I agree that any decentralized stablecoin has to build up organic use uncorrelated with price moves. But do you think that's enough?
DAI's overcollateralization makes it safer than algostables but overcollateralization slows adoption (people are definitely not going to use it for payments) and reduces the money multiplier effect (e.g. $1 in loans causes $2 growth in economy) which further slows growth and prevents DAI from being used like a real world fiat currency would be. There would still be a use for DAI in more volatile situations.
Thus undercollateralization is needed, and perhaps FRAX's method of starting from almost collateralized (~90%) and progressively becoming less collateralized over time with a dynamically adjustable collateralization ratio through their algo market operations could work. This also coincidentally follows how fiats have progressed through time - pegged to gold, then floating and based off trust alone.
Then the question is whether such an algostable has real demand. So far, algostables' only tools that have gotten them to market caps of $20bn are: financial engineering/patently unsustainable blitzscaling by offering eyewatering APYs through burning VC money.
But even that doesn't solve the problem of people just simply not using the algostable. People in Argentina already have some ability to get USD, so FRAX needs to build ramps and payment methods independent of the ebbs and flows of the crypto market. Those ramps also need to be politically condoned since governments won't want to give up monetary policy. An algostable can afford to fly under the radar when its mcap is just $1bn, but for the future where it is backed more by faith than by collateral and more by real-world use than by lockups on Anchor, its mcap will be $100bn or more and somehow governments will have to be fine with it?
Even then, we have seen real world fiats with governments able to force citizens to use the currency get broken, notably by Soros twice. If that happens does a virtual currency with a smaller mcap and no real ability to force users to continue using it without kicking the can down the road (like Terra) stand a chance?
Interested to hear your thoughts on this. Also, it seems like it is much harder to get people to use stablecoins IRL for non-degen purposes, why do you think that's the case?
So first of all -- so many interesting things here. To address "is that enough"? I'd actually argue that organic use (or heck, manufactured use cases that aren't based on fake yield-farming or some circular demand equation) is the most important basis for the value of a currency, especially in the case of crypto because hopefully that organic use also signals some value-creating activity in the ecosystem -- this is kind of funny because in crypto currency almost all usage is conflated with value. Partially because it's a currency and partially because there's not a whole lot of other arguments for value although maybe one day owning a rare Ape NFT can give you the same social clout as owning a Monet who knows.
Anyways, I digress. But essentially I think collateralization is meant to serve as a 'safeguard' of value rather than the actual originator of value. This also holds true in Tradfi, when you borrow loans with collateral -- if you borrow money from a bank (or any lender) and get a loan, the bank would measure the value of the loan itself somewhere on your ability to payback the money, the cashflow that's generated from your paying the interest rates on the loan, but hey if you realllly can't pay back this loan then the bank can take the stuff you gave as collateral which makes them feel better.
On FRAX vs DAI: Not all collateral is the same. It's interesting, DAI is based only on ETH for collateral and so even though there's a steady requirement on 150% collateralization, in times like these when for ex. ETH literally dropped something like 40% today, if you got your DAI through putting down ETH on collateral, you're pretty fucked. Even though FRAX is not fully collateralized, they heavily favor USDC as collateral and I believe they have a fancy algo that calculates what the desirable risk thresholds are for the mix of assets they hold. Essentially FRAX should be a lot more stable than DAI due to the nature of collateral they ask for.
If you don't want to take my word for it, Jon Wu of Aztec mentions the riskiness of DAI using ETH as collateral in this breakdown of the Celsius spiral today (https://twitter.com/jonwu_/status/1536476104986267648?s=21&t=RnWaxs0vojip8DcJzyRNfw). I'm not sure whether I buy that overcollateralization necessarily slows down adoption on a number of points, and given that crypto isn't its own closed economy I don't know if the money multiplier effect can be applied here but that's an interesting take and would love to hear more.
