Hacker News Reader

Comment by beefield | Hacker News Reader

I admit, I did not read that in detail. Can someone explain how the "proof of liabilities" is proving that it contains all liabilites of an exchange? for example the electricity bill that is coming to be paid next week or the off-book loan of x billion cryptocoin from your fellow exchange that you need to pay back also next week?

Awfully lot of trust you seem to need in this fancy world of trustless money of the future.

View on HN

If you close your eyes and ears, then it is possible to imagine that such filthy thing as offchain liabilities doesn't exist. At least when Kraken posts about their "proof of liabilities" without actual audit of offline liabilities on Reddit r/cc, they readily eat that claim, no one challenges it. And then in every single post about this new trend, they will write that Kraken is somehow solvent due to this. I guess this was the point - if people already believe in something, it doesn't need to be real :) .

I don't think it solves that at all, but, as I understand it, that's not been the problem _so far_. If someone has lent an exchange assets against its users' assets then it's not at all clear that they have the superior claim to those assets if the exchange goes bankrupt can't repay those liabilities.

What? I just checked news last week and there was a small thing about an exchange called FTX that seemed to have exactly this problem.

My understanding is that FTX lent customer deposits to Alameda, who lost them, not that FTX owes money to Alameda. In that scenario, FTX wouldn't have been able to prove they held customer deposits.

Didn’t FTX just hold the customer deposits in a scam token after some financial shenanigans? You would still need auditors to determine what assets on a company’s books were worth their stated value and attest to their liquidity.

Right, but "hold the customer deposits in a scam token" means they no longer hold them in the real tokens, which would be obvious using the method in the linked article.

Wouldn't this effectively reduce the role of the custodian from "bank-ish entity" to "safe deposit box operator"? The custodial entity would need to hold on to the exact assets provided by the customer (the equivalent of your bank holding on to the exact $ bills you originally deposited), which means they couldn't be used for revenue generating activities like investing or lending, which in turn means the custodian could only make money through user fees.

Once the custodian is allowed to do anything that causes the exact tokens a customer deposited to be exchanged for another asset, that opens the door to FTX-style malfeasance.

Yes, and I mean I think that's exactly the point: people aren't expecting their exchanges to be a bank that puts their assets at risk, they're expecting them to be a custodian only, and ding them with fees when they trade, or possibly to actively charge them fees for holding deposits. The only reason actual banks get with putting customer funds at risk is that deposits (up to a certain point) are insured by the government.

If the exchange wants to offer interest-bearing deposits (eg: loan out your crypto), well then you'd expect to no longer be able to verify your deposits because you'd know they were loaned out, or that the exchange was trading with them, or whatever else. You'd also be forced to confront the fact that your assets are at risk at that point.

Other creditors may or may not have preferential claims over depositors. The point is that you can't prove that the exchange is solvent (which was the entire point of this exercise) unless all assets and liabilities are considered.

> If someone has lent an exchange assets against its users' assets then it's not at all clear that they have the superior claim to those assets if the exchange goes bankrupt can't repay those liabilities.

Right, but if an exchange took out a bank loan, the loan was recalled, and the exchange repaid it with customers' deposited cash and later goes bankrupt - possession is nine-tenths of the law.

Especially if the exchange intentionally has a structure that can avoid international tax laws - who's to say it won't also avoid international bankruptcy laws?

> and the exchange repaid it with customers' deposited cash

Then in this scheme depositors would be able to see that straight away, and force bankruptcy and asset recovery on the exchange while those assets still existed. This would (presumably!) stop the exchange repaying debts with customer deposits, because those customers would know straight away and could seek legal recourse against the bank and the exchange.

> Then in this scheme depositors would be able to see that straight away,

If you trust the fiat bank account statements, certainly.

This is explicitly addressed in the article: use stable coins for that. I'm very much a crypto-skeptic, but I do think this particular solution will provide some level of increased safety for coin-heads.

If you trust the solvency of the likes of Tether and Luna, certainly.

Seems to me the problem of solvency at the boundary between the worlds of fiat and cryptocurrency is always present, and shuffling it around doesn't make it disappear.

But that's a choice for customers to make, at the end of the day. The exchange says "we're holding dollars in Tether and we can prove it" and if customers aren't happy with that, then they can not store fiat on the exchange.

All benefits of cryptocurrencies goes out the window when you introduce centralized exchanges. "trustless" is referring to the protocols, not the ecosystems.

