and we're doing it with a small fraction of the original headcount. you mentioned outages there. there have been several, and we have actually spoken to an engineer who works at twitter, and they said that the plumbing is broken here and it is on fire and there could be problems at any minute. do you accept that? i mean, there have been a few outages, but not for very long. and it's currently working fine. it doesn't keep you up at night that twitter might go offline again? at this point we've got a pretty good handle on what makes twitter work, and we are also doing it with two data centres instead of three. so we used to have three data centres, we shut down one of them. we are actually at two thirds of the prior compute capability. we've made so many improvements to the core algorithm and in some cases we've improved the core algorithm by 80%, so the actual cpu