Enough people were upset with my previous post claiming that Mt. Gox is suffering a bitcoin bank run because it is holding only a fractional bitcoin reserve that it caused me to think twice about the claim.Posted on 02/11/2014 4:32:38 PM PST by BfloGuy
Enough people were upset with my previous post claiming that Mt. Gox is suffering a bitcoin bank run because it is holding only a fractional bitcoin reserve that it caused me to think twice about the claim.
I could have been wrong, but I doubt it.
Actually, the backlash was strong enough that it made me think that perhaps I was using some common terms in strange ways. I referred to Mt. Gox as a bank, which is a stretch by some definitions, but it definitely stores bitcoin for clients and as such, acts as a custodian.
Maybe it was the term “bank run” that drew ire. Well, I’m pretty sure I know what a bank run is, but I turned to trusty Wikipedia just to be sure.
A bank run (also known as a run on the bank) occurs in a fractional reserve banking system when a large number of customers withdraw their deposits from a financial institution at the same time and either demand cash or transfer those funds into government bonds, precious metals or stones, or a safer institution because they believe that the financial institution is, or might become, insolvent.
Maybe I should have called it “a run on a bank”, but it´s six one way, a half dozen the other. Both terms describe exactly what I meant.
It could have been that what I was calling “fractional reserves” was somehow misplaced. (As I’ve written extensively on fractional-reserve banking, my potential misuse of the term troubled me greatly.) Again, turning to trusty Wikipedia:
Fractional-reserve banking is the practice whereby a bank retains reserves in an amount equal to only a portion of the amount of its customers’ deposits to satisfy potential demands for withdrawals.
Phew, so far so good. That’s exactly what I meant by the term.
So to summarize my claim, I have accused Mt. Gox of holding an insufficient amount of bitcoin in its “vault” to back all the claims to the cryptocurrency that its clients have.
I made this claim based on several basic pieces of economic theory. All of them involved ways that fractional-reserve banks operated back in a time before deposit insurance. Since there is no deposit insurance on bitcoin, Mt. Gox should be operating in a similar way to these banks, but only if it only has fractional reserves.
1. When fractional-reserve banks are faced with more withdrawal requests than they have money in reserve, they must halt withdrawals or face insolvency. This is what happens when a bank chooses to close its doors and not allow you to redeem your deposit. Mt. Gox has halted its redemption requests on-and-off for many months now, and as of February 7 has halted them indefinitely.
2. Before a fractional-reserve bank closes its doors completely, some historical episodes involved using an “option clause” to stymie the redemption requests of its depositors. The option clause, widely used in the Scottish fractional-reserve free banking era, allowed a bank to not honour a depositor’s withdrawal request, though it also obliged the bank to give the depositor an interest payment in lieu of the immediate return of his money. In sum, the bank would convert your deposit (which you could ask for on demand) to a loan (which it would pay back to you at its own discretion) and pay you an interest payment for your troubles.
Prior to shutting its doors this time ‘round, Mt. Gox charged clients a fee to receive their bitcoin quicker than a standard transfer would allow for. In effect, you could get your bitcoin on demand, but you would have to pay for it. Economically, this is just the mirror image of the option clause with the same end result – holders of bitcoin were motivated to not ask for their money, or were enticed to wait longer to receive it, than would otherwise be the case. (The fee to receive your transfer earlier was a hefty 5%, enough to dissuade depositors from asking for their large transactions to be completed quickly.)
3. Finally, fractional-reserve banks which over-issued bank notes typically saw their notes trade at a discount to those banks that did not over-issue. Since an over-issuing bank (i.e., one that operated with fractional reserves, or one with a lower reserve ratio than its competitors) would run a greater risk of illiquidity (or insolvency), customers would only accept notes issues by them at a discount.
As of my writing of this note, bitcoin at Mt. Gox is selling at about a 20% discount to its competitors (e.g., bitstamp and btc-e).
If it exercises an option clause like a fractional-reserve bank, sells its money at a discount like a fractional-reserve bank, and halts redemptions like a fractional-reserve bank, it’s probably a fractional-reserve bank.
So the question becomes, how did Mt. Gox become a fractional-reserve bank?
Many commentators noted that there is not much of a lending market for bitcoin, so as a result Mt. Gox could not be holding fractional reserves. This claim arises because all modern fractional reserve banks operate primarily by loaning out the money deposited in them. This is not the only way that a bank can end up with fractional reserves.
In the 19th century some grain elevators in the United States also held fractional reserves. They sold some of the grain deposited in them, and thus kept only a portion of it on hand for when farmers came around to make a withdrawal. (Incidentally, the practice of fractional-reserve grain storage caused such disruption in the market that it was made illegal – all grain elevators today must operate with a 100% reserve.) Under the gold standard, despite a thin lending market for gold, banks still managed to hold fractional reserves. All they had to do was sell the gold deposited in them for other assets.
