January 27, 2023: Token Emissions Debate
Benny vs Defi Made Here
Our first in a potentially recurring series of debates on the big topics of the day. We wanted to get two divergent takes on the subject of token emissions. The conversation ended up going into a million directions, but nonetheless interesting and worth publishing. Enjoy!
Welcome to our first debate, on the subject of protocol token emissions within DeFi. Emissions have generated a lot of controversy: What is the proper use of token emissions? How should one properly account for them on the balance sheet? Are they a good or a bad thing for a protocol? To explore the subject, we've invited two intelligent observers to weigh in on the subject. Asked to represent the proposition side is benny, and the opposition side is DeFi Made Here.
We'll begin with opening statements -- please give a brief introduction and lay out the general construction of your argument. We'll start with @benny
Hi, my position is that token emissions can work as a decentralized mechanism to fund the growth of a project and are therefore an essential tool for DeFi protocols. Emissions that follow an immutable schedule also provide transparency and predictability to investors.
Now that doesn't mean that emissions are a panacea or that they're always implemented properly. You can piss away emissions on single staking to try and prop up the price of your shitcoin, or you can spend it on things that will bring value to the protocol. So the real debate to me is about how emissions should be used. This will vary depending on the nature of the protocol (L1, DeFi, privacy, ...) and what one considers to be most important to its long term success (security, decentralization, market share, liquidity, innovation, visibility/branding...). Anyways, looking forward to discussing this further in depth and hearing different perspectives.
Thank you... @DeFi Made Here?
Hi. IMO token emissions and tokenomics in general are very complex topics. In fact, most of the projects do not need a token, and nothing will change if the token will be removed at one point. There are projects that depend on the native token, but in most cases, these models are unsustainable and are at the expense of liquidity providers. Token emissions might be good, bad, or neutral for a protocol, depending on how the token is integrated into the protocol. I also think that most of the protocols' tokens are a cash grab. In 2021 protocols had several billion valuations for a few lines of code and 5 active users.
Thank you. benny, you may respond, and DeFi Made Here will have the opportunity to reply as well
I think we're generally in agreement here, so it might be worth narrowing the discussion down to the tokenomics of specific protocols. Projects that raise millions on hot air are indeed a significant part of the industry, but there's not much point discussing their tokenomics since, well, it's all vaporware. I'd also agree that not all projects need a token, just like not all companies need to raise or go public. There are lots of DeFi projects that provide a valuable service, can generate enough revenue to pay for a decent dev team and have no need for a token. Yield aggregators almost invariably fall under this category. Arbi also now has a couple of small lending protocols (Sentiment, Vendor) that work fine without a token. But if you're planning to launch an L1, if you need to incentivize large amounts of capital, to compensate specific on-chain actions at scale or if you want to build an actually decentralized governance system for your dApp, you're going to need a token and you can use an emission strategy to your advantage. So it might be more productive to focus our debate on this subset of protocols and discuss those that we think are employing emissions successfully and those that are not. For instance, I'd argue that Ethereum, Curve and Chainlink are all using emissions sensibly. On the other hand, protocols like Sushi, Olympus or Alchemix offer let's say "nuanced" examples of emissions strategies.
But if you're planning to launch an L1, if you need to incentivize large amounts of capital, to compensate specific on-chain actions at scale, or if you want to build an actually decentralized governance system for your dApp, you're going to need a token and you can use an emission strategy to your advantage.
The question here is who will provide liquidity for your token and why? Normally these are VCs who have a big stake and are incentivized to keep prices high to be able to dump later.
L1 tokenomics is also a completely different topic as not all L1s are built equally. If the transactions are too cheap (eg Solana) then there is no value accrual (and economic incentive for validators etc), if the transactions are too expensive then the chain will not be able, for example, to have a solid LOB perp DEX with HFT bots / MM as they will be losing money due to high amount of transactions.
Governance is quite often a meme. Normally it is controlled by founders or 2-3 whales/entities. Look at Curve - 1 multisig controls 51% of governance power. Ethereum - 3 centralized entities have 51+% of staked ETH. And a bunch of DeFi projects where the founder decides what will be the outcome of the vote. More than that, the most decentralized DeFi protocol Liquity does not have governance.
I would also not include Olympus in the bad examples, it was quite innovative, the problem was that a lot of people misunderstood it (or simply did not care to dyor). But there are countless protocols with really bad emission strategies. However, it is often done on purpose. I spoke with the people who developed tokenomics for big DeFi projects (bad tokenomics with 2pool design) and they explained to me, that the idea is to sell % of supply to VCs in exchange for initial liquidity, get trading volume, TVL, and social media hype to get listed on the exchange. Then there is a big pump in the price, founders dump their bags and become rich.
benny, your response
I don't think you can so casually dismiss governance as a meme, at least not for the protocols and reasons you mentioned. Sure you'll find plenty of projects whose governance is just decentralization theatrics but that's more a symptom of the space's endemic vaporware problem than of decentralized governance itself. A more valid criticism imo is that it's ultimately impossible to decentralize a protocol's fundamental design decisions. But that's not specific to crypto, you can't have a democratic government without an original act of sovereign power establishing its fundamental principles (i.e. its constitution). So with governance the question, to me, is more about the decentralization of operations around the protocol's core mechanism.
