Bitcoin Forum

Lets recap:

During the rally we've seen short selloffs which were all accompanied by lag.
Lag is an internal mechanism of the mtgox infrastructure and not subject to network latency.
Neither the "lots of small orders" nor the "insufficiently funded orders" explanation makes sense.

The obvious conclusion: They hired inept coders which constructed a system which doesn't scale and covering it up.
This is core infrastructure which has become sustainable, a fundamental reason.

The worst part of this is market intervention by scapegoating, discontinuation of service and propping up of the volume.

I would also attribute it to incompetence. It's quite simple, actually.

If it was DDoS and the problem was with the network, the API would return ZERO as the API lag value, as the API should be unaware of any network issues.

As soon as their API returns a lag value, it's the trading engine itself that can't handle it, irrespective of the network connectivity.

That's the reason why even the best DDoS protection company in the world won't change a thing.

Whatever it is, this market will not stabilize until some better, more reliable exchanges open up.

An exchange which offers an API is subject to flooding attacks either through malice or incompetence.

Automated trades should be forced to operate through a QOS mechanism which limits both individual activity, and the aggregate automated action to a small percentage of total exchange capacity.  A fee for API driven trades would also be an effective mechanism to moderate the load caused by automated trading systems.

More importantly, until no one exchange can basically exacerbate a panic just by incompetently failing when it is most needed.  Any exchange with 80% of volume needs cut down to size.

I maintain that streaming a HD video and decoding it takes more resources than operating the entirety of mtgox on a single machine. Processing an order should be a trivial operation.