All Mt. Gox needs to do in order to qualify as a fractional-reserve bank is not have 100% of the bitcoin in its vault that its clients have claim to.
Some people are keen to jump to the conclusion that Mt. Gox, or its owner Mark Karpeles, has fraudulently absconded with the missing bitcoins. This could be, and since the exchange is not exactly the most transparent financial institution in the world (and that says a lot!), it is not unreasonable. I think there is a better explanation for the missing bitcoins however.
The bitcoin protocol suffers from what is called “transaction malleability”. Long story short, it is possible (though as will see not assured), that a node in the network changes the associated hash in a way that invalidates it. The result is that two recipients can lay claim to the same transaction before it has been verified, or what is the same, that a transfer is paid twice.
Mt. Gox notably suffers from this debilitation, and as such operates in a way that allows some transfers to be sent twice. In effect, this would take bitcoins from its vault while depositors still have claim to them.
Mt. Gox has blamed this on a “design issue” of the bitcoin protocol, and gives this as the reason for the recent cessation of withdrawals. Nothing could be further from the truth, however, and to blame its current troubles on the general bitcoin protocol is disingenuous, if not an outright lie.
Other exchanges work around this feature of the bitcoin protocol, and do not suffer as a result. Mt. Gox is the only exchange, to my knowledge, which has this problem.
So, back to the original question – is Mt. Gox operating with fractional reserves? I think it’s clear that there is a pretty good chance of it. The methods it is employing at the moment mimic those of the fractional-reserve banks that existed before deposit insurance, and it is known to suffer from a technical glitch that exposes it to this problem.
To be clear, this is not a bitcoin issue but is rather a problem with one exchange which stores them.
So I end in the same way that I began several days ago
Ever wonder what a digital bank run looks like? Nearly one million Mt. Gox users are finding out first hand.
(Originally posted at Mises Canada.)
I certainly can't say, though, I think the author makes a good point.
Could the pristine e-currency possibly fall prey to the base motives of bankers?
Very interesting article and questions. Thanks for posting.
Timber!!!!!!!
Man who would have anticipated a liquidity crisis in a currency with 40%+ daily volatility......
Turkish Lira are a safer bet.
Click to be Added / Removed.
I thought this was going to be about Bitstamp cutting off withdrawals. Mt. Gox is old news. The problem seems to be fairly widespread at this point. End of the bubble?
The bitcoin protocol suffers from what is called “transaction malleability”. Long story short, it is possible (though as will see not assured), that a node in the network changes the associated hash in a way that invalidates it. The result is that two recipients can lay claim to the same transaction before it has been verified, or what is the same, that a transfer is paid twice.
This poorly written paragraph leaves me scratching my head. Transaction malleability is problematic precisely because certain items in the transaction can be modified in such a way that the modified transaction is not invalidated, resulting in more than one copy of the transaction (with different transaction ids) in the system. Only one of these transactions will make it into the blockchain, so no double spending can occur due to this. Mt Gox' problem is that they are resending withdrawal funds, because they don't see that the withdrawal transaction has completed under a modified transaction id.
That's not the right question. FTFY:
Could the pristine e-currency possibly NOT fall prey to the base motives of bankers?
BTC is suffering heavily from fees, fees and more fees. You pay to buy the coin, twice - you pay half of the bid-ask spread, and you pay the exchange. Then you pay for every payment that you make with your new shiny BTC. And, if you are tired of fees, you can sell your holdings... for a fee to the exchange, and paying the other half of the bid-ask spread. There is nothing in BTC that is not associated with fees - simply because the industry is staffed with young technocrats who suddenly found themselves in position of bankers. The BTC is not regulated in any way, and there is no self-restraint anywhere in the system. There is no competence either - exchanges are ran by computer geeks, not by professional bankers. As the article mentions, Mt.Gox has audacity to charge you 5% for the privilege of getting your own deposit!
Humanity went through those pains from thousands to hundreds of years ago. Economists know all those mistakes pretty well. However BTC financiers do not want to know any of that - they are busy making money, and whenever other people (including central banks of large countries!) tell them about dangers they ignore them. Well, you can always ignore the danger; this doesn't mean, however, that the danger will ignore you. In this case BTC will be inevitably suffering from the same problems that plagued any other currency in the world because BTC is not much different from them.
This can be explained in a simpler way:
Bitcoin protocol does not have an easy way to track transactions. Financial services cannot trivially follow the paper trail. As result, if they do not see the transaction that they expect, they may be willing to resend it, resulting in double expenditure.
