of elected validators
  5. Take a massive short position on Celo
  6. Shutdown 1/3 of validators forever

Instead of trying to argue about whether this is an attack vector that can halt the network or not, why not spend more brain cycles on coming up with an improvement that doesn’t involve validators twiddling reward rates and putting validators in direct competition like this?

Again, as said, this is not an attack vector specific to PoD. You can just as easily just directly pay all your voters wherever you are selected to create a block.

You should be raising this with the protocol team to upgrade from pBFT which puts the constraint on the number of validators.

In regards to your comment on competition… that’s literally what keeps a system decentralised? Or are you suggesting a centralised monopoly is what we should be aiming for.

If we’re not actively innovating to expand the surface area to allow small validators to compete favourably against validators run by whales, what do you think will happen when yearly revenues from being a validator reaches billions or higher? The current “competition” over the cheapest cloud provider and who can bulk buy CELO at the biggest discount is hardly sufficient from preventing a situation akin to custodian banks (of which the top 5 control well over 50% of global AUM)

1 Like

POD makes it possible to do this kind of rebates in a way that is fully baked into the protocol, so a lot of users who would use Valora to vote on where to stake will just haphazardly pick the one that has the highest actual earnings.

Celo’s pBFT for better or worse has its current limitations, if you propose POD on another network such as Solana or Avalanche with 1000+ validators, then there’s less of a concern, but you can’t simply punt your proposal’s attack surface over to the protocol team. It’s your proposal that’s introducing this vulnerability.

I’m 100% for competition, but not when the game being invented has a flaw that allows a clever player to ruin the party. My criticism is not on competition, but the type of competitive game you’re inventing for validators, which is not fun imo.

The current validator rewards is a capped $75k / year cUSD. Being a fixed reward means validators are more like a utility and get paid to run a neutral network infrastructure.

Even when Celo network 100x growth, validators will not earn any more than the fixed amount currently paid, so no validator will ever have yearly revenue in the billions.