Mt Gox.....ineptness or competence (controlled lag).

[Manticore]

Does anyone else find it interesting that these large lags only happen on the way down?? AFAIK (correct me if I'm wrong) recently lags never occur on higher volume during upward momentum. The market looked as though it was about to drop a few seconds ago and PRESTO, the lag magically appears. Lately, why does the lag always appear after the price starts dropping (not before) and never when the price is rising on even higher volume??

I'm probably alone here, but I'm not convinced by the Mt. Gox DDoS attack excuse that supposedly drove down prices yesterday. Sure, there may have been a DDoS attack. But I've watched the last two Gox malfunctions fairly closely and the only things I saw, in both cases, were the beginnings of natural, orderly sell-offs (although the previous instance on March 28th was more dramatic) followed by a trading lag (DDoS attack?) and cancellations that actually stemmed the sell-off.

If these 'DDoS attacks' were meant to drive down the price substantially, clearly it's not working. It looks more like planned lags that act as trading curbs. Trading curbs are meant to halt sell-offs, giving everyone time to regain a clearer head and trade with less emotion. These 'DDoS attacks' have become the perfect excuse on which to blame any natural sell-off and subsequent trade curbing mechanism, IMO. The problem is that these last two lags occurred only after the sell-offs began, not before.

Pair this with the bizarre pricing action on Mt Gox, where minuscule sub-coin buys above the ask move the bid/ask up several dollars and counteract 100's of coins being dumped......no, I think Mt Gox is doing everyone a favor in doing everything it can to keep the price up. I have no evidence (only empirical); this is unfounded conjecture.....

Of course, there is an argument that a mysterious, nefarious group of Bitcoin bandits who hold enough coins to create simulated sell-offs are then DDoSing the system in an effort to buy cheaper coins (at risk of destroying the value of the coins they hold). And Gox is so competent that they have caught and resolved the situation before they became larger problems in both instances. But, I simply did not see anything remotely resembling this yesterday. On March 28th? Perhaps.

Gox has a vested interested in keeping this propped at the expense of angering current customers.....because they have many new customers with which to replace you all. Either they have an ridiculously inept system or they are extremely competent at quelling panic selling. I'm not sure if they are idiots or geniuses.....

[jimbobway]

I'm glad OP put this in the Speculation section because all of it is speculation.

Here is my speculation:  I think Mt. Gox is doing its best to ward of the DDoS attacks.  They even hired a firm to handle it but even the firm could not stop it 100%.

There is big money involved now so people will do anything, include DDoS attacks to bring the price down, and as Mt. Gox said they can only do so much but they cannot stop DDoS attacks 100%.

Most exchanges have been hacked in one point in time or another.  CampBX has not been hacked, I don't think.

Bitfloor, Mt. Gox, bitcoin-central, they've all been hacked. Pick your poison.

The solution is P2P exchanges.  With litecoin, ripple, bitcoin etc. some can make a P2P exchange a reality.

[Manticore]

I'm glad OP put this in the Speculation section because all of it is speculation.

Here is my speculation:  I think Mt. Gox is doing its best to ward of the DDoS attacks.  They even hired a firm to handle it but even the firm could not stop it 100%.

There is big money involved now so people will do anything, include DDoS attacks to bring the price down, and as Mt. Gox said they can only do so much but they cannot stop DDoS attacks 100%.

Most exchanges have been hacked in one point in time or another.  CampBX has not been hacked, I don't think.

Bitfloor, Mt. Gox, bitcoin-central, they've all been hacked. Pick your poison.

The solution is P2P exchanges.  With litecoin, ripple, bitcoin etc. some can make a P2P exchange a reality.

Yes, 'unfounded conjecture' as I stated. Oh, I did not dispute the fact that there are attacks on any of the above entities. I dispute that these attacks are causing price drops. The price began rolling over and dropping before either attack. And these lags, while understandably unavoidable, seem one-sided. I've only noticed bizarre, mismatched pricing action on the downside. I'm not debating the existence of DDoS attacks. Even if they exist (certainly), these lags (DDoS, Gox induced, or combination of both) serve to muffle panic selling. Lags are doing quite the opposite of what everyone here is thinking IMO. We were overbought (among other things), due for a pullback. Everyone is blaming the DDoS attack on the drop when it was largely due to technicals. We would probably have dropped harder if it weren't for lag (DDoS or otherwise).

[Manticore]

If these 'DDoS attacks' were meant to drive down the price substantially, clearly it's not working. It looks more like planned lags that act as trading curbs. Trading curbs are meant to halt sell-offs, giving everyone time to regain a clearer head and trade with less emotion. These 'DDoS attacks' have become the perfect excuse on which to blame any natural sell-off and subsequent trade curbing mechanism, IMO. The problem is that these last two lags occurred only after the sell-offs began, not before.

Pair this with the bizarre pricing action on Mt Gox, where minuscule sub-coin buys above the ask move the bid/ask up several dollars and counteract 100's of coins being dumped......no, I think Mt Gox is doing everyone a favor in doing everything it can to keep the price up. I have no evidence (only empirical); this is unfounded conjecture.....

It could also be an effect of the inherent friction of the system, it just tends to bump its head every time there is an overbought condition because everyone is trying to do everything at once. I would rather credit the height to which the market has reached to improvement in Gox's systems to handle the exponentially increasing workflow. But yes. The lags are extremely frustrating.

It is interesting though, to compare the current price action to the price action surrounding the $32 peak. Uncanny, in fact.

Yes, I've noticed the similarity . Nice money flow analysis, btw.

[kiko]

Just spitballin' here, but it could also be algorithmic.

Orders are stored in a btree, many small bids flow in and few large asks. The tree gets unbalanced, so during sell-offs. The engine has to recurse much deeper than desired to match trades.

PHP object creation/destruction bottlenecks the whole thing.

[dg2010]

I'm pretty sure MTGox are controlling things. Just my personal feeling.

[mp420]

What I find astonishing is that they say in their statement:

There are a few things that we can implement to help fight the attacks, such as disconnecting the trade engine backend from the Internet. By separating the data center from the Mt.Gox website, we will continue to be able to trade.

As MtLagz isn't really a small garage business anymore, I find it quite worrying that they hadn't done this already.

[b!z]

I'm sure the lag is unintentional, as it usually only happens when a large amount of people are attempting to trade.

[hashman]

Keep a log or two of https fetch request times w/ spot price for us.

 

[Beta-coiner1]

I'm sure the lag is unintentional, as it usually only happens when a large amount of people are attempting to trade.

I'm thinking the same,the same lag that was on the Gox was on this very website in the earlier part of rally,in ~Feb.

[MikeH]

I'm curious how Mt Gox is setup, is there a separate server for handling API requests and chart data so that the database isn't constantly hammered?

[zoinky]

I'm curious how Mt Gox is setup, is there a separate server for handling API requests and chart data so that the database isn't constantly hammered?

Try tracerouting them, believe they show different servers.

[oakpacific]

I have done enough observation of the Gox lag, and I think I can confidently say that the lag occured both during upmove and downmove.