Because absolutely nobody has given an answer that works on generic iterables,
from itertools import chain
items = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
zip(*[iter(items)]*2)
#>>> <zip object at 0x7fd673afd050>
[itms for itms in zip(*[iter(items)]*2)]
#>>> [(1, 2), (3, 4), (5, 6), (7, 8), (9, 10)]
So zip(*[iter(x)]*2)
means ix = iter(x); zip(ix, ix)
which pairs each element.
Then you can reverse:
[(y, x) for (x, y) in zip(*[iter(items)]*2)]
#>>> [(2, 1), (4, 3), (6, 5), (8, 7), (10, 9)]
Putting it all together and flattening:
[itm for (x, y) in zip(*[iter(items)]*2) for itm in (y, x)]
#>>> [2, 1, 4, 3, 6, 5, 8, 7, 10, 9]
It's generic and short!
If you want something faster at the expense of genericism, you'll be hard pressed to better this:
new = list(items)
new[::2], new[1::2] = new[1::2], new[::2]
new
#>>> [2, 1, 4, 3, 6, 5, 8, 7, 10, 9]
Note that this still works on arbitrary iterables, but there are fewer layers of abstraction; you can't bump up the size of the flipped sub-lists as easily and can't output iterables, etc.