One of Liquity's main feature is the the automation of operations through governance-free algorithmic monetary policy, so sure they don't need governance. You mention that they are "the most decentralized DeFi protocol" but I'm not sure they could have achieved this without their emissions schedule: LQTY token rewards were essential to bootsrapping liquidity from a large number of different LPs for the stability pool.
When it comes to Curve, the level of complexity that comes with integrating with a wide variety of tokens (as opposed to just ETH for Liquity) means that you can't fully automate operations. The complexity also means that operational decisions actually need to be decentralized, because there's simply too much information going around for a centralized entity to process before deciding. This is what's happening with Curve where community members largely independent of the core dev team run simulations and analyze protocols in-depth to provide information to voters who can weigh it against their own knowledge. Now, saying a multisig owns 51% of the decision power is only half the story. That multisig is Convex's and their vote mirrors the vote of vlCVX holders, none of whom hold more than 10% of the supply. Of course the voting process relies on a centralized solution (Snapshot) and the multisig signers' willingness to follow through with the users' votes. However, there's an economic incentive for the Convex team to be honest (otherwise it'd quickly lose its majority and its whitelist could be killed). And, more importantly, Convex's end goal is to move its governance away from Snapshot and to an L2 where voting can be done on chain. However the tech for this isn't there yet. The current situation is still based on trust (that Convex will deliver and that L2s themselves will effectively be decentralized in the future), but decentralization is a goal and the road towards it is neither short nor easy.
I'd make the same argument with regard to Ethereum, or web3 as a whole: users almost universally rely on centralized RPC providers to interact with web3 protocols, for Ethereum 3 centralized entities have the majority of the staked supply and a single client has over 80% dominance. But all of these are widely discussed, and you've got new ideas and new protocols continuously emerging to try and address those problems. The risk of decentralization-doomerism is that the space becomes apathetic to these issues ("It's all just a meme anyways"). It's good to keep an eye out for and call out centralization vectors, but the process should be constructive rather than dismissive.
This is already getting long, so I'm going to stop here. I realize I didn't address some of your points so feel free to raise them again if you'd like to debate them further.
I don't think you can so casually dismiss governance as a meme, at least not for the protocols and reasons you mentioned.
If the reasons I mentioned are not the problem then what is the problem... I am also not really sure that a governance system where more tokens = more voting power can be also considered fair. There are so many cases when 1 big token holder outweighs all other votes and you can not do anything about that. That is a problem even with relatively big protocols like Balancer.
I'm not sure they could have achieved this without their emissions schedule: LQTY token rewards were essential to bootstrapping liquidity from a large number of different LPs for the stability pool.
No doubt. Token emissions are super cool for any protocol as far as someone is eager to provide liquidity for it. Wondering who is so generous to provide liquidity for LQTY and why? It is a farm-and-dump coin with -99% performance.
Now, saying a multisig owns 51% of the decision power is only half the story. That multisig is Convex's and their vote mirrors the vote of vlCVX holders
Do you want to say that if multisig decides to vote on anything they can not do it without approval from vlCVX holders?
Another problem with governance is that most often token holders do not have enough expertise in complex financial systems to make their own decision on a particular proposal. In the best case, they will read someone's else explanation and vote based on that. Roughly saying one can not decide on Apple business decisions by holding an iPhone. For example, I am considering the MakerDAO governance forum and discussions to be really cool. But I am definitely not the right person to vote on that proposals as they are sometimes too complex. This raises the question if an average token holder should have an influence on these proposals.
To make it clear: I am pro-governance system, but I think that the current system is rather broken.
I am also not really sure that a governance system where more tokens = more voting power can be also considered fair. There are so many cases when 1 big token holder outweighs all other votes and you can not do anything about that. That is a problem even with relatively big protocols like Balancer.
True. Even if not all protocols face this issue they are all at risk of an attacker with deep pockets. But a decentralized protocol can be forked by its community (leaving the attacker controlling a protocol with no user base and a token with no value). Delegation systems can be used to add balance. Limiting the scope of governance to parameter tweaks or contract updates in the way Curve does also limits the surface of attack. There are also a number of projects working on sybil resistance through proof of personhood that would allow for fairer and more efficient voting set ups like quadratic voting.
No doubt. Token emissions are super cool for any protocol as far as someone is eager to provide liquidity for it. Wondering who is so generous to provide liquidity for LQTY and why? It is a farm-and-dump coin with -99% performance.
Well LQTY entitles holders to a share of revenue doesn't it ? So ultimately it comes down to whether people think the protocol will generate enough revenue to make their investment at the current price worth it. Whether or not blockchain adoption and the need for a decentralized stablecoin will allow for this to happen is for investors to assess.
Do you want to say that if multisig decides to vote on anything they can not do it without approval from vlCVX holders?
No to the contrary, there is currently no programmatic mechanism ensuring Convex follows the outcome of vlCVX holders. But Convex has nonetheless done so so far and is actively working on a solution that would bind the protocol's vote to vlCVX holders' on-chain votes.
