Hacker News Reader

Comment by bitJericho | Hacker News Reader

This doesn't happen with regular cash because banks usually take this loss and pass it onto their customers as fees. The exchanges that have closed operate too unintelligently to be able to survive an attack. Almost 900 bitcoins in hot storage? That's almost 500,000 dollars being left in the open. It's like a bank keeping 500,000 dollars in a vault with no lock, no security, and no laws... With p2p coins, hot storage should be just enough for the day's operations and no more. Overdrew for the day? Make the customers wait, it's worth it for the safety.

View on HN

This doesn't happen with regular cash because banks have the scale to absorb it and pass it on to their customers. Bitcoin doesn't yet; and even if it did, there's no real way to restrict it.

Just as an example, Bank of America has over $2 trillion in deposits. If any minimally significant portion of that amount goes missing, it's pretty easy to track just due to scale. There are policies in place that ensure any transaction above a certain size gets looked at. If there are too many large transactions in a day, that gets investigated too. In exchange for having these safeguards in place, the government is willing to guarantee these deposits in the form of FDIC insurance.

Banking regulation is a good thing, especially when you're talking about an anonymous currency where transactions can't be rolled back.

EDIT: Just wanted to add that while Bitcoin itself probably will never be a globally significant currency, some form of cryptocurrency is likely to obtain relevance. But some people are going to get burned along the way, and these are the risks that you need to accept if you want to dabble in what amounts to unregulated banking. The regulations exist for a reason.

What advantage will another currency have over Bitcoin, that will trump it's lack of branding? If Bitcoin (with first mover advantage and network effects) fails, wouldn't it be twice as hard for the next to start from scratch?

My uninformed intuition tells me it's more likely that there will be a Gold and a Silver - one better, one worse, each used for different things.

Bitcoin's complete lack of centralized trust and inherent deflation seem like they may become big problems over the long run. I get that the peer-to-peer nature of it is considered a strong point by many -- but it also exposes users to fraud with little recourse.

It's basically like carrying cash: you can walk around with $10,000 in your pocket, but if someone robs you and gets away, you're out $10,000. If someone steals your credit card, the bank has fraud protection measures in place that limit their losses, and they often can roll back transactions. Unless you impose a similarly regulated structure on top of Bitcoin (where banks are super-secure and won't transfer large amounts of BTC without an auditable authorization chain so people become personally liable for any fraud that may occur) you're not going to be able to solve these problems.

Deflation is a bigger issue, in my mind. The algorithmic scaling of Bitcoin basically ensures that it will either not be very much in demand at all, or it will become increasingly scarce relative to demand over time. This creates an incentive to buy and hold Bitcoin as its value has tended to go up over time. However, most of the modern economy is based on the assumption that money today is worth slightly more than money tomorrow. This creates pressure to spend or invest, rather than hold onto currency for any reason other than liquidity. If Bitcoin remains popular, it won't be because of its virtues as a currency.

As I understand it, banks are insured against theft by central banks/governments in most countries.

Plus there is that whole thing of regulations about bank responsibilities.

That's when the bank goes out of business. Banks are hacked all the time the government does not (usually) bail them out.

Yeah, normal robbery is handled with normal insurance. The police are called, a police report is made, forensic accountants determine how much got swiped, the insurance company writes a check.

(I am not a banker.)

Ah you are correct, FDIC/CDIC/etc do not in fact provide theft insurance, but you help with the point I was making though - Banks have been robbed for years, and individuals don't lose money when that happens (how could they? your deposit doesn't sit as cash somewhere waiting for you to claim it)

According to the FDIC, for most banks theft is covered by the banks insurance policy (they refer to it as a "banker's blanket bond") which also covers loss (of money) by fire, flood, and even things like embezzlement etc.

So until BTC exchanges/etc actually have insurance policies (literal policies, not figurative "insurance policies") the risk seems higher.

Also worth considering - I remember reading about a bit coin site that uses a Safe Deposit Box to store the majority of its holding "offline", but safe deposit boxes aren't insured by either FDIC (even in the event of bank failure) or by the bank, so if their safe deposit box is breached (either as a theft or just damage) there is no safety net.

Bitcoin n00b here.

What exactly is a hot wallet/storage?

Hot storage is a wallet that is accessible online, eg, n wallet used by the "bank" to transfer coins from their own account to a customer that wants to be paid out.

A cold wallet is one where the keys are kept offline and not plugged into anything, eg, a printout, or a USB key.

So, is keeping a majority of bitcoins (>98%) bitcoins in cold storage the only way to keep them safe?

If they secured their servers (from hacking), they could theoretically leave everything in hot storage (although it's not necessary for normal operations as others have stated).