I'm thinking that maybe my next hot take would just be comparing different stablecoins but my impression so far is that FRAX is actually more stale than DAI. But DAI has been around I think since '17/18 and FRAX was released late '20. If you check on Etherscan I think DAI has like 3x more transactions than FRAX on the ETH chain which kind of makes sense. Will write more about how to think about DAI/FRAX adoption versus other stablecoins.
To your why only degens in stablecoin thought: My opinion is that unfortunately, Crypto still has a long way to go in terms of 'linking up with the real world' as a viable, efficient, and truly reliable payment alternative. Reliability aside, Tradfi itself has a ton of friction when it comes to being used as a means of exchange, we just don't see it as regular consumers paying for coffee at the coffee shop but Stripe/Plaid are both multi-billion companies that literally just make regular way TradFi USD transactions cheaper and easier from one banking connection to another.
I guess I was feeling more like a crypto-skeptic when I was asking if that was enough. I definitely agree that organic use is the most important, but I am just unsure if this organic use is going to come. Are we going to live in metaverses? If we are going to, which version of the digital economy will actually succeed? Will successful implementations of GameFi beyond Axie and Stepn, on the level of games with huge microtransactional economies (e.g. Fortnite, League of Legends (yuck)) ever emerge? Game development takes time, so I suppose we'll know in a few years. Wondering what you think!
And spot on about how usage is conflated with value, though we should probably disambiguate and say that there are 2 types of usage: economic activity usage, aka buying a latte, paying for spin classes and financial activity usage, aka yield-farming, total return swaps etc. Both fiats and crypto are correlated with the latter, yet crypto has much much higher beta. And it seems to me that without engagement with economic activity usage total crypto mcap is just going to stay rangebound, below $3T and all flow in crypto is just going to be circular - people who made money being early in BTC and ETH investing in alts, then becoming "VCs" and investing in hugely illiquid web3 projects etc. Without new money and greater fools coming in this market cap isn't going to increase and the realistic outcome for all crypto is going to be mid at most and crypto in 2030 is going to look much like today's crypto, except with more regulation and more institutions. Don't know how you feel about the future of crypto (apologies for the possibly unwanted segue), but curious as to what you think, especially if you work in the space.
Anyways, I too digress. On collateralization, maybe I could explain a little by elaborating on the multiplier thing. Collateralization in tradfi is pretty much always undercollateralized, which enables most people to buy houses, raise debt for companies etc. This raised level of demand facilitates a raised level of supply, which just brings more money into the economy in general. People can buy houses, but they also have more cashflow freed to spend in the economy etc. It's weird because in some regions of crypto you used to be able to find 125x, 100x leverage (I don't remember the source, it's not an extraordinary claim anyway, but there was an article for how Bitfinex pretty much always liquidated people who tried 100x leverage) which certainly helped pump up volume. But in the case of something that requires assurance (stables) you then need a lot of collateral which kills economic activity usage (since real world transactions need stable price levels). This is another convoluted reason as to why I'm skeptical about the digital/metaverse economy/think the future will just be regulation and stables like USDC/USDP. L1s/crypto need some kind of way/some promise that allows newcomers to make money, and I'm not sure if ETH become deflationary/disinflationary is secularly a bullish thing. The way economies grow is to have inflation more or less keeping pace with the growth of the economy itself, and ETH being deflationary is just going to price people who want to participate in economic activity out.
You make an interesting point about DAI, ETH and the quality of collateral. I don't think you're materially wrong but I think DAI takes other cryptos like WBTC, BAT even USDT as collateral. Apart from the stables though these are all fairly correlated tokens, especially during high-stress downturns so your point roughly stands - DAI takes as collateral coins that explicitly are going to dip during high-pressure moments. In March 2020 ETH dipped 50% and while DAI actually went up a little, though I believe conditions are obviously different - much like how UST survived a small depeg in 2021 but then got annihilated this time round. The only way to tell if overcollateralization is enough is time, though overcollateralization is fairly safe as long as the protocol is able to liquidate before the price dips too much. When $150 of ETH backs $100 of DAI then the integrity of 1DAI = $1 would be maintained as long as the protocol can sell the $1.5 of ETH on average for at least $1 of DAI. Even then it's quite a long road to $0 so that's an argument for overcollateralized DAI having higher collateral quality than slightly undercollateralized FRAX.