Another problem with governance is that most often token holders do not have enough expertise in complex financial systems to make their own decision on a particular proposal. In the best case, they will read someone's else explanation and vote based on that. Roughly saying one can not decide on Apple business decisions by holding an iPhone. For example, I am considering the MakerDAO governance forum and discussions to be really cool. But I am definitely not the right person to vote on that proposals as they are sometimes too complex. This raises the question if an average token holder should have an influence on these proposals.
In that case you could just abstain or delegate your vote, for governance models that allow for something like liquid democracy. But even expertise is not always enough. Kahneman has a paper where he shows that there's huge variability in experts' opinion especially on complex questions. And it's much better to develop simple rules to apply to your data to come up with an answer. Incidentally I think a lot of the recent work you see coming from the DAO in Curve pushes in that direction: i.e. making pool simulations more accessible, trying to formalize LP risk criteria or improving analytics. It comes down to taking the complex questions the DAO votes on (pool parameter changes, gauge addition) and trying to apply some heuristics in order to answer them.
Another way to handle this complexity is to more directly decentralize information processing through market mechanisms, which is what bribe markets are currently doing. It's not necessarily a perfect solution, and the current iterations are also not always free markets. For instance, in the hypothetical, it'd be interesting to see how protocol competition and economic incentives align for certain types of votes.
To make it clear: I am pro-governance system, but I think that the current system is rather broken.
Of course it's going to be broken. That's what makes it exciting. DAOs have been around for less than a decade. They're an experimental space with innovations, awkwardness and failures. You could be pessimistic if the system was failing and people stopped looking for solutions and stuck to their old failing ways. But it's not what's happening.
Limiting the scope of governance to parameter tweaks or contract updates in the way Curve does also limits the surface of attack. There are also a number of projects working on sybil resistance through proof of personhood that would allow for fairer and more efficient voting set ups like quadratic voting.
Some of the governance attacks may lead to irreversible consequences (eg Beans Farm totally drained).
So ultimately it comes down to whether people think the protocol will generate enough revenue to make their investment at the current price worth it.
So if we assume that liquidity providers are the community, then emissions incentives are distributed at their expense. And this is a problem. Most of the token emissions are zero-sum PvP where liquidity providers (for the native token) are on average always in loss (in the best case they lose a little less cause they get a tiny % of revenue). I do not like that and I wish there were better designs in existence.
No to the contrary, there is currently no programmatic mechanism ensuring Convex follows the outcome of vlCVX holders.
So we can agree that Curve is centralized af?
But even expertise is not always enough. Kahneman has a paper where he shows that there's huge variability in experts' opinion especially on complex questions.
This is the thing. It is difficult to come to a certain conclusion even for the experts. How can a protocol like Maker, for example, rely on random token holders, if they deal with complex financial systems, deals with real banks, etc etc.
Some of the governance attacks may lead to irreversible consequences (eg Beans Farm totally drained).
Bean Farm was a poor design choice (no timelock) rather than a flaw of governance in general. If a DeFi contract is poorly permissioned and lets anyone withdraw funds, it's not an indictment of DeFi just bad development practices and poor due diligence.
So if we assume that liquidity providers are the community, then emissions incentives are distributed at their expense. And this is a problem. Most of the token emissions are zero-sum PvP where liquidity providers (for the native token) are on average always in loss (in the best case they lose a little less cause they get a tiny % of revenue). I do not like that and I wish there were better designs in existence.
Not all emissions need to be directed to the protocol's token. Alchemix is a good example of that issue, they drop ALCX for single staking and ETH/ALCX emissions (soon to be replaced by a 80/20 BAL I hear, but it's not much better) when the emissions would be better spent on peg maintenance. but emissions are not some big secret, even protocols with a huge diluted/circulating mcap ratio are transparent about how much will be emitted and how it will be spent. it's up to investors to decide whether the value lost on dilution will be necessary and profitable in the long run. some may also want to take advantage of a big emission schedule and its downward pressure on price to accumulate.
So we can agree that Curve is centralized af?
No, because you're disregarding that most of the protocol is still entirely automated and permissionless. If you're unhappy with Convex's vote on a pool's parameter, you can go and set up your own pool with your own params. Convex can't kill your new pool, nobody can take the LPs money or stop people from trading there. And the revenue generated by the protocol will continue to accrue to all veCRV holders (anybody can burn fees and call the contract to distribute them). We're talking here about the core functionalities of the protocol: setting up pools, trading on them and distributing the revenue they generate. Regarding governance specifically (parameter votes, gauge weight, ownership votes), I would say that there is a centralization vector, but one which has never manifested itself as such in practice (cf. my previous comments on why the multisig has an incentive to always follow through with what vlCVX voters decide) and which both Curve and Convex are actively working towards mitigating.
This is the thing. It is difficult to come to a certain conclusion even for the experts. How can a protocol like Maker, for example, rely on random token holders, if they deal with complex financial systems, deals with real banks, etc etc.
I've listed several options. Random token holders can still chose delegates they agree with and the outcome may not be less efficient than what the experts come up with. Liberal democracies are run by random vote holders but they've greatly outperformed technocratic regimes like the USSR where, at least since Khruschev, qualified experts were in charge of coming up with optimal decisions.
