Making photons attract?

No.
Why do you think so?

 

I think the answer is yes, and it is mentioned in Feynman's NZ lectures. I don't think this is what you were thinking of, but it goes like this:
Time is up, space is horizontal, ignore the dots:
A . . B
| . . |
| . . |
| . . |
| . . |
a . . b
The probability that you will get photons at A and B a short time after them being at a and b is calculated using the two events shown. But there is also the possibility of this:
A . . B
. \. /
.. \/
.. /\
. /. \
a . . b
Which is less likely as it involves a larger distance for one thing.
But taking the two together, it comes out that the most likely configuration for A and B is slightly closer together than a and b.

 

Lights don't have to move in straight lines, if you don't know.
Light could move in any path in space, but the path which have the highest probability will be observed.

Read this: http://www.scribd.com/doc/7119268/2physics-Qed-Feynman-Qed-the-Strange-Theory-of-Light-and-Matter-Princeton-University-Pre" [Broken] Chapter 1

 

Yes, as Black Integra said, light can take any path it likes, the highest probability path is usually a straight line. See page 73 of the above link.