DAI coin is just diversifying risk by using multiple underlying assets- So instead of a 10% chance of the value dropping 90% due to a failure of an asset guarantor, it will just have a 90% chance of dropping 10% in value. The MakerDAO organization can likely cover this 10% loss in asset value by using their enormous capital, but this is not a sustainable strategy for creating a stable synthetic asset.

I think sustainable synthetic blockchain assets are possible, but they will always have complicated risk/reward profiles that won't fully mirror the underlying asset they are designed to model: The dream of a truly "synthetic dollar" will always remain a dream.

EDIT: I should clarify that I think DAI coin may still be a useful construct for some situations, but it's going to be an asset with very different properties compared to any target real-world asset.

> sustainable synthetic blockchain assets are possible, but they will always have complicated risk/reward profiles that won't fully mirror the underlying asset they are designed to model

We figured out this doesn't work in the 19th century.

Reserves don't remove volatility, they just hides it. This is how banks work. And like a bank, a system of keeping "enormous capital" on the sidelines, ready to buy, works 90% of the time. When it doesn't, however, when people fear "enormous" is not enormous enough, they withdraw (i.e. sell), which prompts more selling, until eventually, since "enormous" isn't 100%, the buck breaks and the cards come crashing down.

Also people like to steal the "enormous capital," which is why we have regulations.

>this is not a sustainable strategy for creating a stable synthetic asset

Could you give some pointers for further reading on the subject please?

There's not really much reading on this- You simply can't take 10 assets that are worth less than $1 individually (because of guarantor risk) and mix them together to create an asset that's pegged at exactly $1- This is not a problem that can be solved by "diversification".

For other history on synthetic assets, read Preston's posts and also Vitalik's posts such as https://blog.ethereum.org/2014/03/28/schellingcoin-a-minimal...

MakerDAO's system over-collateralizes on DAI. Currently, it takes at least $1.50 of ETH to create $1 of DAI. This is based on the volatility and risk profile of ETH. These parameters are adjustable by MakerDAO token holders. Suppose they added gold to the asset basket where 50% of DAI was backed by gold and 50% was backed by ETH. They could require 125% position on gold and 150% position on ETH. In this case, 50% of the supply of DAI would be subject to the whims of ETH price while 50% would depend on gold. If gold or ETH were both to drop to near zero at nearly the same time so rapidly that the CDPs could not be liquidated in time, then the system would be at risk. Seeing as how Gold and ETH are not correlated, this is an unlikely event. Adding 8 more uncorrelated assets to the protocol would only further improve the stability of the system.

The inherent problem is that even if you are heavily diversified, in order to back by 50% gold you'd need to have 50% gold PLUS 50% MULTIPLIED BY THE PROBABILITY THE GUARANTOR FAILS.

For ether, such overcollateralization isn't a problem, because you can package it as an ether derivative and have no counterparty risk... But for a gold collateral you would have a risk that cannot be mitigated in this way and the risk will need to increase the slippage of the asset.

As time progresses, other guarantors will place their assets on the blockchain similarly to Digix. And when other guarantors come online, Maker will be free to add those to the pool. As it stands, Digix provides Assay certificates and undergoes third party auditing. MakerDAO is off to a great start. They're doing all the right things to build a sustainable stable coin.

I'd be curious what your thoughts are on the structure I propose in this comment of mine (different thread): https://news.ycombinator.com/item?id=16867880

If wanting to read from my parent comment: https://news.ycombinator.com/item?id=16851187

> sustainable synthetic blockchain assets are possible

Want do you have in mind?

Well, maybe something like an ethereum on-chain dollar ETF and a decentralized oracle for determining the exchange rate, which is built out of bonds that allow people to get leverage against the underlying ether currency in exchange for providing collateral for the ETF. The problem with such a construct is that (1) the people providing the collateral will insist on high fees and (2) the amount of collateral provided will need to be limited for enough people to participate, so the design would always have a risk for a "broken peg" during extreme fluctuations.

It's not quite the same thing, but Stellar allows the implementation of pegged Assets (e.g. USD), though they are issued by an anchor that you must explicitly trust to redeem those deposits. That could be a bank or other well-capitalized institution however.

This is basically the BTC Tether model, except hopefully some anchors will step up that can actually complete an audit without breaking up with their auditors.

I'm not all that convinced that it will be possible to create a stable synthetic blockchain asset without either explicitly pegging to fiat (a la Stellar) or having a big and diverse enough slice of GDP flowing through the system so that speculative activities are a minority of transaction volume.

This is more or less how DAI works. And in the case of a broken peg, there's a global settlement option that can liquidate all the collateral. https://developer.makerdao.com/dai/1/stability

I think the defense of "nothing has happened yet" is a tough one to back, in particular given that during the build up to the last recession, 1) collateralized debt was claimed by wall street to be "as good as cash" and 2) the housing market would "always go up". It works until it doesn't.

DAI is set up so that it's not dependent on ETH always going up. People create DAI by locking up ETH (or in the future, other assets) as collateral. Currently for ETH, you have to lock up 150% of the value that you take out in DAI (but you can collateralize higher than that if you like). If the price of ETH falls below a certain value, then a position can be liquidated automatically by the system, to cover the position.

DAI has a goal of adding additional forms of collateral in the future. One could see a coin like DAI being backed by a mix of Gold, real estate, commodities, or securities to achieve greater stability.

> several black swan events

That word doesn't mean what you think it does. Nothing that's happened since it launched is that unexpected.

Couldn't agree more.

The author of the OP has an article about DAI too: https://prestonbyrne.com/2018/01/11/epicaricacy/

Doesn't seem so hot to me, and what happens when the bot runs out of capital?