It's not necessarily about covering all liabilities of an enterprise nor all its assets. It's about building a proof that you hold your customer assets and you are not running a fractional reserve. You may still run an unprofitable business and perhaps at some point you go bankrupt. But at least all the customer assets are there and segregated from the business assets and you cannot use them as the piggy bank for your business.

So, lets imagine you have been running this continuously from the beginning and there is absolutely nothing fishy on the chain of the events of the ledger that makes this proof. All customer deposits (1B worth of coins) are backed by respective assets (1B) in the portfolio. Unfortunately the business has been run badly, and in addition to this portfolio, the exchange has assets of one worthless laptop, but there are some tax liabilities worth 2B and an just found loan payable to Italian Mafia worth also 2B. Tax authorities file for bankruptcy.

1. How does the exchange ensure money is paid to the customers instead of Mafia or tax authorities?

2. If/when it can't, how these liabilities are included in the proof of liabilities?

(note: this is a real and difficult problem. That's why there are laws, regulations and deposit insurances around customer funds in finance, which, yes, fail occasionally. I just do not see how that can be solved by blockchain.)

You make good points, this by itself is not sufficient. That doesn't mean it's not useful, though. For Centralized Exchange you need additional regulation to ensure that level of customer protection. And yet, regulation itself is not enough, as you can still defraud in a regulated entity. You need that the weight of criminal punishment is hard enough to disincentivize it. And you possibly need insurance (FDIC and equivalents) too.

Hence why proponents argue to do things on-chain, where we have built-in guarantees and this issue disappears entirely.

> But at least all the customer assets are there and segregated from the business assets and you cannot use them as the piggy bank for your business.

Why would they be segregated? The priority of creditors in a bankruptcy proceeding is controlled by courts that will order assets handed over to senior creditors whether they are holding "on chain" liabilities or not. It is the disclosure of such a contract that is the problem of understanding all liabilities, both on and off chain, as bankruptcy court doesn't care about the distinction.

https://www.interactivebrokers.com/en/index.php?f=2334&p=fin...

> Client money is segregated in special bank or custody accounts, which are designated for the exclusive benefit of clients of IBKR. This protection (the SEC term is "reserve" and the CFTC term is "segregation") is a core principle of securities and commodities brokerage. By properly segregating the client's assets, if no money or stock is borrowed and no futures positions are held by the client, then the client's assets are available to be returned to the client in the event of a default by or bankruptcy of the broker.

Because segregation is a necessary prior to enshrine further protection on customer assets.

That seems to be a case in which regulation is actually protecting clients of brokerages. Of course, thise regulations are bad if you are a, say, crypto exchange, and want to, say, engage in "investment" activities using those client deposits (regardless if those a fiat currency, crypto or something else).

Absolutely. Centralized crypto exchanges should have the same regulation and oversight as brokerages/banks/other financial institutions. Because they are that type of institution, just having crypto under custody doesn't make it magically solve any issues.

> > Client money is segregated in special bank or custody accounts, which are designated for the exclusive benefit of clients of IBKR.

No, originally you talked about segregation between crypto and non-crypto. I responded that there is no such thing. You then changed to talking about segregation between client finds and company funds -- do you see how this has nothing to do with the original point? How do you know a brokerage doesn't have both crypto and non-crypto liabilities? You don't. And the non-crypto can be senior to the crypto. There is no segregation between non-crypto and crypto and no guarantees about what kind of liabilities are held by both clients and investors.

> if no money or stock is borrowed and no futures positions are held by the client, then the client's assets are available to be returned

Right, this is exactly why you need to know what all the off chain liabilities are. Because when they come due, you can lose all your crypto because of a brokerage's off chain liability. Thus there is no such thing as a proof of solvency.

The idea is that the exchange publishes a sum of their liabilities, and each individual user can check that their balance was uniquely included in the sum, cryptographically.

Unfortunately this does not prove in any way that the exchange has included all liabilities in the sum, it only proves your deposit is included. A very, very different thing.

Of course. But it lets every individual check this, which means if any individual's balance is not included, they can publish that. It is a vast improvement over the current state.

The problem is not the ones who want to publish if their balance is missing. The problem is the ones who want to hide their balance is missing.

For instance, if Alameda owes FTX billions of dollars, but both are owned by the same people.

This is addressed in the article.

Users could falsely claim that their balance is wrong.

This would be easy to cryptographically certify, this is not a problem.

It doesn’t cover any off-chain liabilities.

Note that for most exchanges a material portion (most?) of their liabilities will be off-chain—e.g. fiat customer deposits.

Reply on news.ycombinator.com