Tracking of transactions is one of most fundamental properties of good accounting. You never receive money from a thin air, and you never send it out for no reason, and you do not internally transfer it without a paper that says why this should be done. When you receive or send money you get receipts or other confirmations from your nearest bank. The bank itself maintains a chain of other confirmations as the money is transferred between various places. You must know where your money is at any given time. (Perhaps this will take effort to trace, like being on a phone for 30 minutes, but all the data is in the system.) Bitcoin does not help you in tracking. Transactions look like these. There are no tags. You do have access to the source and destination address, and the amount. But this is insufficient. There may be many transfers from that source to that destination with that amount. Other (non-Mt.Gox) exchanges use complex logic to match transactions; perhaps they do a better job than Mt.Gox. It's not always mathematically impossible to track money - it's just difficult because there is no tracking number that you would put into your paperwork, like a purchase number. Those numbers are even preprinted for you on many documents, like FedEx waybills. Those numbers are sufficient to track the item. Bitcoin does not have them.
Tonto say:
Competition for gubbermint: Bitcoin
They not like competition. Monopoly more fun.
They find problem. Make illegal.
Shut all down, take everything away...
Only matter of time.
But that is far from saying that Bitcoin transactions are difficult to track.
This is not an ID that you attached to the transaction. The transaction structure (tx) does not have such fields at all. As I said, BTC was not designed by an accountant, so the protocol is lacking certain important features.
The ID that you see is simply the double SHA-256 hash of a serialized transaction. It is volatile; if any field within the transaction changes (regardless of how significant these changes are) the ID changes as well. This is exactly the problem that Mt. Gox and others are describing. Here you can see how transaction sequence field, along with the lock time field, can be used to manipulate transactions after they have been sent into the network. Here you can see a more detailed explanation from a developer.
This means that there are no permanent links (or GUIDs, or unique primary keys) in a table of your BTC transactions. You can generate a key from other fields... but it's not guaranteed to be unique. Similarly, two different keys, generated at different time, may refer to the same row. This is a major issue for any modern accounting system.
“This is a major issue for any modern accounting system.”
Heck, that is a major issue for just about any modern data system, accounting or no. Even a wet behind the ears programmer should know the importance of having a unique primary key if they want to track any kind of data.
Again, the transaction hash is volatile only until the transaction confirms. At that point the hash is fixed and will never change. I admit, the hash is not guaranteed to be unique, but you're much more likely to find a specific grain of sand on all the beaches of the world than to find a duplicate transaction hash.
As you point out via your link, there are relatively simple methods to track the transaction while it is being confirmed in contrast to your previous "complex logic" characterization. The transaction inputs will always be a fixed reference as a last resort.
Yes, but the sender of the money has no way to predict what it will be in the future. You have a ledger, and you want to record an outgoing transfer, and you want to track it. What can you use to do so? Only the fields that are static - the source address, and the input, and the outputs. In transfers between companies the source address and the input and even the outputs may be identical (if you buy 1 BTC every day, for example.) There is no provably reliable way (none at all!) to track those transactions.
I agree that the hash will become fixed after the transaction is confirmed... but which transaction? If you cannot identify the transaction without using the hash, then the hash itself becomes useless. As matter of fact, the hash contains less information than the transaction itself. It then becomes secondary to worry about uniqueness of the hash, as long as the primary question - "which transaction?" - remains unanswered.
If you look into the protocol you will see that the developer failed to include an immutable - say, 128-bit - field that would be entirely defined by the sender and could be used for tracking purposes. But it's not there today. It's possible to add the field... but imagine the work that it takes to up-rev the protocol and rebuild all the software.
Yes, there are ways to sufficiently accurately identify transactions, as long as you are the originator. But what does it mean "sufficiently" here? If you send one transaction per day then it's probably not a big deal; but if you send a million transactions per day, to same or different destinations, using same or different values, it may quickly become marginal.
Some sources (like the article above) theorize that Mt.Gox was a victim of death by 1,000 cuts, when they lost a little here and a little there. The more successful you are the higher are the chances that you will suffer those losses.
Looking at all this mess from high elevation, one can easily say that all those guys are $idiots, and that no one sane would do $this, and I, the genius, would do $that, and it would guarantee reliability. But we are not in their shoes. Mt.Gox was one of the first exchanges, and chances are that it is still running on the early code that was never audited (ha!) for such bugs. Maybe they did not even understand that transactions may be changed - the feature of the protocol that allows that is "not used currently," so they could assume that it won't ever be used. That could be a good plan back then, when BTC was ran by a few friends and experimenters. But today anyone can hack the client and do whatever he wants on the network, and he won't be your friend.
Whatever the true reason is, Mt. Gox claims that this bug is a problem for them. Is it truly such a disaster? Probably not, in vast majority of cases, and with sufficiently good software that looks for your transactions "out there." But they say it's a problem, and that's their story, so far.
The inputs to will never be identical to another as they are each a reference to a previous transaction output. Any given output may only be spent once, so will be unique as an input the transaction (unless you are attempting to double spend them).
If I was faced with this, I would define a transaction reference as a hash of the non-malleable portions of the transaction. In fact, hashing the transaction inputs should be enough.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.