Topic: Mtgox statement on facebook (Read 43402 times)

The market carried on without gox. When it reopened the adjustment was in line with the other exchanges - a good sign?

It is not in Mtgox interest to stop a sell off.  They make a more in a few hours of panic selling then they do in weeks of regular trading.

You really think they would try to stop a sell off? 

Their infer-structure simply cannot handle the demand.  Plain and simple.

No, the panic selling began well before the lag. This is an absurd statement. "As expected"?? No, I do not expect that new accounts on a trading platform would devastate a currency. And I don't believe it did. They use this lag as an excuse for any sell-off. We were going down well before this sell-off started. The lag actually slowed the sell-off down.

IMO This is an absurd pump....basically saying 'Don't worry, we have tons of new buyers coming'. Can you imagine a broker in any other asset class making a statement like this?

In another statement during the previous minor sell-off (not from yesterday but April 3rd) -- 'I understand that many of you have a lot at stake here, but remember that Bitcoin, despite being designed to have its value increase over time, will always be the victim of people trying to abuse the system, or even the value of Bitcoin decreasing occasionally. These are not new phenomena and have been present since the beginning of time when humans first started trading'.

Can you imagine a brokerage in the US being able to say that the underlying is 'designed to increase in value' or that it's value 'decreases occasionally'? They couldn't because it would be illegal.

Twitter: 'Trading is suspended until 2013-04-12 02:00am UTC for market cooldown. Once back trading will be also faster.'

And now we're halted. I question their timing. We were in a slow sell-off at a key psychological support level (yesterday's low). We were not in a panic sell today, so why halt? They could have halted yesterday. Yes, they need to upgrade their servers....but I think it's interesting that they chose a key psychological support level to halt and not when it was trading higher today.

The drop and the publicity about it could also dampen the public interest.

+1

Seems like they may have a bottleneck somewhere that adding servers won't fix. High probability that it's in the database, which is pretty hard to scale by just adding servers. But i don't see how the amount of traffic they get should be this much of a problem for them. They should be  able buy a powerful enough database server with one or two day's worth of the profits they make (assuming the $100K+ per day profit I saw posted in another thread is true). I've worked on websites that got 1M+ hits per day in traffic spikes, not quite the same type of app as an online exchange, but I know a little bit about performance and scaling.

For the amount of money they make, and the amount of traffic they do, there is simply no excuse. Their public statements don't instill confidence. First they said "you guys have no idea what it's like, come work with us for a day, we can handle DoS attacks, where withstood two at the same time without any lag", then the next day they are down after the maintenance because of a DoS attack.

Let me take a guess... PHP and MySQL??

There are some excellent alternatives to Mt. Gox that already exist: bitstamp, campbx, bitfloor.  But most of you all won't switch or strongly suggest that your friends/family go to one of these alternatives.  Why in the world would we with technical know-how want to spend time building yet another exchange when it's clear most people just go with the best-known name.

elastic computing service might be too expensive for mt.gox?

As much as I think Gox has some very serious issues and is currently unfit for purpose (clearly they have some horrible legacy and crazy architecture to content with - e.g. I heard someone mention the web front is not decoupled from the trading engine - WTF?!) it just not that simple.

Anybody who knows about how to build fast highly secure exchanges that have to guarantee perfect consistency with regards to state/transactions would think twice about using any sort of public virtualised infrastructure. Physical hardware is still the norm in this domain because raw single threaded clock speed and fast IO (e.g. solid state drives) often matters a great deal for exchanges and trading platforms. To give you to example I have direct knowledge of: I worked at Betfair in the past and they were considering building bespoke hardware chips to run the code of their core exchange engine to ensure they could scale their platform for the future and I know a friend at Merrill Lynch who told me about their servers which ran in oil baths to enable over-clocking raw CPU speed matters. There are many other example of custom hardware in use by banks and the like. To think anyone would use Heroku for an exchange is laughable. And I speak as someone who choose to build an app on Heroku or any other PaaS/IaaS provider 99% of the time. And I didn't even mention security.

What I do hold Gox accountable for though is not investing in their platform a long time ago. I can't believe they haven't had the money to build an entirely new trading engine and hire a really good team to build and support it. They have been very short-sighted. Any good start-up has a plan in place to scale if they start to experience exponential growth.

I think we should be using some of our Bitcoins to fund the building worthy alternative - I certainty don't hold much faith in any of the others so far. An exchange for bitcoiners, built by bitcoiners.

Does anyone know of a good exchange? BitStamp lagged like hell as well, and I heard it's not the only one. (well, even the forum and bitcoincharts did)

Really, we read every day about the wonders offered to today's computing by ever progressing cloud/virtualization/whatnot technologies -and the only ones that know not how to properly use them are the ones managing our money?

They should get rid of the tiny trades or charge more for anything less than a certain limit.

Thousands of tiny orders are not required anymore and they're annoying.

I am sure mtgox earning a decent amount of money since the bitcoin price rise. They should spend them all to upgrade the infrastructure and trading platform to prevent any problem. Bitcoin trades 24/7 and requires a high end reliable system. I was very disappointed with the service yesterday

What's it going to take before the Bitcoin community gets together and creates a viable decentralized exchange to replace Mtgox?

How loud do we need to scream it!  Get a decentralized exchange platform going.  Please, please, those with the technical know-how, lets get the discussion started.

LOL, I think that Maged surely know this stuff  But anyway did you heard of Linode?

Agreed. It's good for Bitcoin.

The micro trading bots spamming are interesting. Very important not to react like the USSA and ban such things but to react to them in a fluid market response. Established markets also have this problem of spamming - I expect this community to be better than those trillion pound markets in the response.

If it's 20k accounts a day then we have to presume that KYC is outsourced to a specialist company for this? Or is it in someone's garage handling passport scans and bank statements?

I hope a lot of Mtgox users saw this FB message and will move to another exchange... more balanced traffic on different exchanges would be healthier for Bitcoin.

You don't have to get verefied to sign up and play with small amounts. 20K new users would be possible, I'm sure that verification que will skyrocket soon.

Have you considered the glut of orders is preventing execution, thus our noticing it in the first place? I'd say that's the root of the whole problem. If they had all their trades going through we wouldn't be worried.

If you look at their trade feed, there are barely a few trades per minute in times of high lag. I still don't understand how they can capitulate under what is essentially so little volume.

The engine isn't even capable of >one trade per second most times, Nasdaq does at least 35,000 per second and is tested up to 60,000.

http://www.nasdaq.com/services/homw.stm

The order of magnitude difference doesn't make sense to me. Sure, some is to be expected, but not that much.