2: kmullin (1100) 2017-01-17 17:08 (Edited at 2017-01-20 07:06)

I've wondered about this myself before. So, just to elaborate on lordjackes's point, i think the tactic is to get White's lance [香車] on 1c and White's pawn [歩] on 1d so that the 1d-pawn blocks the lance from moving up further. This configuration allows Black to attack the lance on 1c with their right knight [桂馬].

So: 
☗P-1f [１六歩] ... ☗P-1e [１五歩] ... ☗P-3f [３六歩] ... ☗N-3g [３七桂] ... ☗P-1d [１四歩], 
then (assuming starting with ☖P1c [１三歩]): 
☖Px1d [１四歩] ☗P*1c [１三歩] ☖Lx1c [１三香] ☗N-2e [２五桂]
(And, of course, even if you don't attack 1c with the knight, there's still lordjackes's point about the king being blocked from escaping – White would need to move two pieces in order for the king to travel up the first file.)

This knight attack doesn't work if White moves the pawn up.
So: ☗P-1f [１六歩] ☖P-1d [１四歩] and 
☗P-3f [３六歩] ... ☗N-3g [３七桂]... ☗P-1e [１五歩] ☖Px1e [１五歩] ☗P*1c [１三歩] ☖Lx1c [１三香] ☗N-2e [２五桂],
then, White just moves: 
☖L-1d [１四香]
to avoid capture (as there's no pawn on 1d [１四])

Anyway, that's one explanation i've seen before.

It also allows White's bishop [角] to peep around the corner at ☖B-1c [１三角] defending 2d [２四]. If Black tries to do a Tsukada Special attack without thinking carefully, White's bishop may be able to attack the rook [飛] on 2d and then promote the bishop on 5g (i.e., ☖Bx5g+ [５七角成]).

There are of course variations where White doesn't move to 1d as well as variations where White moves to 1d even though Black doesn't move to 1f.

1: lordjackes (1828) 2017-01-17 15:20

It's not like white "should" respond with moving the same pawn but it can be useful in the future. Black can play next P-1e so he is gaining more and more space. Even in the future, it could be possible for black to play P-1d and when white takes on Px1d put a pawn on 1c.
If the kings move to that side is good to have some escape route.