Live data from Hacker News

DeFi risks and the decentralisation illusion

bis.org

121–130 of 140 posts

Re: DeFi risks and the decentralisation illusion

#121
post #109

Earlier quoted context omitted.

Because if I already have $10,000 to put as collateral, to get a $10,000 loan to buy a $10,000 car, it means I didn't have a need for borrowing funds to begin with. I can go straight to buying the car without borrowing money. Whereas if I need to borrow the funds, I won't be able to get the loan because I don't have the 100% collateral required.

Using the BTC as collateral lets you stay long BTC and get the car. You also avoid paying capital gains since you aren’t selling. The positions are fundamentally different. If you take the loan you are long BTC and short dollars. If you sell BTC for the car you have zero of both.

Yes, we had already established that. The point is that requiring 100% collateral does not allow the borrower to trade future consumption for present consumption, which is the whole point of borrowing as far as consumer and businesses are concerned. This is why DeFi loans are unsuitable when it comes to funding consumption or productive investment.

Re: DeFi risks and the decentralisation illusion

#122

Earlier quoted context omitted.

The psychological momentum of cryptobros far exceeds that of central bankers. Cryptobros and affiliated parties are comprised of thousand of vanguard who have built their entire financial future and self worth on their very limited life experience and ignorance of even the most basic economic principles and history, and a technological screw driver behaving as a hammer looking for a nail. BIS participants are, at min…

Your comment takes me back to the 90s when fax machines were superior to email. You're clearly not paying attention to the space. Everybody in this space is 20-30 years old. Bet against demographics, I dare you.

Email and crypto are definitely comparable in their utility, usefulness, and technological advancement /s

Your comment takes me back to the late 90s in the dotcom bubble when everything was revolutionary and definitely not a scam

Re: DeFi risks and the decentralisation illusion

#123

Lost me at "lack of shock absorbers such as banks" - DeFi has lots of flaws, but it takes truly a sheltered economists to think that banks absorb shocks! I will pay attention when Goldman Sachs starts to hand out stimulus money or even loans in a crisis to absorb the shock

Anyone with sufficient capital and willing to take risks can be a shock absorber. The particular organization of such individuals such as into banks is merely an abstraction.

And this means ... What exactly? If it takes individuals willing to take risks then the crypto space has plenty of them. Are you saying that the point the authors are making is therefore not valid?

Re: DeFi risks and the decentralisation illusion

#124

Earlier quoted context omitted.

1. I wonder how much of this still holds in a potential future world where machines are making decisions in most aspects of life already. 2. It doesn't follow that you can't have some recognized authorities within decentralized finance to negotiate fallback cases. For instance, the role of banks could become merely to supply information related to human authentication, not most of market operation.

I guess you could set it up that way, but then you create the problem of competing authorities - it's the same problem that the internet has with DNS. At some point, it has to have a single source of truth. The blockchain itself isn't enough when what is recorded on the blockchain is potentially not what the actors intended (or is what a criminal intended in a fraud scheme).

Competing authorities seem like a feature not a bug. It allows for experimentation and evolution of governance models. International law works this way and is what allows people to “vote with their feet”. There is no actual requirement for a single source of truth.

Re: DeFi risks and the decentralisation illusion

#125
post #119

Earlier quoted context omitted.

> So the risk is limited as long as the loans can be liquidated in time in case of a price crash. If I may guess, it seems unlikely there are too many folks in DeFi circles who have ever heard acronym LTCM. (TL;DR: A bunch of actual Nobel laureates (no kidding, or at least as much as Nobel price in economics is an actual Nobel) founded a huge and famous hedge fund with a trading strategy assuming they can liquidate t…

Collateral is sold off automatically once they breach some threshold (varies, but it's always > 100%), and there are bots that compete to do this as quickly as possible. This is in no means perfect, but it is quite responsive, and did survive crypto's March 2020 price crash.