It's also interesting that FRAX prefers USDC as collateral, because then that just introduces more centralization, which I suppose is half the point of FRAX (the other being the lofty goal of replacing USD/becoming increasingly based on trust alone). I think that lofty goal requires decentralization, since the US Govt would be able to target transactions with USDC, so it remains to be seen how that turns out.
Ultimately, it seems like FRAX and DAI take different kinds of risk and thus have different purposes and different liquidities as stablecoins, much like your M0, M1, HQLA whatever. Thus it stands to reason that both could and should coexist, much like the multichain thesis for the future (though that is under attack these days).
All this talk about FRAX and DAI also makes me think that there are no shortcuts to growing algo stables, because compared to other protocols algostables face the same risks (smart contract, key person, etc) but also face risk that their incentives and mechanisms fail. Any growth-hacking would introduce mercenary capital that amplifies volatility and inhibits ability to ensure that the stablecoin actually adheres to principles that protect its stability. Unless you have an idea there, then we could go raise some millions from a16z together!
I'm fairly unaware of the FRAX ecosystem vs the DAI ecosystem so if you're writing about that I'll look forward to it.
You also make a very very good point about Stripe and Plaid. I am probably going to verge on a rant/verbal vomit here so apologies in advance. I think the transactions space is very ripe for further innovations (blockchain or not) that massively cut tx fees, yet for some reason we don't really see them (we don't talk about Chai).
As for why that is I think firstly one entire class of blockchain payment processors will never become as big as Stripe. These are the ones higher up the ladder of ambition, who operate in payments where one or more leg of the payment is in a stablecoin. Alice Financial on Terra was one of them. I think doesn’t get big because it targets crypto-natives, as such its growth is circumscribed by the growth of the respective L1 ecosystem itself. Even among big payment processors, only Visa currently supports USDC. There is a chance that people in Argentina, Turkey, Russia would end up using these payment processors, but I doubt such a niche product would ever go mainstream in the US.
The second class of blockchain payment processors are ones that simply have the blockchain living under the hood, and either have transaction approval executed through blockchain, or execute a fiat -> stable -> fiat operation under the hook. I’m going to go with a very theoretical argument against this because I’m very far from a payments expert. I think efficient systems tend towards centralization and thus a decentralized blockchain just doesn’t make much sense. The fastest blockchains are either theoretical (Celestia, some kind of roll up), or have a lot of downtime (Solano). It would take forever to grow a proprietary chain with nodes, fault tolerance, a good consensus mechanism etc to support payments (with what incentives too?) and have that somehow challenge the speed of centralized servers. It may happen one day, but all of that speed needs to also be accompanied with the other baseline features payment processors need to have, such as ease-of-use, supporting everything and integrating with everything else. Basically composability. Stripe does this with their 7 lines of code, integration with SquareSpace, Shopify etc, as does Adyen. I don’t want to say that something that hasn’t been done before is impossible, but the challenge is fairly immense and we haven’t really seen any examples of blockchains successfully replacing private networks. Perhaps it makes more sense to start with serving the crypto-native and eventually work on payments solutions as the technology gets better.
There is also the massive opportunity of facilitating bank/tradfi financial transactions/trades using SWIFT or whatever darkpool thing it is banks use and cutting confirmation times from T+2 to T+1, T+30min but presumably regulation is slowing much change on that front. Even still I think most banks are working on it (JPM tokenizing stuff, other banks looking into smart contracts/permissioned blockchains, FTX petitioning CFTC to get rid of FCMs in derivatives trading) I actually think that's where the real alpha is because while there are regulatory headaches there are maybe less programming headaches because there aren't really bad actors who are going to drain Citibank's FX treasury for example, and this system is far more antiquated than regular payments so there is a lot of opportunity.
