NS2 program specifications

Hello all,

I am using the Network Simulator NS2 program
(https://www.isi.edu/nsnam/ns/) in
order to simulate a simple network topology with 1 UDP agent and 1 CBR
Application Traffic sending data from a node over to another node
further away.
Somewhere in the middle of the topology, there is a choice of network path
involved since there are two alternative options (either e.g. 1-2-3 or
1-4-3
where node 1 is sending data and 3 receiving). I am also using the
following
commands for dynamic routing:

Agent/rtProto/Direct set preference_ 200
$ns rtproto DV

I also use an rtmodel stop/start event in order to simulate the following
network topology change: connection of nodes 1 and 2 is "cut off" and then
reinstated later. My questions are as follows:

1) How is the time that exists between the path change after the 1-2
connection
drop calculated? I can see some 11-byte rtProtoDV packets being transmitted
through the network but I don't exactly understand how the time between the
connection drop and the packet-delivery network path change is calculated.

2) Which one of all the nodes defines where the packets will be
directed? So,
does the traffic source define the whole path that one of its sent
packets will
follow, or do the intermittent nodes decide seperately where to forward the
packets ad-hoc?

3) What does the Agent/rtProto/Direct set preference_ 200 command
exactly do? I
have read on the NS2 manual that the default value is 100, but what does it
exactly do differently when it is 100 vs. when it is 200? The NS2 manual
isn't
clear at all at that point...
(https://www.isi.edu/nsnam/ns/doc/node311.html)

Thank you in advance,
Jason