It's a biggest problem that there is a routing inconsistency which occurs with any distance vector protocol in which the update messages carry only pairs of destination network and distance to that network. Basically the router R1 has a direction connection to the network 1 so that there is a route in which the table has a distance 1 which will be included in the periodic broadcasts, Router R2 has learnt from router R1, installed the route in its routing table which advertises the route at the distance 2. Finally, the R3 has learnt the route from the R2 and advertises it at the distance 3. Now, suppose the router R1 connection to the network fails then R1 will update its routing table for making the distance to the infinity. In the next broadcast R1 will report the higher cost route.
Unless and until the protocol includes the extra mechanisms to prevent the other router which broadcasts its routes before R1, it will affect the R1 status. In particular, if the R1 receives the routing message from R2 the route is advertised on the route R2 and hence the list which reaches the network earlier. At this point, if either R1 or R2 receives a datagram destined for the network 1, then they will route the datagram back and forth until the datagram's time to live counter expires.
In the next round of routing exchanges the R1 broadcasts the routing table entries. So, when it learns that R1 route to the network 1 has the distance when its routing table entries are written. When it learns that R1S route to the network 1 has the distance 3, R3 calculates the new distance for its route making it to the 4. In the third round, R1 receives 3, R2 calculates a new distance for its route making it to 4. In the third round, R1 receives a report from the R2 which includes the increased distance. And then increase in the distance in its table to 5. Hence the  two hosting routers which continue to RIP to infinity.