They take a load of space, they look like a leftover from the nineties, they take different space on every browser so if you want a precise layout positioning you can't, they don't reliably report positions or events, they can't be reliably scrolled at a specific relative position on every browser and more.
I'm fine with the downvotes, people around here downvote to disagree even if they don't actually know what suffering and misery cross browser development is.
You're getting downvoted because your complaints are either rather vague (doesn't measuring viewport size exclude scrollbar size, so their size is irrelevant to you?) and indicate a desire to mess around with what is very much not yours to mess around with (size, styling, position of scrollbars) which simply means you don't respect your users.
that's a load of work for some fake internet point.
and no on some browser the scrollbars consume layout space and on some they don't. (remember to add the ios/safari monstrosities into the mix)
heck, just open this http://www.matanich.com/test/viewport-width/ on edge/ie11/chrome and see yourself how reliable clientWidth is (especially test edge fullscreen vs edge windowed but fullscreen it will blow your mind)
Alright, so some browsers like Edge will report the full screen width available even with scrollbar visible, because it fades out the bar after a while. You could've just said that.
That is understandable as a reason. Although i still think it's a pretty bad reason to use to force "smoOoOo--oOoth" scroll on everyone, instead of only browsers that are known to be broken.
I'm fine with the downvotes, people around here downvote to disagree even if they don't actually know what suffering and misery cross browser development is.