Black can reply at a, making ko. The ko is complicated, but does not seem to gain anything on average, by comparison with simply filling. The worst that Black threatens is to take , yielding the same result as if Black played first. The only way Black could gain, on average, would be if , taking the ko, did not threaten much. But it does.
If Black does make the ko with Black a, for White to connect at is usually inferior. Normally it is better to play elsewhere instead.
Now let's look at the follow-ups.
Now we can make a game tree:
A / \ / \ 5 B / \ 2 0
A stands for the original position, B stands for the position after White plays.
Similarly, we reckon that Black has 3 points of potential territory at A, and that a play at A is worth 2 points.
(Note: B is actually ambiguous. Instead of taking gote White can simply play at 3 with sente. Well, sort of sente. The position after White 3 is also ambiguous. None of that affects the value of B, or A.)
-- Bill Spight