With regards to call waits it is not exactly rocket science but merely simple call vectoring.
No doubt like many call centres NTL use CMS (Call Monitoring Software) which will provide them stats for their peak calling periods.
Any such switch can be modified at a moments notice to change the way calls route.
It can be configured instantly to for example if calls waiting are equal or greater than 20 then give engaged signal.
If a major service outage occurs then it will play an instant announcement of the fault and then continue into the vectoring system of if persistent call are greater than say 20 then disconnect calls.
The variations are steps and can pre defined manually by usually the controller are the there are endless configurations of these.
One old classic example from NTL would be in the days of Tech Support in Newport when they abandoned the night shift and moved it to the Swansea Call Centre - what would happen at 12 midnight is any calls waiting would be dropped while they routed the calls to Swansea.
The vector for this would be to allow say for example unlimited call to get into the queue but come 12 midnight the vector would be to drop all current calls and then reroute all new calls to Swansea.
So at 11.59pm there could be 50 callers waiting for however amount of time but come 12 midnight they would be disconnected and when they redialled they would be routed to Swansea etc.
Totally unacceptable from the customers point but that is how it used to happen and no doubt with clever configuration still does.