Bean Farm was a poor design choice (no timelock)
There is no point to focus on a single example, as there is a number of different where everything was ok with smart contracts. For example, GEV attack from DCFGOD where he accumulates gov tokens and then liquidates the treasury.
Not all emissions need to be directed to the protocol's token
Wherever emissions are directed (in most tokenomics designs) the token price depends on liquidity providers who are taking the hit for one or another reason. Being an LP = is always a bet against the market. You are buying, when everyone is selling and you are selling when everyone is buying (if we are talking about AMMs). But this is a completely different story. What I wanted to say is that token emissions for the protocol are a no-brainer, just have to find someone who will provide liquidity to exit for everyone. Or find a friendly lending protocol that will add your token and founders can borrow against the native token (as there is no way to sell so much supply at a time).
I've listed several options. Random token holders can still choose delegates
Can or can not choose, can also skip the vote, etc. How does it save the governance from random token holders having an influence on the proposals? Especially in the system when 1 token = 1 vote. Not saying that random token holders might not have enough understanding to vote in the best intentions of the protocol, it is also a big risk for every protocol, when someone can accumulate enough gov tokens to vote to rug the protocol in one or another way.
Liberal democracies are run by random vote holders but they've greatly outperformed technocratic regimes like the USSR where, at least since Khruschev, qualified experts were in charge of coming up with optimal decisions.
First of all, 1 person = 1 vote. Secondly, they select people who have enough expertise to run the country. They do not vote for the decisions of the elected experts.
> > So we can agree that Curve is centralized af?
No
Interesting enough that Curve itself admits this problem
Wherever emissions are directed (in most tokenomics designs) the token price depends on liquidity providers who are taking the hit for one or another reason. Being an LP = is always a bet against the market. You are buying, when everyone is selling and you are selling when everyone is buying (if we are talking about AMMs). But this is a completely different story. What I wanted to say is that token emissions for the protocol are a no-brainer, just have to find someone who will provide liquidity to exit for everyone. Or find a friendly lending protocol that will add your token and founders can borrow against the native token (as there is no way to sell so much supply at a time).
You make it sounds like LPs are unwitting victims waiting to get fleeced as everyone uses them as exit liquidity. But some protocols hire, and pay a high price for, a professional market maker to LP their token. Other may use POL and do it themselves. Or have their VCs agree to LP as part of an initial deal. Other LPs may run more or less sophisticated strategies, or simply willing to accept IL to accumulate more of the emission token. There's nothing criminal going on here unless the project shilling vaporware and actively deceiving investors about the prospects and value of what is being built. Protocols are searching for capital to LP and those who provide it do so because they expect a return (from price appreciation due to the growth generated by the initial capital allocation, from LP fee returns or emissions, etc.). They may not always be right, but that's how markets work. I don't really see the issue with lending either. Unless the founders are not actively trying to get liquidated to exit scam, it's a sensible use of their allocation. Founders need to eat too.
Can or can not choose, can also skip the vote, etc. How does it save the governance from random token holders having an influence on the proposals? Especially in the system when 1 token = 1 vote. Not saying that random token holders might not have enough understanding to vote in the best intentions of the protocol, it is also a big risk for every protocol, when someone can accumulate enough gov tokens to vote to rug the protocol in one or another way.
If there is a risk that governance can rug the protocol, it means your protocol has a rug feature baked in it. Whether the decision to use that feature is decentralized or not is an implementation detail. The real problem is that your protocol has a major design flaw to begin with. If you look at Curve, Uni, Convex, Balancer, Aave or Yearn, the DAO can't do anything with users' money. Its role is to make operational decisions to ensure that the protocol continues to run smoothly (tweak parameters, update variables like contract addresses, etc.)
Interesting enough that Curve itself admits this problem
It's not particularly new. There are a few other vectors of centralization around the protocol (the single front-end, gauge management, registry curation) and pragmatic discussion on potential solutions and mitigation is always welcome. It's not like there's an hostile take-over of Curve by Convex either, both teams talk regularly and are eager to deploy a better voting system for vlCVX once it's technically feasible.
You make it sounds like LPs are unwitting victims
I am saying that most of the token emissions designs are 0-sum PvP
Founders need to eat too.
Quite often their appetites are unreasonably high.
If there is a risk that governance can rug the protocol, it means your protocol has a rug feature baked in it.
By saying "rug the protocol in one or another way" I do not mean literally withdrawing TVL from the protocol. For example when one big token holder can stream a lot of emissions to his own pool when a big token holder can liquidate the treasury when he can change the settings of the desired pool to fuck up LPers.
I would also say that Luna/UST scam could have been less harmful to the ecosystem if there were no possibility for Terra to influence gauge emissions in favor of UST. There was no rug, but people lost billions.
It's not particularly new. There are a few other vectors of centralization around the protocol
So Curve is centralized?
both teams talk regularly and are eager to deploy a better voting system
Today they talk, tomorrow they don't. Until they deploy smth it doesn't mean anything. (Don't get me wrong, I am not thinking that Convex wants to harm Curve or whatever, but saying that the system is decentralized cause "teams talk regularly" isn't serious).
Anyway, we declined far from the initial topic:
What is the proper use of token emissions? How should one properly account for them on the balance sheet? Are they a good or a bad thing for a protocol?
