If you want the
Cliff's Notes version of Bandwidth versus Latency, think of 
a circuit as being a bridge
(the kind you drive your car across to get over water). Bandwidth
would be the amount of lanes on the bridge, which would determine how
many cars can get across during rush hour traffic. Latency would then
be the speed the cars can get across that bridge. In this circuit
bridge, though, all the cars go roughly the same speed and are limited
by the speed of light. So if you want to download a file fast, you
want the bridge to be really wide rather than one lane. This would
give your bridge a lot of thruput, and it would have high
bandwidth. The latency would then be determined by how long your
bridge was, and how fast cars could get onto the bridge. 


Latency can slow
things down, of course, but even if you see RTTs of, say, 600ms, which
is pretty high, that's still only an additional slowdown of .6 seconds
between the time a file is requested until it is sent. When you have
packet loss that packets need to be resent. So if you are downloading
a big file with 30% packet loss to some place, 30% of the packets
would actually need to be resent.