We are Mt. Gox: AMA | Hacker News Reader

We are Mt. Gox: AMA

reddit.com

51 pointsnico47 comments

Talk about putting up a shingle and calling yourself an exchange.

The clamor from the crowd that they need to do better PR gets me chuckling the most. They are fairly transparent about their lack of technical expertise, hinting at major issues in their understanding or seriousness of what they are doing.

When one noisy person tells you how to do your job and you know they're wrong, whatever. Brush it off.

When a fistful of redditors are telling you 15 different ways that you should be doing your job and it's pretty easy to show that they understand the problem domain and issues better than the team... that's a big problem.

I guess people hope MtGoxs new trading platform scales better, but that is not the sort of thought/wonder to be having when you are about to press submit on thousands in trades.

Bitcoin. Currency like kids and potato cannons. What could go wrong? Fire it up!

"Bitcoin. Currency like kids and potato cannons. What could go wrong? Fire it up!"

You highlight the weakness in MtGox and then by analogy suggest the whole platform is similary amateuristic.

the bitcoin protocol itself is solid, tested and has some very smart people working on it.

Isn't the idea of a central exchange (which "Magic the Gathering online exchange" is), where most of the trading occurs, a little antithetical to notion of a distributed currency?

Absolutely, but a true p2p, decentralized Bitcoin exchange service has yet to happen.

Paper IOUs seem almost ideal, if we didn't have to rely on a central issuing authority to make them universally accepted. IOUs are easy to replicate with blind signatures (see Section 4, Examples: http://web.mit.edu/6.857/OldStuff/Fall02/handouts/L15-voting... ) but noone wants to hunt down as many people as they have bills, just to trade the individually-issued digital IOUs... This is what Ripple ( http://ripple.com ) claims to do, though: act as a distributed clearing house to resolve and cancel debt via a social graph. I haven't had a chance to actually use Ripple though, so I am certainly not recommending it specifically, just observing that it seems to address the right problem - that money is just debt - whereas Bitcoin simply creates a finite resource to be fought over.

The problem is moving the counter currency around. I can zap you some bitcoins without any problems, but it's much harder for you to zap me back some USD (or EUR, JPY, whatever).

Fair, the ForEx role will never go away, but it seems that MtGox has also become a trading platform for the Bitcoins themselves, with the obvious downside that it is susceptible to DDoS' like any centralized service. It makes more sense for a P2P currency that this simply be done at the protocol level with at most a custom client, but one that is still a full peer. The problem I, and others, are seeing is that the blockchain is becoming huge and will only get bigger, meaning only those with the resources to maintain it will participate as full peers - hardly ideal for a distributed system since it will inevitably require a certain amount of centralization to scale.

Fair, the ForEx role will never go away, but it seems that MtGox has also become a trading platform for the Bitcoins themselves

Wait, whats the difference between those two things? Converting bitcoins into anything else _is_ ForEx.

MtGox has become something of a speculative trading platform. While direct exchange (cash in, cash out) will always be necessary (unless you are paid in Bitcoins and live off of Bitcoins), I don't think MtGox is mostly being used that way. Rather, MtGox is becoming synonymous with Bitcoins for traders, and people want to use it like a bank (with all the problems that entails, and Bitcoin was partially hoped to relieve). As such, Bitcoin is only as reliable as MtGox. If People's main interaction with MtGox was buying Bitcoins for use on real goods and services, then I would agree that it is just another exchange, but people aren't using it like that. People buy BTC with USD one day, and buy CAD with BTC the next, with every intention of going back through the exchange and the Bitcoin currency again in order to cash out. MtGox has an API, and in fact, you are right, that is precisely what a ForEx is used for... I'm not knocking it, only pointing out that many of the problems people want Bitcoin to solve are being recreated by re-imposing conventional institutional roles. As long as there is more than one currency, people will need to exchange them. Who am I to say Bitcoin "shouldn't" have a trading platform other than to note that, IMHO, anyone who uses MtGox speculatively is relying on it's centralization, with all that that entails. I think it's ultimately just a matter of scale.

People buy BTC with USD one day, and buy CAD with BTC the next, with every intention of going back through the exchange and the Bitcoin currency again in order to cash out.