These questions do not have a simple answer, cause it really depends on how the token is integrated into the protocol. You can not just compare eg UNI and CRV. Curve is built around CRV, UNI is nonsense/use token.
If the protocol is built around the native token, then the emissions should be assessed very carefully, as they might harm the protocol. If the token has nothing to do with the protocol, then emissions are cool as far as someone is willing to provide liquidity.
I would also say that Luna/UST scam could have been less harmful to the ecosystem if there were no possibility for Terra to influence gauge emissions in favor of UST. There was no rug, but people lost billions.
That's putting a lot of blame on Curve and I don't think it's warranted. Luna/UST was $60 billions and the vast majority of that value was in Anchor. Combined, the Curve UST pools' TVL topped at a bit above $1 billion. Luna was able to direct emissions to their pool through a competitive bidding process (colloquially referred to as "bribes") and some of their own voting power (via vlCVX). They never reached near majority voting power, and the fact that gauge weights are always incremental (meaning that even if one entity obtains 100% of the gauge weight vote one round, they won't get 100% of the emissions) mitigates the risk of one voter suddenly sucking up all the emissions. Terra/Luna failed because of a flawed seigniorage design, Anchor's irrealisticly high returns and overconfidence in the stability of the LUNA price, among other things (a report by the risk team from December 21 highlighted some of these risks for LPs, in addition to oracle and smart contract risks).
So Curve is centralized?
This is the second time you're asking this. Maybe I wasn't clear the first time or maybe we've grown so accustomed to decentralization that we now take it for granted.
Curve allows anyone to trade a variety of assets, to create trading pools for any token, to deposit and earn yield on one's capital and soon to take out loans. And no one, not even the DAO, can prevent you or anyone else from doing any of that. You remain fully in control of your assets throughout. That's decentralization. This is not something that was possible 10 years ago, and it's not something that's possible in TradFi today.
Now of course someone could pull the plug on the front end because it's centralized. A government or the hosting company could shut it down. But the contracts are still around. You can still use them and trade by going to 1inch or paraswap or any other aggregator. You can go to etherscan and interact directly with them. You can load up web3js and write a script to run your trades. It's a very resilient system. Sure, ideally we'd have alternative front ends. We'd have a front end that anyone can run locally. Or we'd have something usable and reliable on IPFS. We're not there yet, but we'll get there eventually.
Today they talk, tomorrow they don't. Until they deploy smth it doesn't mean anything. (Don't get me wrong, I am not thinking that Convex wants to harm Curve or whatever, but saying that the system is decentralized cause "teams talk regularly" isn't serious).
My comment was about collaboration on technical solutions, not about decentralization itself (cf. comment above for that). The teams may stop talking and Convex could still roll out a voting system on its own. If the relationship turned openly antagonistic, the cvxCRV peg would likely be the first casualty, Convex would lose users and voting power to the benefit of competing protocols (keep in mind that the farmed CRV on Convex goes back to LPs as CRV, only locking CRV for cvxCRV increases Convex's voting power). Eventually the protocol's whitelist could get cancelled which would be a death knell. Curve would still work fine.
These questions do not have a simple answer, cause it really depends on how the token is integrated into the protocol. You can not just compare eg UNI and CRV. Curve is built around CRV, UNI is nonsense/use token.
Agreed, these questions can only be answered for a specific protocol or category of protocols and what you call "integrations". Unless there's some combination of utility, revenue share or claim to treasury/assets (am I missing anything?), the protocol is peddling snake oil.
Terra/Luna failed because of a flawed seigniorage design
This is not a question why anchor failed and nobody blames Curve for that. What I am saying is that these governance designs where 1 token = 1 vote and can influence economic outcomes might cause such situations where fraud can accumulate governance tokens and cash out through another protocol.
This is the second time you're asking this.
I have asked this for the second time because I am not very clear with your answer:
So we can agree that Curve is centralized af? No
Interesting enough that Curve itself admits this problem It's not particularly new.
So, if one multisig controls 51% of the other protocol governance, does it make the later protocol centralized (of course governance can not change immutable contracts, but there are other aspects)? Just yes or no?
If the relationship turned openly antagonistic, the cvxCRV peg would likely be the first casualty
The relationship does not influence the peg, but the economic incentives for others to maintain the peg.
Frens, while I would be very interested in hearing about the level of Curve decentralization, it seems the line of discourse there is at a wrap. Let's do as Ser Made Here suggested and revisit the topic of token emissions. DeFi Made Here -- your last paragraph, you cited "then the emissions should be assessed very carefully, as they might harm the protocol" -- are you contending that $CRV emissions are harming the protocol? Or is it an exception in your view?
I have explained my point in some of my threads before. Emissions cause selling pressure which is not offset by buying pressure in the case of Curve. At the moment Curve has the highest TVL, but trading volume is significantly lower than on Uni. This means that LPs are incentivized and are farming CRV tokens. If CRV price goes further down then LPs will pull the liquidity -> less trading volume -> less fees -> etc. I can not say that the emissions are bad - Curve is built around this system, I am saying that Curve will suffer a lot if there will be no buyers for these emissions. One would argue that CRV has utility in form of fee sharing and bribes, etc, but the more tokens emitted - the more diluted the yield. Let's say I have never believed in this system in the long run, but I consider that crvUSD might drastically change the situation.
