Dapper425

I believe Elan does a great job in explaining this concept.
In this particular problem, your
H0 : u >=50
and just think that you desperately want to reject this SOB hypothesis so that you can proudly claim that the mean of the population might be less that 50 in reality. Your only hope is the critical region where it gets rejected. At a 10% level of significance( this is one sided) ,
10% of the values of the standard normal distribution falls onto the critical region. You look in the table to find out the z value whose probability matching .01 (means 10% of the values are in the right tail thats how the table is constructed). That happens to be 1.285
Try drawing a normal distribution and shading out the lousy region where you want the rejection to happen if the test statistic falls there. That will help.