[BIP-57] Introduce Gauge Framework v1

Governance General Proposal
81

copy the spreadsheet and you can do what you want. it will not be updated on an ongoing basis.

Every change of the cap requires a tx from the gov multisig. Additionally, there are implications for unclaimed rewards with cap adjustments. i.e. if a gauge has a 2% cap but gets 3% of the vote then next week cap is raised to 3%, people can now claim rewards from the past week as if the gauge did actually get 3% of emissions. Basically we cannot really do a huge amount of cap adjusting or we will run into issues/ways to game the system.

General update - Badger has signaled they cannot vote on Balancer’s snapshot after the update to vlAURA’s strategy. We will wait until they have corrected this. They’ll provide an update but there is no firm ETA at this time. Cheers.

1 Like
82

We will try to get this done as quickly as possible. This change to governance practices requires us to make a simple change to the graviAURA strategy to support new balancer delegation. We have a timelock and a review process for strategy changes that usually takes around 5 business days. We are working to the change proposed and start the review process with haste.

I’ll provide more details as to when we expect this to be complete as soon as they are available. I don’t see why this wouldn’t be in place by next Thursday(August 1), and would even understand if balancer carried on with governance if it were not, but asking 2 weeks to adapt to a technical change in governance function seems reasonable.

83

Effectively, yes.
The emissions are spent by veBAL and vlAURA holders, but the incentive to vote comes from their earnings.
The fact that these bribes are cost-inefficient, i.e. more money is spent on bribes compared to the amount earned from the pools, makes them net negative.
As a veBAL or auraBAL holder I would receive more yield without the core pool bribing than with it.

Let me dive deeper into the calculations, and maybe you can point to where I’m wrong.
Let’s take a look at the bribes posted between August 4 and August 17 and estimate their cost-efficiency for the DAO.

From what I can gather from the gauge-vote-incentives channel on the Balancer Discord and llama.airforce, during that period there were around 130 $k USDC posted as core bribes for veBAL holders and 49.7 $k USDC for vlAURA holders.

WBTC/WETH/USDC (A)

The first example is WBTC/WETH/USDC pool on Arbitrum. 27.8 $k USDC were spent on bribes. What were the results revenue-wise?
The pool earned 12.6 $k in fees over the following week. Let’s assume it earns the same next week as these are two-week bribing cycles.
Afaik 50% of the fees go to LPs and 50% goes to the DAO, thus the total number is 12.6 $k in revenue for 27.8 $k in expenses should sound about right, correct?

So, the pool earns 25% of the revenue to the treasury (= 3.15 $k), and the rest is recycled into more cost-inefficient bribes, while veBAL and vlAURA holders earn nothing from all of the trading activity on Arbitrum. Again, I ask you to correct me if I’m wrong as I might miss some arrangements.

wstETH/ETH

The second example is the wstETH/ETH pool on mainnet.
Afaik it received 103.25 $k core bribes, 78.8 $k to veBAL and 27.5 $k to vlAURA holders.
On top of that, it got 255.5 $k worth of veBAL bribes in LDO over the two week period.

So the core bribe share was 28.8%. For simplicity, let’s assume that this is Balancer’s share in wstETH emissions and revenue. All bribes bought around 17% of BAL emissions.
Let’s also assume that the grand plan of farming 50% of stETH staking yield is already live.
What do we end up with?

  1. The pool earned 9.3 $k in fees over the following week, which we can extrapolate as the protocol revenue from the fees.
  2. The staking rewards share. Lido shows 3.9 % APR in staking rewards, the pool has 224 $M TVL and has 71.5% stETH in it.
    160 $M * 3.9% * 50% = 3.12 $M that the protocol would earn from the pool in a year total.
    28.8% of that is 900 $k.
    900 $k a year means 34.5 $k every two weeks.

Thus, the protocol spent 103.25 $k on bribes that result in 43.8 $k revenue. 25% of that (~11$k) would go to the treasury, and the rest would be recycled into more cost-inefficient bribes.

Do these calculations accurately represent that the protocol loses money on core bribes, or am I missing something? Wouldn’t veBAL holders be better off if the core bribes would just be distributed to them directly?

During the same period, Badger pool bought 4.47% of emissions by spending close to 80 $k on bribes, which annualized means close to 2 $M of yearly revenue for veBAL and vlAURA holders.
The competitive bribe market allows the bribe revenue to scale and new bribers to enter the competition, increasing the emissions cost-efficiency for the DAO.