In the real financial world, matching and settlement occur asynchronously from each other, on different systems. Matching is "X tried to buy Y at Z, Q tried to sell Y at Z, their orders match." Settlement is physically delivering Y to X while physically debiting Z from Q.

Bitcoin developers haven't quite cottoned onto the wisdom of separating these functions architecturally. (One of many advantages is "If your matching system is compromised, you shut it down and investigate, but no money actually leaves. The settlement system is in your back office and much more protected than the matching system, because the settlement system doesn't have to talk to customers directly.")

Bitcoin developers instead have developed a security pattern called hot wallet/cold wallet, where BTC which are available to the system are "hot" and BTC which are not available to the system are "cold." The idea is that, in any given day, you might only require 2% or so of your company's total reserves to go in or out. You keep the private keys to, say, 5% of it on the live system. That's your hot wallet. You keep the private keys to the remaining 95% somewhere else. That's your cold wallet. Even if your live system is rooted, you should not (the thinking goes) lose the private keys to the cold wallet.

The Bitcoin community widely believes that this pattern is sufficient to prevent events like the recent Mt. Gox debacle, where the system was compromised and both the hot wallet and cold wallet were drained.

@patio11 - I dont understand. From what you wrote about matching-settlement vs hot-cold it seems there are still two systems at play there. I don't understand why there is a difference in security unless there is a time element in play (settlement at EOD).

Doesn't that violate the real time nature of bitcoin then? I have built e-commerce settlement systems in the past and I thought that the big challenge with bitcoin was always the instantaneous element.

You're correct, injecting extra time delay between transactions and settlement is one of the reasons why that architecture is more secure. That's a feature of it, not a bug. That is not the only difference: at almost all Bitcoin exchanges, your hot wallet is on your web tier and exposed to the adversary (so successful adversaries have authority to disburse 5% of your deposits), with the matching/settlement separation, a successful adversary still has no authority to disburse any percentage of your deposits. The hot/cold system also doesn't require e.g. intelligent accounting and reconciliation of those accounts, which is a major reason why the financial system actually works.

BTW: Bitcoin isn't a real-time system. The community widely believes it is, but people who actually understand what is happening would say "cough Yeah by 'real-time' we mean 'an hour later' cough."

Bitcoin developers haven't quite cottoned onto the wisdom of separating these functions architecturally.

I'm not sure this is true. Any off blockchain transaction is basically an unsettled (and therefore reversible) bitcoin transaction. So for example, trades on bitcoin exchanges and payments between web wallets will have separate and distinct settlement phases. Generally bitcoin enthusiasts gloss over this though, because they don't like the idea of reversible transactions.

The current maximum transaction rate for the bitcoin networks is something like seven transactions per second. So either they'll have to figure out how to increase that or move to a more conventional clearing and settlement system if bitcoin-as-a-payment-network ever takes off in real size.

> The current maximum transaction rate for the bitcoin networks is something like seven transactions per second

What? I'm not entirely sure that I understand this correctly: Do you say that the whole bitcoin network, with all that computing power, can't compute more than 7 transactions per second?

They can compute more than 7 transactions per second, but it is at present a hardcoded artificial limit (one of many) because if the network tried to sustain e.g. 100 TPS like an actual bank, the block chain would balloon in size to terabytes very quickly, and since every Bitcoin client needs every transaction in history, that would force most people to abandon it.

I don't know the details, but I think this arises from the rate at which blocks are discovered and the amount of space each transaction takes within a block. 7 tps is actually pretty high. All of paypal only does about 100 tps.

> Today the Bitcoin network is restricted to a sustained rate of 7 tps by some artificial limits. These were put in place to stop people from ballooning the size of the block chain before the network and community was ready for it. Once those limits are lifted, the maximum transaction rate will go up significantly.

See here: https://en.bitcoin.it/wiki/Scalability

The limit here is one of design: each block is currently limited to X MB, each transaction takes Y bytes, and each block is designed to happen every Z minutes; for Bitcoin's current values of 1MB (expected to be raised at some point if the size becomes a limit), something like 1k, and 10 minutes, that works out to 7 transactions per second.

Altcoins which have chosen blocktimes of say 1 minute will be able to do more transactions per second, and ones which lift the 1MB cap likewise.

Execution (match engines are just a part of that) and settlement aren't just different parts of an exchange architecture; they are often different business units, or even different companies.

Could you clarify your position?

Are you saying that Bitcoin-based financial systems cannot introduce a more secure settlement system without fundamental architectural changes to the Bitcoin protocol?

Or are you simply saying that nobody has apparently done so?

If the former, I would like to challenge that assumption. If the latter - what are you really trying to get at?

I think Mt Gox is the exception to the rule.

Yet nobody will pay fees for a bitcoin wallet.

Reply on news.ycombinator.com