"over 50% of websites are now needlessly slow"
Yeah, that thing wouldn't be not useful to webmonkeys or they wouldn't use it. Webmonkeys who all use the latest in browsers with the best of javascript accellerators on a nicely fast machine with a wide screen attached. But not all of us do. And that code is eating cycles on the client side, not the server side.
This isn't even about form, but about preference in what brand gears to power your website with. Who cares? What happened to function? To delivering the content the site visitor came to get? Or is "UX XP" the only thing that still counts? What?
As such, this "study" is a bit useless. Including the glossing over that plenty of people just get some package from somewhere that then just happens to use this thing. Or they'll use it themselves but haven't realised that indeed, sucking a lot of code just for one or two functions, and from elsewhere to boot, is a recipe for a needlessly slow website.
Because, you know, everybody has a computer just like mine, with a screen just like mine, with an os just like mine, with a browser just like mine that runs permanently in a maximised window just like mine. These are perfectly valid assumptions and so is the assumption that the popularity of the code behind it matters down to the percentage points of how much dodgy studies say it is used.
Dear webmonkeys, be a little considerate of the other people's cycles you're deciding to waste, there's a good simian, thanks.