Hey
Why would you think so? 
0.5 would be the natural threshold if we use a sigmoid function (as in the following chapter) which pushes all predictions between 0 and 1 and lets us interprete the values as probabilites. I tried to give an explanation how the with-sigmoid-threshold 0.5 and the without-sigmoid-thrshold 0.0 are related here.
Hope that helps you make sense of it.
Besides that please have a look at the Forum etiquette regarding @ mentioning people in the forums.