@benny, your response to any of the above?
So, if one multisig controls 51% of the other protocol governance, does it make the later protocol centralized (of course governance can not change immutable contracts, but there are other aspects)? Just yes or no?
No. You're conflating the protocol and its governance when the latter is a strict subset of the former.
I have explained my point in some of my threads before. Emissions cause selling pressure which is not offset by buying pressure in the case of Curve.
This statement is true when the price goes down and it's not true when the price goes up. CRV is roughly at the same USD price it was 2 years ago and in that time period the supply increased by half a billion tokens so clearly it's been offset. But take other reference points and you can make widely diverging arguments. It's not very helpful. Also, the impact of emissions is often overstated because locks are not taken into account. We've seen a lock rate of over 50% for most of 2022 and we're now still above 30% in what is (hopefully) the bottom of the bear, so the effective inflation rate is much lower than the 28% advertised.
At the moment Curve has the highest TVL, but trading volume is significantly lower than on Uni. This means that LPs are incentivized and are farming CRV tokens.
More volume is not always good. A lot of volume on Uni is toxic flow, to the detriment of LPs. We see the same configuration on a few other DEXes that artificially boost volume numbers with ultra low fees. By contrast, Curve sees less toxic flow and captures more volume from non-arb traders. So Curve is still a better place to LP on-chain even if you entirely discard the additional CRV token incentives. Returns are only one factor of course. Brand recognition, first-mover advantage, VC backing also play a role, and explain some of Uni's ability to continue to attract liquidity despite its comparatively lower returns for LPs. Emissions are a way to give Curve a competitive advantage since it doesn't have any of these things. It allows the DAO to attract LPs in a decentralized manner by offering them all the same fair deal.
If CRV price goes further down then LPs will pull the liquidity -> less trading volume -> less fees -> etc.
The assumptions here need to be nuanced:
LP will not pull their liquidity if the price goes down. They will pull it if the price goes down enough that alternatives become more attractive. And on-chain, today, there aren't that many. For LPs, Curve is still competitive with Uni and other DEXes even with low-priced or no CRV emissions. Other options like GLP or similar products carry a very different risk profile and don't scale in the same way. Off-chain, you'd have to trust a CEX or a bank. The former option isn't that sexy after the FTX collapse but the latter is still decent. T-bills are definitely a serious competitor for DeFi right now.
Liquidity does not always correlate with trading volume. There are plenty of pools with low TVL but high liquidity utilization on Curve. CNC/ETH has 4m TVL but generates great volume (and fees). FRAX/USDC sits a half a billion in TVL with very low utilization. But that's because the pool sells peg stability rather than volume and provides revenue to the DAO via bribes instead of trading fees.
On the other hand there are other factors driving volume. Some are out of Curve's hands (general market conditions and adoption), but some are not. Better integration with DEX aggregators, fee adjustments to capture (ideally non-toxic) volume from pools on other DEXes, gas cost reductions can all increase curve's share of overall on-chain volume. Coincidentally these are all areas that are seeing new developments. For instance, a swap on the tricrypto pool costs anywhere between 250k and 600k gas. This is much more than on Uni where it'd set you back 150k gas. So Curve is at a disadvantage on aggregators - and ETH-stable pairs are some of the largest volume drivers on mainnet. The upcoming version of the tricrypto pool will shave this down to less than 100k gas and be much more competitive.Trading volume correlates with fees but not all trading volume is created equal. Having better dynamic fees (and having them on stable swap pools as well) can help increase revenue (cet. par.) and reduce toxic volume. This is also currently an active area of research at Curve.
I can not say that the emissions are bad - Curve is built around this system, I am saying that Curve will suffer a lot if there will be no buyers for these emissions. One would argue that CRV has utility in form of fee sharing and bribes, etc, but the more tokens emitted - the more diluted the yield. Let's say I have never believed in this system in the long run,
I guess it depends what your vision for the long term is. If you believe, like I do, that stablecoins will continue to grow and be a major factor of DeFi adoption, that adoption will increase with new generations who are growing up with crypto as their first and main gateway to finance, and that trust in CEXes will continue to erode and drive volume and pricing power on-chain, then you're expecting an increase in on-chain volume that will make AMMs self-sustainable. Emissions are a means to that end.
Now maybe that will not happen. Maybe AMMs never capture enough volume. Or maybe it won't happen on ethereum or the chains where Curve is available. Or maybe the emission schedule is not appropriate to get there, or not flexible enough to allow the protocol to effectively react to market changes. These are all valid possibilities/criticisms. IMO if we are truly talking long term, the discussion should be on these points. Three years from now inflation rate will already be down significantly.
but I consider that crvUSD might drastically change the situation.
crvUSD is generous icing on the cake and a few cherries on top. But Curve's v1 and v2 pools are already competitive products and will continue to improve. TVL and volume may be going down under current market conditions, but that is an ecosystem-wide phenomenon.
This statement is true when the price goes down and it's not true when the price goes up.
?? Yes, when the price goes up that means there are more market buyers than market sellers.
