First of all, the events are very, very different.
Let me speak to the event of July 8. The way to characterize it is that there was no way for the engineer at the time.... When it had been performed five times prior with no incident whatsoever, there was no belief, no information at the time, that there was going to be any issue.
What happened was that when the code change was executed and the filter removed, the behaviour of the equipment, in the way it's designed between one vendor and a second vendor, was very different. That was the unknown at the time: the behaviour of one device, one manufacturer, who executes a standard one way, versus another manufacturer. It was that sequence that caused essentially the event to then flood the core network.