I've already submitted this as feedback through the Send feedback tool, but I wanted to upload a video demonstrating the issue.
Earlier this week, the scroll speed for Edge Canary was tweaked to carry more momentum. By that, I mean webpages travel farther when you flick the touch screen. It is now in line with my other Windows 10 apps. I really like this change, and I would have said it makes the experience smoother... except that, in upping the speed, the Edge team also brought back the choppy, stuttering effects with a vengeance. To put it in comparison, for those who can't see the attached video, I can only really tell that Chrome is imperfect when I put it side by side with Edge Dev. With the latest versions of Canary, I can tell there is a serious issue as soon as I try to use the browser.
I have (hopefully successfully) attached a video of this problem in action. On the left is Edge Dev, which you can tell decelerates too quickly but is buttery smooth. On the right is Edge Canary, which keeps moving after you lift your finger, but is noticeably choppy. If you're having trouble seeing it, try glancing back and forth between the two browsers a few times, and watch what happens when I lift my fingers, not when I'm actively controlling where I scroll to.
I'd also like to note that the new scrolling in Edge Canary is equally choppy when using a trackpad; it's just easier to show what's going on with the touch screen. Mouse wheel based scrolling seems fine, but don't quote me on that.
Bonus fact: If you try to trim your video length using the photos app for Windows 10, it will cut the frame-rate to 30 fps. This makes the effect much harder to notice. If you want to contribute to this thread, make sure you keep your video frame-rate at 60 fps.
Now that I pay attention to my scrolling with mouse with smooth scrolling flag enabled, I see it's slow at first, then its speed increases and finally decreases at the end again. like it has built in acceleration. don't know if that's their definition of making scrolling smooth or not..