There were some denial of service issues that on Ethereum during the March 2020 crash that caused Maker DAO to have some Zero bid liquidations and the protocol had to mint and sell off some of Maker's token to keep all the issued DAI properly backed. This did not happen again in the crash in May, which was nearly as steep, but the protocols all kept functioning as expected.

Re: DeFi risks and the decentralisation illusion

#126

Earlier quoted context omitted.

I guess you could set it up that way, but then you create the problem of competing authorities - it's the same problem that the internet has with DNS. At some point, it has to have a single source of truth. The blockchain itself isn't enough when what is recorded on the blockchain is potentially not what the actors intended (or is what a criminal intended in a fraud scheme).

Competing authorities seem like a feature not a bug. It allows for experimentation and evolution of governance models. International law works this way and is what allows people to “vote with their feet”. There is no actual requirement for a single source of truth.

That seems kind of different in a way, the different authorities cover different countries where laws are different.

It almost seems like to make it work you would need...regulation.

Re: DeFi risks and the decentralisation illusion

#127
post #121

Earlier quoted context omitted.

Using the BTC as collateral lets you stay long BTC and get the car. You also avoid paying capital gains since you aren’t selling. The positions are fundamentally different. If you take the loan you are long BTC and short dollars. If you sell BTC for the car you have zero of both.

Yes, we had already established that. The point is that requiring 100% collateral does not allow the borrower to trade future consumption for present consumption, which is the whole point of borrowing as far as consumer and businesses are concerned. This is why DeFi loans are unsuitable when it comes to funding consumption or productive investment.

Uncollateralized lending in DeFi is very nascent (right now primarily targets crypto businesses, and are typically ran by centralized companies who have launched a protocol on chain).

TrueFi, Maple Finance, and Goldfinch are the biggest and primarily have permissionless lenders and kyc'd borrowers. Some of those borrowers may make consumer loans (Goldfinch is like this).

Permissionless uncollateralized borrowing has yet to take off (even though contracts for this already exist and are live), but I suspect it will once decentralized stablecoin on chain supply gets decoupled from current centralized stablecoin supply (decentralized credit based stablecoins built on top of incentivized permissions management of on off chain flows [via over collaterlized decentralized stablecoins and centralized stablecoins alike] and on/off chain risk [via derivatives]). Decentralized derivatives protocols will be key to permissionless uncollateralized lending growth imo, but we are not there yet (I think we need to continue to see global markets break down more in OTC/CCP IRD's and tradfi counterparties continue to lose trust with one another in derivative transactions for this to grow faster in DeFi).

I can see that in the next 10-20 years, 20% of the eurodollar system with be contained within (multichain) permissionless DeFi protocols as HNW individuals and tradfi institutions outside of the US abandon CCPs and typical OTC derivatives txs.

I won't have to argue with folks at ihsmarkit like I do now for making EOD CDX data public (like it was before they were acquired by shit & pee global), when I can pull it from on chain contracts in real time.

Re: DeFi risks and the decentralisation illusion

#128
post #121

Earlier quoted context omitted.

Yes, we had already established that. The point is that requiring 100% collateral does not allow the borrower to trade future consumption for present consumption, which is the whole point of borrowing as far as consumer and businesses are concerned. This is why DeFi loans are unsuitable when it comes to funding consumption or productive investment.

Uncollateralized lending in DeFi is very nascent (right now primarily targets crypto businesses, and are typically ran by centralized companies who have launched a protocol on chain). TrueFi, Maple Finance, and Goldfinch are the biggest and primarily have permissionless lenders and kyc'd borrowers. Some of those borrowers may make consumer loans (Goldfinch is like this). Permissionless uncollateralized borrowing has…

I think you're not understanding the fundamental problem that uncolletarelised DeFi lending faces, which is the fact that the borrower can simply walk away with the money. None of the companies that supposedly offer uncollaterilised borrowing do what they claim to do. This is obvious if you read the fine print. And it's to be expected, because if they did, borrowers would borrow all the available funds and walk away, never to be seen again. That's not a sustainable business model.

