C++ random_shuffle vs rand

I have been trying to solve the problem http://ift.tt/2moTuFS and logic is same as Mastermind (board game). For the implementation I need some random position/index. For that I first tried

srand ( time ( NULL ) ) ; 

with

init = store [ rand () % store.size () ] ;  // store is the vector containing values

but this did not pass all the test cases ( I made multiple submissions to ensure that this is not due to chance ). Then tried

srand ( 2333333 ) ;

with

random_shuffle ( store.begin () , store.end () ) ;
init = store [ 0 ] ;

passed all the testcases ( for this also I made multiple submissions to ensure that this is not due to chance ). Then also tried

srand ( 2333333 ) ; 

with

init = store [ rand () % store.size () ] ;  // store is the vector containing values

but this also did not pass all the test cases ( I made multiple submissions to ensure that this is not due to chance ). Then tried

srand ( time ( NULL ) ) ; 

with

random_shuffle ( store.begin () , store.end () ) ;
init = store [ 0 ] ;

but this also did not pass all the test cases ( I made multiple submissions to ensure that this is not due to chance ).

Link for the submission/code : http://ift.tt/2n1waee

So, what is that special in second case ( with seed 2333333 and random_shuffle ). Is there any reason for getting desired results ? Also, I thought srand ( time ( NULL ) ) is one of the best choice but why not in this case ? Using random_shuffle is better than using rand ?

Update : Also most of the other accepted submissions also used srand ( 2333333 ) not srand ( time ( NULL ) ) ;