I agree that any decentralized stablecoin has to build up organic use uncorrelated with price moves. But do you think that's enough?
DAI's overcollateralization makes it safer than algostables but overcollateralization slows adoption (people are definitely not going to use it for payments) and reduces the money multiplier effect (e.g. $1 in loans causes $2 growth in economy) which further slows growth and prevents DAI from being used like a real world fiat currency would be. There would still be a use for DAI in more volatile situations.
Thus undercollateralization is needed, and perhaps FRAX's method of starting from almost collateralized (~90%) and progressively becoming less collateralized over time with a dynamically adjustable collateralization ratio through their algo market operations could work. This also coincidentally follows how fiats have progressed through time - pegged to gold, then floating and based off trust alone.
Then the question is whether such an algostable has real demand. So far, algostables' only tools that have gotten them to market caps of $20bn are: financial engineering/patently unsustainable blitzscaling by offering eyewatering APYs through burning VC money.
But even that doesn't solve the problem of people just simply not using the algostable. People in Argentina already have some ability to get USD, so FRAX needs to build ramps and payment methods independent of the ebbs and flows of the crypto market. Those ramps also need to be politically condoned since governments won't want to give up monetary policy. An algostable can afford to fly under the radar when its mcap is just $1bn, but for the future where it is backed more by faith than by collateral and more by real-world use than by lockups on Anchor, its mcap will be $100bn or more and somehow governments will have to be fine with it?
Even then, we have seen real world fiats with governments able to force citizens to use the currency get broken, notably by Soros twice. If that happens does a virtual currency with a smaller mcap and no real ability to force users to continue using it without kicking the can down the road (like Terra) stand a chance?
Interested to hear your thoughts on this. Also, it seems like it is much harder to get people to use stablecoins IRL for non-degen purposes, why do you think that's the case?
So first of all -- so many interesting things here. To address "is that enough"? I'd actually argue that organic use (or heck, manufactured use cases that aren't based on fake yield-farming or some circular demand equation) is the most important basis for the value of a currency, especially in the case of crypto because hopefully that organic use also signals some value-creating activity in the ecosystem -- this is kind of funny because in crypto currency almost all usage is conflated with value. Partially because it's a currency and partially because there's not a whole lot of other arguments for value although maybe one day owning a rare Ape NFT can give you the same social clout as owning a Monet who knows.
Anyways, I digress. But essentially I think collateralization is meant to serve as a 'safeguard' of value rather than the actual originator of value. This also holds true in Tradfi, when you borrow loans with collateral -- if you borrow money from a bank (or any lender) and get a loan, the bank would measure the value of the loan itself somewhere on your ability to payback the money, the cashflow that's generated from your paying the interest rates on the loan, but hey if you realllly can't pay back this loan then the bank can take the stuff you gave as collateral which makes them feel better.
On FRAX vs DAI: Not all collateral is the same. It's interesting, DAI is based only on ETH for collateral and so even though there's a steady requirement on 150% collateralization, in times like these when for ex. ETH literally dropped something like 40% today, if you got your DAI through putting down ETH on collateral, you're pretty fucked. Even though FRAX is not fully collateralized, they heavily favor USDC as collateral and I believe they have a fancy algo that calculates what the desirable risk thresholds are for the mix of assets they hold. Essentially FRAX should be a lot more stable than DAI due to the nature of collateral they ask for.
If you don't want to take my word for it, Jon Wu of Aztec mentions the riskiness of DAI using ETH as collateral in this breakdown of the Celsius spiral today (https://twitter.com/jonwu_/status/1536476104986267648?s=21&t=RnWaxs0vojip8DcJzyRNfw). I'm not sure whether I buy that overcollateralization necessarily slows down adoption on a number of points, and given that crypto isn't its own closed economy I don't know if the money multiplier effect can be applied here but that's an interesting take and would love to hear more.