Are you sure they're doing that?? Bitcoins volatility would render it pointless for Forex trading. (edit: by which I mean trading between two old school currencies via bitcoin)

I am taking it for granted that speculative trading is a pretty common, if not the main activity on MtGox:

- "Bitcoin day trading": https://bitcointalk.org/index.php?topic=57169.0

- "Mt. Gox Temporarily Halts Bitcoin Trading to Allow the Market to Calm the Hell Down": http://betabeat.com/2013/04/mt-gox-halts-trading-temporarily...

But I could be wrong.

Oh yeah, people are swapping BitCoin for (old school) currency Y and then back again like crazy.

I thought you meant people using Bitcoin to do arbitage between (old school) currencies X and Y

I agree that speculation is harming bitcoin. It needs less volatility to be truely useful. But then when a genuinely innovative currency/technology suddenly gets a ton of press, I don't see how a spike could be avoided.

"I thought you meant people using Bitcoin to do arbitage between (old school) currencies X and Y"

That's precisely what I suspect (and can only guess at this point) that it is being used for:

http://bitcoinmagazine.com/btc-trader-bitcoin-arbitrage-made...

http://lichtman.ca/is-bitcoin-arbitrage-feasible/

https://github.com/michaelcdillon/BitCoin-Arbitrage

At the same time, MtGox also serves as an entrypoint for non-speculating users and has effectively become the "Bitcoin bank." It's voluntary - people are choosing to use MtGox, but in this regard MtGox's success seems to be a Bitcoiner's weakness: a DDoS against MtGox can cut the value of a BTC (and their lunch money) in half.

MtGox isn't the only exchange, just the largest.

> Isn't the idea of a central exchange (which "Magic the Gathering online exchange" is), where most of the trading occurs, a little antithetical to notion of a distributed currency?

No, its a demonstration of network effects and natural monopoly in an unregulated market.

It seems they were not aware that AWS exists. They should really consider hiring engineers specialized in scaling.

The more fundamental problem is the lack of separation between their trading engine and their website. While a website would be perfectly fine on AWS (and likely the more significant part of their scaling problems), the trading engine belongs on a very tightly controlled platform.

I don't have any confidence in their platform. It shouldn't take more than a few domain experts to build a better platform...but for some reason its not happening.

> I don't have any confidence in their platform.

Neither should anyone, as evidenced by their track record. It's just sad that it is this platform driving the price of the Bitcoin at the moment, and I'm not sure how it speaks about the future of the currency.

The web frontend commands the trading platform, and the BTC/money outflow. It would be irresponsible for them to host it in an environment they dont control.

To be fair, it's my understanding that EC2 are virtual machines. If they use EC2, they need to be 100% certain that their servers are safe and don't fall victim to the Bitcoinica Linode attack. http://bitcoinmagazine.com/the-bitcoinica-linode-theft-and-w...

A full month to provision new servers seems awfully long. Even if you don't want to go full cloud (which could make sense, I'm not judging that), you can easily get dedicated servers installed and configured in less than a day.

Well, not if you want to do self-hosting of your own equipment.

Turnaround time from Dell is minimum two weeks. Add weeks as the servers get bigger and more custom. Also, most DCs need time to set up racks, power and connectivity.

I think if any of the important parts operated on AWS, governments would be quick to take advantage of that to shut it down ("it's used int eh drugztrade~!").

I don't think that using AWS would help them that much. They need to keep very secure (know any banks that use AWS?), and the main trading engine can't be parallelised anyway (because of order sequence and the blockchain).

(Also: Amazon would chuck them out as soon as the first senator reached for their phone, as with wikileaks.)

Trading on Mt Gox has absolutely nothing to do with the bitcoin blockchain.

Ah, I didn't realise that. so they're independent of the bitcoin address system? I had read that their trading engine can't be parallelised and assumed the blockchain was the reason. I stand corrected.

edit: this guy explains it: https://news.ycombinator.com/item?id=5539387

Yeah, because AWS = instant scaling.

This sort of operation needs physical hardware in their own secure location. This isn't some mobile social webapp.

One of the problems here is that an exchange buying and selling one thing doesn't benefit much from scale. The core of a matching engine relies on knowing everything that happened before it, i.e. every trade submitted which has not been matched or cancelled. This pushes the matching process down to a single thread on a single machine.

I am building a matching engine now (for pleasure) at

https://github.com/fmstephe/matching_engine

(Having said all that I have no idea where their performance problems are coming from - the matching engine may not be the culprit)

wow. After looking at all of the issues and the lack of answers from MtGox..I don't think I'm going to be investing any money in this exchange any time soon.

