User blog comment:LittlePeng9/Color reducing for Turing machines/@comment-10429372-20140426114729/@comment-1605058-20140426134035

Both improvements are correct.

As for the second one, I didn't take it into consideration, as I was trying to make a universal way of reduction, without wondering what happens at any exact values. One can also argue that for 2 colors we have to do nothing, yet my reduction still applies.