There is not much for me to say though if we can’t agree that green is green, spending what otherwise would be veBAL revenue is a cost, and getting higher yield on veBAL and vlAURA is good for the ecosystem and leads to emissions not only becoming more cost-efficient, but more valuable too.

84

The core pool bribes are distributed directly via voting. I’m still quite confused on that point to be honest. If core pool bribes were distributed passively as before all mercenary voters would see a large reduction in their earnings compared to today. There is simply not enough 3rd party bribers to cover the void we’d leave behind.

Saying Badger bribes are revenue but core pool bribes are an expense doesn’t make sense to me. I could definitely not be thinking about it correctly though. If we embrace your thinking won’t that lead to very little protocol revenue from fees as emissions are concentrated into pools where the only revenue is bribing? Does this seem healthy to you? Most people value protocols like Balancer on protocol revenue earned from fees. I struggle to see how this strategy leads to value creation for the protocol.

1 Like
85

This is not the first time I have seen this argument in this thread, and tbh I don’t get where it comes from or why people believe it to be true.
As I mentioned earlier, there is no indication that 3rd party incentivized bribe market lacks longevity.
The Convex bribe market has been up for less than a year, and starting around December/January it has, on average, been way below $2 emissions for $1 in bribes, sitting around $1 for months in the first half of the year.

More to it, if we imagine the scenario where one day somehow all the bribers leave the market, what would be the issue?

Old bribers leaving the market means an opportunity for new bribers to enter it. It’s a free market, and different actors are ready to pay for emissions at different rates.
On Curve/Convex there have been a couple of actors ready to pay extremely high prices for emissions, effectively monopolizing the market and turning other participants away. But it’s okay, it’s how free markets work. At the end of the day, it’s the lockers who win from the market competitiveness.

Curve has basically sold its inflation to other protocols that are willing to buy it at a discount solving the inflation issue. /Ouroboros/

Let’s go further and imagine the worst case scenario where all the bribers have left the market, so there is no bribe market at all.
Well, then we’re back at square one, where:
a) lockers vote for the markets they wish to support
b) lockers that don’t have the preference in what market they wish to support have the incentive to vote for the pools that they expect the protocol to earn the maximum amount of revenue from that is distributed back to them.

In that scenario, the DEX would come back to Curve 1.0 voting & emissions, which effectively means that emissions are much more inflationary and less cost-efficient compared to the ecosystem with a developed bribe market.

I don’t distinguish between what’s best for the protocol and what’s best for vlAURA and veBAL holders, I consider the two to be naturally aligned.
The lockers have natural incentive in optimizing for the higher yield, it’s just doing it through what you tend to call “protocol revenue” is suboptimal. If somehow one day bribe market dies, voting for the pools that you consider “good” would be the only option for them.
But in that case, the ecosystem would arguably be in a worse state.

With the way this proposal approaches the bribe market (as if something that accounts for 90% of lockers revenue doesn’t exist), what would otherwise be an unlikely scenario that no one wins from becomes much more likely. And for what, what is the goal?

The only goal I’ve found a clear incentive for is trying to increase the treasury’s Stablecoin balance sheet. But as I mentioned earlier, there are multiple other ways to do that which don’t affect the lockers this much.

86

Can you explain how you got to a market cap of $217MM (or even $239MM in another row…) for $auraBAL? With a supply of roughly 2MM, at a price of say $17, that should be more something like $34MM…? Similarly $LDO is in your sheet for a market cap of $1388MM, whereas CoinMarketCap only reports $654MM? You ascribe a market cap of $809MM to $WETH, but Etherscan reports a supply of about 4MM, coming to a market cap of >$6000MM?

Also I cannot help but conclude that this proposal basically denies an uncapped gauge to any pool that is not of the 50/50 kind. In my opinion, one of Balancer’s value propositions is its custom pool compositions such as the 80/20, which doesn’t really get taken seriously here. Doesn’t this force the whole gauge system into basically the same system that Curve has been running for years now? What new value does Balancer now add?

87

Tokens which are freely minted are assumed to have the market cap of the token that mints them. aurBAL can be minted with BAL, thus it gets BAL’s market cap. etc. Lido’s market cap on coingecko matches the spreadsheet.

Large cap tokens can still secure 80/20 uncapped gauges. Lido being a prime example of a great one for us. Small cap 80/20 pools have proven to generate near zero revenue, I can show many examples of this. Balancer will slowly die if we continue to allow these and similar pools to have a large amount of emissions directed to them.

2 Likes
88

I consider the situation where the absolute majority of the revenue comes from bribing healthy as long as the emissions cost-efficiency overall is high enough.