No kidding - they appear to be focusing on answering question #1 and completely avoiding #2 & #3. Does not instill confidence.

You guys need to understand that these guys are super Japanese, complete with the culture of never wanting to show dishonor. If they were American I would be much more suspicious, but they have consistently fixed the problems that they have had and are desperate to hire western developers to move to Japan to help the not only build the exchange, but expand the community's trust in them.

"You guys need to understand that these guys are super Japanese, complete with the culture of never wanting to show dishonor."

Then why have a Reddit Q&A about their failures?

Huge opportunity here for a proper exchange / trading platform to be built. There is obviously a demand for it.

1) Regulatory issues are a big question mark.

2) There isn't actually that much volume compared real electronic exchanges. Just looking at the Mt Gox page from today it looks like $30MM USD worth of bitcoins have been traded. Hard to make any real money off of that.

If it was a sure bet that doing it would make you rich tomorrow, don't you think there would be scores of people racing to get it done? It's hard to find fruit hanging this low, if coinlab doesn't get it, I have a feeling someone will.

> 2) There isn't actually that much volume compared real electronic exchanges. Just looking at the Mt Gox page from today it looks like $30MM USD worth of bitcoins have been traded. Hard to make any real money off of that.

While it might have much less volume compared to a "real" exchange, MtGox fees are 0.6% of each transaction. $180k daily revenue is certainly "real money" for a startup.

Proper trading platforms (TradeHill, CampBX, etc.) were built and then no one used them.

CampBX web interface is terrible, but I admit, ought to be a competitor since they at least appear to be run better than gox. TradeHill's first act ended badly. In Act II they have a min 10k deposit requirement.

I keep waiting for interesting news about coinlab's taking over mtgox, but so far, not much.

The best piece of all comes from their answer about new servers.

> Upgrading computer systems means ordering more servers (2 weeks timeframe), setting up (1 day), load testing (2 weeks) and deployment (1 day). It's a process that can take up to one month in total.

You've got to bear in mind that they're talking to reddit. I think thats a fair description of why they can't upgrade at the drop of a hat. As they say elsewhere in the AMA:

We are big in the bitcoin world, but compared to a Facebook or a Google or even a bank we are too small and don't have access to their technology

What would you have them do?

> Cloud solutions are not meant for large scale operations such as trading systems

Arguably, hosted cloud solutions aren't ideal for large scalable operations, if you define "large scale" as in "someone with the scale of Amazon, Google, etc. that can afford to run a private cloud."

But MtGox makes it pretty clear that they aren't on that scale. Their argument seems to be simultaneously that they are both too "large scale" for hosted cloud solutions and too small scale to be able to effectively address the challenges they face in-house, which, if we accept it, seems to indicate that they are at a scale that cannot effectively operate.

Right. They couldn't effectively operate in the face of a huge trading spike and a DDOS. Hence they're upgrading. But then they get stick for taking too long to upgrade.

Conversely, if they spent $$$ on a bulletproof platform, they'd probably go bust (as seems to have happened to other more technically proficient exchanges)

The underlying problem is: people have too high expectations of what the bitcoin world can cobble together at this point. Its being built by hackers, not bankers.

> Conversely, if they spent $$$ on a bulletproof platform, they'd probably go bust (as seems to have happened to other more technically proficient exchanges)

If that's true, it would seem to indicate that no one with the resources to eat the cost of operating a reliable exchange through the expected growth period of bitcoin has sufficient long-term confidence in bitcoin to underwrite the growth-phase costs in order to reap the rewards once bitcoin is all grown up.

> The underlying problem is: people have too high expectations of what the bitcoin world can cobble together at this point. Its being built by hackers, not bankers.

Expecting the main exchange for something that is promoted as a currency to meet the reliability expectations of a currency exchange is not unreasonable.

You're assuming that access to investment (VC funding or whatever) is a perfectly efficient system. It clearly isn't.

As far as I can make out from reading #bitcoin, their whole service runs on a single server currently. The very notion that they could be too big for any hosting provider is hilarious.

If they tried to use AWS amazon would drop them like a stone. I think they also are deliberately keeping their hardware in house for security.

It always bothered me to know that there is no other solution to DDos attacks other than shutting down the systems and wait till it gets back to normal load.

Reply on news.ycombinator.com