Re: DeFi risks and the decentralisation illusion

#129
post #5

The decentralisation illusion argument seems weak. One could have claimed the same about the early internet: Early internet wasn't truly decentralized as there were still ISPs, and you still need to register your domain with a centralized entity. This misses that the big change is one of access. Content creators were able to reach a large audience without playing ball with the big publishers or newspapers. With defi,…

Instead of having to curry favor with bankers, you now have to do so with a clique of developers, lest they hard or soft fork your crypto out of existence. “The development community is proposing a soft fork, (with NO ROLLBACK; no transactions or blocks will be “reversed”) which will make any transactions that make any calls/callcodes/delegatecalls that execute code with code hash (ie. The DAO and children) lead to t…

At this stage of maturity, ETH would never survive a fork like that,

which is pretty evident given the umpteen hacks that have happened since going unabated!

Re: DeFi risks and the decentralisation illusion

#130
post #128

Earlier quoted context omitted.

Uncollateralized lending in DeFi is very nascent (right now primarily targets crypto businesses, and are typically ran by centralized companies who have launched a protocol on chain). TrueFi, Maple Finance, and Goldfinch are the biggest and primarily have permissionless lenders and kyc'd borrowers. Some of those borrowers may make consumer loans (Goldfinch is like this). Permissionless uncollateralized borrowing has…

I think you're not understanding the fundamental problem that uncolletarelised DeFi lending faces, which is the fact that the borrower can simply walk away with the money. None of the companies that supposedly offer uncollaterilised borrowing do what they claim to do. This is obvious if you read the fine print. And it's to be expected, because if they did, borrowers would borrow all the available funds and walk away,…

> I think you're not understanding the fundamental problem that uncolletarelised DeFi lending faces, which is the fact that the borrower can simply walk away with the money.

I understand it very well, that's pretty much the risk to be mitigated (or not) by who the loans are extended to on the protocol level (when not trying to do it in the KYC/ofchain legal agreement way which is how its done now for the most part). Pools of capital can be lent to specific actors in a non permissioned way that can be governed by the the protocols users or on/off by the on chain contracts themselves automatically when certain on chain conditions are met.

Also, for the non corporate uncollateralized lending in defi now through flash loans (i.e. via Aave), it is impossible for the borrower to walk away from borrowing the funds because the loan must be paid back in the same transaction or entire transaction reverts. However this isn't appropriate for typical consumer loans.

Currently, a lot of the centralized companies with their protocols on chain mitigate the risk just by restricting the pool of borrowers to those who they can legally go after to recoup any losses in the event of a default (just like in tradfi, but still the risks remain).

In the case of a derivatives protocol, writers can borrow against buyers deposits (instead of having to put up their own stablecoin deposits to back the writing) to open positions with the expectation that the writers can write enough volume to net out the delta most of the time while capturing a spread. If/when they (the writers in the derivative liquidity pools) can't and if enough addresses choose to withdraw the decentralized overcollateralized/centralized stablecoins from the protocol (rather than transferring/swapping their protocol credit to another address who wants to buy or write derivatives, or use as a unit of accounting outside of the protocol) and there is a shortfall, decentralized overcollateralized/centralized stablecoin yielding debt tokens can be issued by the protocol automatically (as well as raising decentralized overcollateralized/centralized stablecoin collateral requirements across the board for writers who haven't been cleared by protocols risk management contracts or by some kind of on chain governance) to those trying to withdraw who can sell it on a dex at a premium or discount to par value of the stablecoin yielding debt token.

The risk doesn't go away in tradfi with all the uncollateralized lending now, it gets spread throughout all the actors of the system in various ways, much of which isn't very transparent to all actors in the system (and even for those in the know, it is not in real time). The same (spreading risk through various actors that engage with the protocol) can be done in a DeFi context minus the opacity we have now (we all can see what addresses have/done what, regardless of whom/what is behind the address).

There isn't going to be a one size fits all approach to uncollateralized lending in DeFi. Protocols will do it differently based on what the users see fit to do with their funds and will manage the risks in many different ways (some of which will be better than others).

Post reply on HN