CRV is roughly at the same USD price it was 2 years ago and in that time period the supply increased by half a billion tokens so clearly it's been offset.
For the last 12 months, the selling pressure was not offset by buyers + lockers. By it is true indeed that CRV price is the same as when TVL was ~$1B.
Also, the impact of emissions is often overstated because locks are not taken into account.
In my threads, I take them into account.
We've seen a lock rate of over 50% for most of 2022
Unpopular opinion, but a lot of people did not realize what risks they were taking. I personally know whales who were trying to run DN strategy staking cvxCRV and shorting CRV.
and we're now still above 30% in what is (hopefully) the bottom of the bear
In fact, the locking rate went below 30% before yCRV went live and has since then increased to 34% as of today. I do not see a reason for the locking rate to increase in the near future, as all liquid wrappers (cvx/sd/y-CRV) are traded below the peg. If someone wants to liquid-lock CRV it is cheaper to buy from the market, which is not increasing locked supply.
More volume is not always good.
Always more volume is good.
We see the same configuration on a few other DEXes that artificially boost volume numbers with ultra low fees.
How do they attract LPs?
So Curve is still a better place to LP on-chain even if you entirely discard the additional CRV token incentives.
How did you come to this conclusion?
explain some of Uni's ability to continue to attract liquidity despite its comparatively lower returns for LPs
Uni has comparatively higher returns for LPs. Fees generated are 10x higher, while TVL is lower. But the average user won't be able to get great returns passively providing liquidity, active market makers could achieve high returns.
The upcoming version of the tricrypto pool will shave this down to less than 100k gas and be much more competitive.
That would be great, gas factor is a big problem for Curve on mainnet
If you believe, like I do, that stablecoins will continue to grow and be a major factor of DeFi adoption
I believe that stablecoins will continue to grow, but I do not believe in hundreds of various pseudo-DeFi stablecoins. So being a launchpad for stablecoins isn't the most attractive option long-term, imo.
crvUSD is generous icing on the cake and a few cherries on top.
I see it rather as a lifering
@DeFi Made Here I’m curious if you can point to some examples of protocols that do emissions very well in your opinion? $crvUSD notwithstanding, is there any steps you would recommend Curve take to shore up its position if you had unilateral control over the protocol?
I’m curious if you can point to some examples of protocols that do emissions very well in your opinion?
Convex! I am a big fan of performance based emissions. There are some other examples, like Nexus Mutual, but I am not very familiar with its design, so can not tell about them for sure.
$crvUSD notwithstanding, is there any steps you would recommend Curve take to shore up its position if you had unilateral control over the protocol?
I can not oppose myself to the Curve team and shouldn't even try to recommend them smth. But if I have to recommend anything then probably try to optimize gas and integrate Curve into as many UX aggregators as possible. In the future, very few people will be using protocols directly but swapping directly through the wallets on their phones.
I am also thinking that Curve should have deployed on BSC (2nd biggest by TVL and 1st biggest by active users) a long time ago, now there is a Curve fork live there and it would be harder for Curve to attract liquidity over there if possible.
And one more would be to expand to Cosmos. Cosmos would allow interconnecting with all IBC chains and Cosmos seems to get more traction soon, as finally, USDC will be launching there, dYdX launching own chain, etc etc. We can not predict what chain(s) will be dominant in the future, so being live on many chains is an advantage, imo.
benny any thoughts on protocols outside of Curve you think do tokenomics and emissions strategy particularly well?
Always more volume is good.
Only if you consider that LPs are agnostic to toxic flow (see Crocswap's analysis of LP returns on Uni for a counterpoint in which the low-volume 0.3% ETH/USDC pool on Uni outperforms the high-volume 0.05% one). But sure, let's discard metrics like markout or LVR (maybe the opportunity cost they measure doesn't matter to passive LPs) and other potential caveats (about IL, depeg events, etc.). This holds true, from an LP perspective, if all LPs benefit equally from increased volume. That is indeed the case on Curve but not on Uni v3 where if you're out of range you're out of luck.
How did you come to this conclusion? Uni has comparatively higher returns for LPs. Fees generated are 10x higher, while TVL is lower. But the average user won't be able to get great returns passively providing liquidity, active market makers could achieve high returns.
The fee comparison is not that straightforward, precisely because, as you mention, the distribution of returns on Uni is unequal. If you are confident about your ability to set and update your range over time, then sure you might be able to capture these comparatively higher returns. It's worth noting that this comes at a cost (gas, labor) so the fees are not net (gas costs for JIT are close to half of the revenue for instance). Otherwise your options are to LP passively but miss out on trading fees when out of range or eat IL and pay gas to update your range. So Curve is a much better option for the vast majority of LPs. UniV2 generally fails to outperform Curve for LPs. Balancer does offer an interesting alternative. They now combine an implementation of Curve's stableswap with their monolithic vault design that is very gas efficient. But you have to be comfortable with having such a single point of failure.
Another aspect of Curve that's lost in the comparison is that many protocols want stability/peg not volume and they compensate Curve LPs for the former without caring about the latter. They do that by directing CRV emissions via bribes (but could still post direct incentives via a gauge if CRV were worth $0). This then skews the TVL/volume ratio. You could say that protocols overpay for this to boost a vanity metric (TVL), but that's a different debate.