Among the current pools, even the one that would produce the most fee+staking flow - stETH - would still have most of its value coming from bribes.

The ratio of fee & staking revenue to bribe revenue would be different for different pools.
It is also reasonable to assume that the pools with gauges that would have the fee revenue as a dominant factor would have less cost-efficient emissions.

Balancer used to distribute BAL rewards through the Committee for a long time, right?
I’m curious to know what ballpark of cost-efficiency the emissions / protocol revenue equation was back then.

How many $ in emissions were spent on $1 in revenue back then? 3? 5? 7?

In my valuation framework revenue earned from the fees is a factor, but I would probably give it around 5%-10% weight.
The fee revenue is a minor demand factor in a rational actor’s equation of choosing to lock the ecosystem token or not.

It becomes more apparent if we isolate the fee revenue, make it the only factor, and see if the thing would work or not.

Would you be interested to buy and effectively “burn” CRV to get 7% APR in Stablecoins on a volatile token? Or would you want to get into CRV because of something else?

89

So I seem to have misunderstood and a colleague pointed it out. Under this governance. graviAURA would capped both by a little bit for being in a 40/40/20 and by a lot for being 20% in a pool. As a result, we estimate AURA would at least a 120 million dollar market cap before graviAURA pools could grow over 2% voting.

Do I understand correctly?

90

roughly yes, though remember revenue factor can help exceed the threshold as well. it is not solely driven by market cap.

1 Like
91

Ok. Then I don’t see anyway I can stand behind this proposal as it stands. Are you open to rethinking it more or does this just go to vote?

I just want graviAURA to be able to not be globally capped at 2% for a very long time, as well as BADGER.

10% Seems right to me. 5% seemed a reasonable compromise, but this doesn’t work.

1 Like
92

I think all things considered we can go to 5% instead of 2% if that means you can support it. 10% is not realistic - we may as well not do this framework at all and save ourselves the hassle.

5 Likes
93

Followed this discussion from the sidelines.

I like 10%

2 Likes
94

5% is a good starting point, can always revaluate later if its not sufficient

95

@TheOne would you agree to this framework if the cap for low mcap pools <50 million mcap was 10% of veBAL, including the current weighting of token mcap based on the pool type and their pool weight?

IMO the second multiplier shouldn’t be applied. So it’s ok to take Fernando’s efficiency number, but then not multiply it by how much of the pool it is, however at 10% it’s workable either way.

I think this would be a step in a good direction. I think if there were consensus between major holders and delegates it is something that could pass.

Maybe if everyone can just comply, we don;'t need to add all this complexity to governance?

96

It is a workable middle ground should there be no kill aim against our small cap investment, we request reciprocity.

97

Here are the things the community agrees on:

  • The framework is necessary to standardize our evaluation in adding, adjusting and rejecting gauges
  • High revenue generating pools will be uncapped
  • Some form of cap on small pools

Disagreements:

  • % cap on smaller pools - swinging between 2, 5, and 10% cap

I suggest we move ahead using those three options on snapshot. The community has signaled that they prefer one of these three options under @Tritium post 2 days ago.

“Implement the framework using a x% cap on small-cap pools”

I prefer 2% but am opting for 5% as the compromise and believe this is the best option to appeal to both sides. This is only a start, and we can reflect on the % cap in the future in a few months.

1 Like
98

I think it’s worth noting that launching a poll with 3 options would highly favour those voters who have concentrated voting power and can coordinate much easier.

The 10% option should not even be in the question, it’s clearly not a “workable middle ground”, but barely different than what is currently out there, as a malicious party could just concentrate/cycle around a small number of gauges.

7 Likes
99

So I noticed that I heavily misunderstood the math, so I played around with the spreadsheet a bit more.
One thing I am wondering now, is a 33/33/33 pool worse for balancer than a 50/50 pool or even multiple 50/50 pools?

I am asking because if let’s say we have token A and token B that are right on the edge and have a mcap of 50M each. LP with tokenA/tokenB/ETH would be capped but each of A/B, A/ETH and B/ETH would be uncapped.

I guess multiplying with the weighting% confuses me in general. If we have a smallish mcap token, why is a 80%Token/20%ETH Pool better than a 20%ETH/80%Token? Shouldn’t they be equally bad?

100

I don’t think the 10% option is viable either, but I was just going, based on the poll tbh. If sentiment has changed and there isn’t much support for 10% then it should be ignored. Going forward with 2% and 5% is fine.