I'm thinking that maybe my next hot take would just be comparing different stablecoins but my impression so far is that FRAX is actually more stale than DAI. But DAI has been around I think since '17/18 and FRAX was released late '20. If you check on Etherscan I think DAI has like 3x more transactions than FRAX on the ETH chain which kind of makes sense. Will write more about how to think about DAI/FRAX adoption versus other stablecoins.
To your why only degens in stablecoin thought: My opinion is that unfortunately, Crypto still has a long way to go in terms of 'linking up with the real world' as a viable, efficient, and truly reliable payment alternative. Reliability aside, Tradfi itself has a ton of friction when it comes to being used as a means of exchange, we just don't see it as regular consumers paying for coffee at the coffee shop but Stripe/Plaid are both multi-billion companies that literally just make regular way TradFi USD transactions cheaper and easier from one banking connection to another.
This exchange is fun !
I guess I was feeling more like a crypto-skeptic when I was asking if that was enough. I definitely agree that organic use is the most important, but I am just unsure if this organic use is going to come. Are we going to live in metaverses? If we are going to, which version of the digital economy will actually succeed? Will successful implementations of GameFi beyond Axie and Stepn, on the level of games with huge microtransactional economies (e.g. Fortnite, League of Legends (yuck)) ever emerge? Game development takes time, so I suppose we'll know in a few years. Wondering what you think!
And spot on about how usage is conflated with value, though we should probably disambiguate and say that there are 2 types of usage: economic activity usage, aka buying a latte, paying for spin classes and financial activity usage, aka yield-farming, total return swaps etc. Both fiats and crypto are correlated with the latter, yet crypto has much much higher beta. And it seems to me that without engagement with economic activity usage total crypto mcap is just going to stay rangebound, below $3T and all flow in crypto is just going to be circular - people who made money being early in BTC and ETH investing in alts, then becoming "VCs" and investing in hugely illiquid web3 projects etc. Without new money and greater fools coming in this market cap isn't going to increase and the realistic outcome for all crypto is going to be mid at most and crypto in 2030 is going to look much like today's crypto, except with more regulation and more institutions. Don't know how you feel about the future of crypto (apologies for the possibly unwanted segue), but curious as to what you think, especially if you work in the space.
Anyways, I too digress. On collateralization, maybe I could explain a little by elaborating on the multiplier thing. Collateralization in tradfi is pretty much always undercollateralized, which enables most people to buy houses, raise debt for companies etc. This raised level of demand facilitates a raised level of supply, which just brings more money into the economy in general. People can buy houses, but they also have more cashflow freed to spend in the economy etc. It's weird because in some regions of crypto you used to be able to find 125x, 100x leverage (I don't remember the source, it's not an extraordinary claim anyway, but there was an article for how Bitfinex pretty much always liquidated people who tried 100x leverage) which certainly helped pump up volume. But in the case of something that requires assurance (stables) you then need a lot of collateral which kills economic activity usage (since real world transactions need stable price levels). This is another convoluted reason as to why I'm skeptical about the digital/metaverse economy/think the future will just be regulation and stables like USDC/USDP. L1s/crypto need some kind of way/some promise that allows newcomers to make money, and I'm not sure if ETH become deflationary/disinflationary is secularly a bullish thing. The way economies grow is to have inflation more or less keeping pace with the growth of the economy itself, and ETH being deflationary is just going to price people who want to participate in economic activity out.
You make an interesting point about DAI, ETH and the quality of collateral. I don't think you're materially wrong but I think DAI takes other cryptos like WBTC, BAT even USDT as collateral. Apart from the stables though these are all fairly correlated tokens, especially during high-stress downturns so your point roughly stands - DAI takes as collateral coins that explicitly are going to dip during high-pressure moments. In March 2020 ETH dipped 50% and while DAI actually went up a little, though I believe conditions are obviously different - much like how UST survived a small depeg in 2021 but then got annihilated this time round. The only way to tell if overcollateralization is enough is time, though overcollateralization is fairly safe as long as the protocol is able to liquidate before the price dips too much. When $150 of ETH backs $100 of DAI then the integrity of 1DAI = $1 would be maintained as long as the protocol can sell the $1.5 of ETH on average for at least $1 of DAI. Even then it's quite a long road to $0 so that's an argument for overcollateralized DAI having higher collateral quality than slightly undercollateralized FRAX.