He has a weird definition of "break". At no point did the value of collateralized ether backing each DAI go below $1 or even anywhere close to that.

There are valid arguments to be made against a DAI-like system like capital inefficiency, the appropriateness of the bounded volatility assumptions, and maybe sell spirals, but Preston Byrne's articles include a lot of invalid arguments.

Thats precisely his point. Contagion causes the break. Analogous to a bank run or the ruinous fall of the Thai baht in 97/98. Nobody has any real info on whether the bank has enough reserves, nor do they care. They just dont want to hold the asset.

In this case, they do. DAI's system is run entirely on the blockchain. It's auditable and decentralized. The biggest risk to DAI is if the value of the collateral crashed below the amount of DAI in circulation. The system has incentives in place to prevent this from happening. I'm not saying it's bullet proof, but I think it's a highly resilient system. I think adding multiple types of collateral should protect against the situation where ETH drops rapidly to near zero.

Bank eg was an analogy. The issue is contagion, which can be external in source.

What do you mean "contagion"?

The crypto hate on HN is enormous, and someone who evaluates MakerDAO for 15 minutes will speak authoritatively over its inevitable failure.

The core thing to realize about Maker is that all Maker does is loan Dai against an asset! It's collateralized. Ethereum may be risky to use as collateral, but something like Digix, where tokens are issued one-to-one with gold stored in a vault, means that you are now issuing Dai against a real asset (gold). If you think Maker will fail, you are arguing that the value of the asset backing Maker will fall.

> Digix, where tokens are issued one-to-one with gold stored in a vault, means that you are now issuing Dai against a real asset (gold)

Marketable collateral is an old idea, and suffers certain intrinsic difficulties. One is counterparty risk. Here we have at least three trust points: the place(s) the gold is physically held, Maker and the mechanism by which one holds Digix.

The classic case: Maker lies about the amount of gold in the vault (or steals the gold). Less classic case: the person holding the gold does the same. More realistic case: someone in this chain runs into financial difficulties, or messes up their AML or sanctions compliance program, and has their assets frozen and/or seized by some authority somewhere in the world.

Once you have possession of a gold token, no one can stop you from trading it, because of the blockchain. Converting your gold token into gold and having it shipped to you may be problematic, but once you have the gold token, it's transferrable.

Maker is decentralized, the problem there would be a bug in the smart contract.

And Digix being a failure / scam, that is indeed a failure point.

But what MakerDAO and Dai represents is not some "magic blockchain thinking", it is based on rational economic incentives.

> Converting your gold token into gold and having it shipped to you may be problematic, but once you have the gold token, it's transferrable

"I want to sell you this gold token. You can't convert it to gold, because the gold was all stolen."

As with any asset you don't physically own, you are trusting that it exists. Digix is insured, and any token backed by a physical asset you would need to trust the organization that issued it. But this is not rocket science - these types of businesses and the industry surrounding them have existed for a long time. The innovation, if you believe it's innovative, is that the representation of the asset exists as an ERC20 token.

> these types of businesses and the industry surrounding them have existed for a long time

And they've failed, via common mechanisms, for as long. Hence why issuers of marketable collateral are tightly regulated. This "innovation" updates an administrative aspect that always worked fine while leaving the dicier back-end not only untouched, but less regulated than before. It's analogous to rolling back to an un-patched OS, changing the color scheme and calling it progress.

> If you think Maker will fail, you are arguing that the value of the asset backing Maker will fall.

Or the entity having the key to the gold vaults decides to buy themselves a nice tropical island. The real world, it seems, does not expose a blockchain API.

Digix is incorporated in Singapore and goes through the same audits as any other gold storage facility, has their gold insured, etc. If you own paper gold that is backed by gold in a similar facility, then this is no different. If you own gold futures then it is ultimately backed by physical gold in facilities (although perhaps warped through various contracts and leveraged 5x).

It just happens that Digix gold is issued as an ERC20 token rather than registered with a physical gold exchange.

The issue is that it is uneconomical. Collateral is set at 4-5x. The market for people who want to make that trade is limited.

I know people who do it. They expect a long-term rise in ETH value, but they want to spend some money now without paying capital gains. They use Maker to essentially take a loan out on their ETH collateral. (And yes, they talked with their tax advisors.)

Thats exactly what I had in mind. A reasonably intelligent trader doesnt want to make his operation more capital intensive however. Has use cases, but limited.

> MakerDAO's DAI coin (currently collateralized by ETH) has already proven quite stable in the face of several black swan events and the price of Ethereum crashing nearly 70% over the past couple of months.

That is an interesting claim. What were these black swan events which DAI passed with flying colors?

I would argue the value of ETH (the asset that backs DAI) crashing from nearly $1400 a couple months ago to a low of almost $350 a few weeks ago is pretty strong. What other asset has these sort of swings in such a low amount of time? The fact that the collateral backing DAI has dipped and yet DAI has remained stable is a testament to the systems underlying it.

What could kill DAI would be ETH crashing to near zero rapidly.

You could argue that. And that is a good point. But it is simply price swings or volatility. The job of the stable coin is to be stable during periods of volatility, even high one as that.

A black swan event is the one you pointed out about ETH crashing to zero. And there were none yet.

From the DAI site:

    > If the value of ether held as collateral is worth less than the amount 
    > of Dai it’s supposed to be backing, then Dai would not be worth one dollar 
    > and the system could collapse.

    > Maker combats this by liquidating CDPs and auctioning off the ether inside before the 
    > value of the ether is less than the amount of Dai it is backing.

Note combats not prevents, it will go to zero with probability 1 [0].

[0] https://en.wikipedia.org/wiki/Gambler%27s_ruin