I believe that stablecoins will continue to grow, but I do not believe in hundreds of various pseudo-DeFi stablecoins. So being a launchpad for stablecoins isn't the most attractive option long-term, imo.
I guess Curve can kind of function as a stablecoin launchpad but that isn't the protocol's core value proposal. If by "various pseudo-DeFi stablecoins" you mean shady stables like UST, USDD and their myriad less illustrious counterparts then sure, there's not much of interest there (although their demise make good fees for the DAO). But we're also seeing new stables being issued by emerging regulated EMIs in Europe or big banks in Australia. None of these are decentralized (nor necessarily noteworthy currently) but they could potentially offer better off/on-ramp experience and interfaces with centralized financial services. And they'll need a marketplace. They might eventually consolidate, but we're still far from being there.
@benny any thoughts on protocols outside of Curve you think do tokenomics and emissions strategy particularly well?
I've mentioned Chainlink before. From a tokenomics POV, the token has utility as it is used both to pay nodes for bringing external data on-chain as a sort of oracle gas and to ensure the quality of that data (node stake LINK which can be slashed if they provide incorrect data). Chainlink is not a DAO and its operations are very centralized so there are no "emissions" to speak of, just tokens that the protocol (via its foundation) hands out as incentives to develop the network, ensure its security, build a moat and establish Chainlink as the industry standard.
Ultimately the situation is not too different from Curve in practice, token's current circulating supply is increased and holders diluted to finance the ecosystem's development and generate value in the future. It's common for early stage ventures to burn money to finance growth. Web3 is no different in that aspect and emissions represent a similar strategy. It's fine if there's a credible path to profitability. I guess the difference with Curve, is that centralization makes things smoother for Chainlink. CRV emissions are immutable and scheduled and they need to be handled by the DAO - so there's less adaptability and interesting problems stemming from decentralized management.
One more question to toss into the mix to both of you as we wind down the discussion -- do you agree or disagree that a protocol's income should take token emissions into account? Is this even a meaningful metric?
Uni v3 where if you're out of range you're out of luck.
Agree, uni v3 is for active (professional) market makers, passive LPs should use true AMMs such as Curve, for example. Don't know why we ended up discussing AMM designs though.
But we're also seeing new stables being issued by emerging regulated EMIs in Europe or big banks in Australia.
Do you think they will chose Curve and bribe someone to provide liquidity for their coins?
do you agree or disagree that a protocol's income should take token emissions into account? Is this even a meaningful metric?
When the token is used to attract TVL then yes. Because there is a strong correlation between TVL and token incentives. If such a protocol removes incentives it is likely to lose TVL as well. However, there are examples when the protocol used early incentives to attract TVL and then gradually reduced token emissions without losing the TVL.
Do you think they will chose Curve and bribe someone to provide liquidity for their coins?
For EMIs and other fintech startups yes and yes. For banks, don't know if "they" will chose Curve but depending on how accessible the stablecoin is to end-users, anybody can technically create a pool. Bribes unlikely even if they get a more politically correct rebrand.
do you agree or disagree that a protocol's income should take token emissions into account? Is this even a meaningful metric?
It seems meaningful to me. Emissions are an expense for the protocol/DAO so you want to try and deduct them as such from your income, even if the actual valuation of that deduction is not always straightforward.
Let's move on to closing statements. @DeFi Made Here, you're up first.
What I can say is that all these topics are not black and white. Sustainable tokenomics is a very complex thing where protocols need to balance between different points. Token and project do not always have a correlation and a good project doesn't mean that its token is also good.
In general emissions for the protocol are good as far as someone is providing liquidity to exit. Projects that are built around native tokens are facing huge risks in case the token goes down in price. Protocols where the token does not have any relation to the operation of the protocol and there is no value accrual do not suffer from the token's price action.
The current governance system is not working as intended, imo. Situations, when one or a group of people have the extreme majority of governance power, might be dangerous for the protocol/users. As well as the system where 1 token = 1 vote does not seem fair to me.
DeFi is the future but there is still a long way ahead.
Wagmi,
DMH
Also thx @gerrit for making this conversation possible and thx @benny for the convo itself. Hope, I was not the worst person to debate with
@benny The last word is yours
The original question was quite large and I'm glad we could get into specifics, even though that did cause us to digress quite a bit - but in a productive way! To me emissions are really just a way to finance growth and indeed emissions can be squandered even by otherwise technically sound projects. Emissions also do offer a (potential not always actual) way to raise capital without backroom deals, by just relying on anonymous randos on the internet. I think DeFi Made Here's concerns about 1 token = 1 vote and concentration of power are well warranted, although I'm not as pessimistic as he is about the outcomes. In that regard, even though we may disagree on the semantics, I'm glad we got to discuss the issue of centralization with regards to Curve/Convex - as it's actually something i've been harping on lately
Thanks to @CurveCap for setting this up, it's been fun and I've truly enjoyed the convo with DeFi Made Here!
And that’s all! Remember that none of this has been financial advice!
Please drop us some comments on this new format. Share thoughts with either debater. Did you enjoy the debate? What could we do better? What topics or debaters would you like to see next?