My silicon chess friend often suggests moves that are hard to understand – at least for me, a mere woodchopper...
As an example, in the Pohlmann defense, Houdini recommends 6.Bb5+ (diagram) after 1.d4 d5 2.e4 dxe4 3.Nc3 f5 4.f3 exf3 ( Scheerer prefers 4...e5, but I doubt his analysis as he relies on Milov-Kramnik, 1990, but Milov could have improved by playing 7.Nb5 instead of 5.dxe5 Qxd1+ 6.Kxd1 Nc6 7.Nd5 ) 5.Nxf3 Nf6. Now I teach chess to children and I always learn them that giving a check like that is not so good, as it allows the defender to chase away the bishop with the simple c6. Applying that golden rule would indicate that 6.Bc4 is preferred over 6.Bb5+. Still Houdini recommends the latter over the logical 6.Bc4. Why ? Well, let's look at the different ways of dealing with the check.
a/ 6...c6 The most simple way of blocking the check 7.Bc4 (diagram)
The attacker has now taken away the natural black move Nc6.
...a2/ 7...e6 8.Qe2 (+=)
...a3/ 7...g6 8.Qe2 (+=)
...a4/ 7...c5 8.Bf4 (+)
...a5/ 7...Nbd7 8.Ng5 (+=)
...a6/ 7...Na6 8.Ng5 (+=)
...a7/ 7...Nd5 8.Bg5 (+=)
...a8/ 7...Qc7 8.00 (+=)
...a9/ 7...Qb6 8.00 (+)
b/ 6...Bd7 7.Qe2 (diagram)
White develops pieces
...b2/ 7...c6 8.Bc4 (+=)
...b3/ 7...a6 8.Bxd7 Qxd7 9.Bg5 (+=)
...b4/ 7...c5 8.Bf4 (+)
...b5/ 7...e6 8.Bf4 (+=)
...b6/ 7...g6 8.Bf4 (+)
c/ 6...Nc6 7.d5 a6 8.Ba4 b5 9.Nxb5 axb5 10.Bxb5 (+=)
d/ 6...Nbd7 7.Ng5 c6 8.Bc4 Nd5 9.Ne6 (+)
So analysis proves that white gets a small advantage after 6.Bb5+. After 6.Bc4, black is able to equalise with 6...e6 (diagram).
So Houdini was right, in this case 6.Bb5+ is better than 6.Bc4 as it denies black to place his knight on the c6 square. But as a general rule, I will stick to telling children that Bc4 is better...