Your result $7/10\cdot6/9\cdot3/8\cdot2/7=1/20$ is the probability to get the four bills in a specific order. But there are $4!/(2!2!)=6$ ways to pick the four bills, so the probability is $6/20=0.3$.
EDIT
If you want to get your result right in terms of probability, in my opinion the best way to avoid errors is that of using a probability tree:

To address your other questions about the right way to count the number of combinations, the rule p=(favorable cases)/(total cases) works only if all the cases one considers are actually a priori equally likely to happen. For the problem at hand John selects at random one out of the possible $10\cdot9\cdot8\cdot7=5040$ ways to extract four bills (just think of the bills as numbered), so these are our elementary events as they are all alike. Of these, only $(7\cdot6)\cdot(3\cdot2)\cdot6=1512$ will contain exactly two counterfeit bills, so the probability we are after is $1512/5040=0.3$.
Notice that I counted all the possible outcomes considering them different even if they contain the same bills in a different order. As favorable cases do not differ for the order of bills, we may as well consider two outcomes to be the same if they contain the same bills in a different order. In that case, John selects at random one out of the possible $\binom{10}{4}=210$ ways to extract four bills and of these only $\binom{7}{2}\binom{3}{2}=63$ will contain exactly two counterfeit bills. Needless to say, the probability is still $63/210=0.3$.