Say we have slice of b such
b:= make([]int, 0, 5) // length: 0, cap: 5
and slice of "c" made from slice of "b"
c:= b[:2] // length: 2 (?), cap: 5
The question if how come we have length of 2 for "c"? I was expecting length of zero as well, like b, since we are making c out of b
Yep, slicing can get you access to the elements beyond the len of the original slice (though not beyond its cap, or who knows what memory you'd be accessing).
This means, for example, that you can implement append-like functionality, returning a "grown" slice with len increased to something closer to the cap. That is, append's access to the region between len and cap isn't only available to built-in functions; you have it as well. Look at Append: an example in the Go blog entry on slices to see it done (and if interested read the whole post; it helps make slices as a whole make sense).
The spec on slice expressions (and slice types) and the Slice Tricks page may also be interesting.