It's also interesting that FRAX prefers USDC as collateral, because then that just introduces more centralization, which I suppose is half the point of FRAX (the other being the lofty goal of replacing USD/becoming increasingly based on trust alone). I think that lofty goal requires decentralization, since the US Govt would be able to target transactions with USDC, so it remains to be seen how that turns out.
Ultimately, it seems like FRAX and DAI take different kinds of risk and thus have different purposes and different liquidities as stablecoins, much like your M0, M1, HQLA whatever. Thus it stands to reason that both could and should coexist, much like the multichain thesis for the future (though that is under attack these days).
All this talk about FRAX and DAI also makes me think that there are no shortcuts to growing algo stables, because compared to other protocols algostables face the same risks (smart contract, key person, etc) but also face risk that their incentives and mechanisms fail. Any growth-hacking would introduce mercenary capital that amplifies volatility and inhibits ability to ensure that the stablecoin actually adheres to principles that protect its stability. Unless you have an idea there, then we could go raise some millions from a16z together!
I'm fairly unaware of the FRAX ecosystem vs the DAI ecosystem so if you're writing about that I'll look forward to it.
You also make a very very good point about Stripe and Plaid. I am probably going to verge on a rant/verbal vomit here so apologies in advance. I think the transactions space is very ripe for further innovations (blockchain or not) that massively cut tx fees, yet for some reason we don't really see them (we don't talk about Chai).
As for why that is I think firstly one entire class of blockchain payment processors will never become as big as Stripe. These are the ones higher up the ladder of ambition, who operate in payments where one or more leg of the payment is in a stablecoin. Alice Financial on Terra was one of them. I think doesn’t get big because it targets crypto-natives, as such its growth is circumscribed by the growth of the respective L1 ecosystem itself. Even among big payment processors, only Visa currently supports USDC. There is a chance that people in Argentina, Turkey, Russia would end up using these payment processors, but I doubt such a niche product would ever go mainstream in the US.
The second class of blockchain payment processors are ones that simply have the blockchain living under the hood, and either have transaction approval executed through blockchain, or execute a fiat -> stable -> fiat operation under the hook. I’m going to go with a very theoretical argument against this because I’m very far from a payments expert. I think efficient systems tend towards centralization and thus a decentralized blockchain just doesn’t make much sense. The fastest blockchains are either theoretical (Celestia, some kind of roll up), or have a lot of downtime (Solano). It would take forever to grow a proprietary chain with nodes, fault tolerance, a good consensus mechanism etc to support payments (with what incentives too?) and have that somehow challenge the speed of centralized servers. It may happen one day, but all of that speed needs to also be accompanied with the other baseline features payment processors need to have, such as ease-of-use, supporting everything and integrating with everything else. Basically composability. Stripe does this with their 7 lines of code, integration with SquareSpace, Shopify etc, as does Adyen. I don’t want to say that something that hasn’t been done before is impossible, but the challenge is fairly immense and we haven’t really seen any examples of blockchains successfully replacing private networks. Perhaps it makes more sense to start with serving the crypto-native and eventually work on payments solutions as the technology gets better.
There is also the massive opportunity of facilitating bank/tradfi financial transactions/trades using SWIFT or whatever darkpool thing it is banks use and cutting confirmation times from T+2 to T+1, T+30min but presumably regulation is slowing much change on that front. Even still I think most banks are working on it (JPM tokenizing stuff, other banks looking into smart contracts/permissioned blockchains, FTX petitioning CFTC to get rid of FCMs in derivatives trading) I actually think that's where the real alpha is because while there are regulatory headaches there are maybe less programming headaches because there aren't really bad actors who are going to drain Citibank's FX treasury for example, and this system is far more antiquated than regular payments so there is a lot of opportunity.