how is .05 more than .2, something is definitely off here.
The correct way to read the graph is that they're equal within the margin of error. The effectiveness is the same, but you get small variations due to small number statistics. If one person rolls two dice 100 times, they might get "7" 18 times, and the other person 16 times, but they're really rolling the same dice and playing the same odds.
Really, 0.05, 0.20, 1.00, and 5.00 mg all produce an equal effect on DHT levels in the scalp, the finasteride response curve is completely flat. Your scalp is thus
saturated in finasteride for all doses of 0.05 mg or greater. However, at 0.01 mg the effectiveness is much lower than at 0.05 mg. Note that they only measured the response at 0.01 and 0.05 mg but not at 0.02, 0.03, and 0.04 mg. We don't know where the declining response happens.
In contrast, the blood DHT level does look like it falls off between 0.05 and 0.20 mg, the former gives you a 50% reduction and the latter a 70% reduction. If you assume that blood DHT levels are what causes side effects, then to decrease your side effects you need to go below 0.20 mg/daily. But again, they only measure 0.05 and 0.20 mg, they don't measure any of the numbers in between and thus we don't know where the flat response curve begins.
If we make the (rational) assumption that the best dose of finasteride is whatever dose maintains a ~70% reduction in DHT but minimizes the offset in blood DHT, then that is some dose between 0.01 and 0.05 mg daily.
It is impossible to cut a pill this small, your only solution would be a solution, pun intended

You'd have to crush the finasteride tablets into fine powder, and dissolve them in everclear, and drink something like a teaspoon a day.