If I may just inject someting....
There has been several quotes of the 737 FCTM so I won't add any more (mainly because I don't have it available)
Now, If we are manually man-handling the thrust levers we bug the whole headwind and half the gust and then aim for the bug. Later we bleed of some of that but that has already been discussed.
The autothrottle does things a little bit different (no surprise there, it is a 737 after all) it wants the bug to be set at +5knots but tht doesn't mean that it's aiming for it. Boeing engineerd this thing in the pre-byte (bite?) era and solved the problems by setting different speeds on the forward and retarding motions. The effect of this is that the autothrottle is incapable of maintaining the bug speed in anything else then conditions where the wind is completely stable and critically doesn't change with altitude. So what may appear as a rubbish autothrottle is a design feature so that the actual speed is always higher than the bug speed. the more gust or wind change that is encountered the higher the margin to the bug.
The real dilemma occurs if you intend to disengage the autothrottle at say 500'. If you bug +5 the autotrottle will fly at a speed in excess of it and one might argue that you would have to change the bug setting when you disengage.
If you choose the other option the autotrottle will now fly much to fast (in theory twice as much wind addative as you wanted and bugged) and it will be a handful to get rid of that speed eventually and you could potentially exceed the flap limit.
All above not withstanding the fact that the autothrottle is a not-to-be-trusted-piece-of &^%%^ in most scenarios!
signing off