Re: Why would anyone want onscreen buttons???
Software buttons have been the minimum standard for 3 years now. Last year, Kit Kat made full screen "immersive" mode the easiest thing in the world for devs to add to their apps. I think losing .4" on a 5.2" screen is a little aggressive; by pixel count the area has typically been 5%, which on 5.2" would be .26". So in a worst case scenario (on the home screen) things are 5.26% bigger with off-screen buttons - but nothing is added, things are just enlarged slightly to be spread out over the different areas. In a poorly made app, more content could be shown... but most don't, they just stretch because they want to show a preset amount of stuff. On well made content apps, they use immersive for anything you'd want to enlarge (images, video, text) and it's a non-issue.
I can't see any argument against meeting the minimum standard in design guidelines. Any utility "loss" is at best negligible and no functionality loss is possible. That being said, the guidelines do allow for capacative buttons - but there is a significant problem created and it is one that the Turbo and Note 4 (as well as all other hardware keyed devices created so far) share: the button schema is already legacy. There are ICS buttons on devices that will presumably run Lollipop and possibly the M version.
Another obvious drawback to those who have used onscreen buttons that try to use buttons that are offscreen - onscreen buttons can move, change, be customized, be resized, contextually adjust to apps, etc. If you pick up your phone or tablet upside down in landscape, those buttons can be in their normal position if you have them set up to do so... meaning you don't have to fumble around for the "correct" side of the device. Not as important on phones for most, but on tablets that is a pretty handy value add.
I understand that people have different preferences, but saying that on-screen buttons are objectively worse seems to be a mistake. The potential value add in usability combined with the improvement in design language both put on-screen in the objectively better category. It is fine to prefer something that is different then the best practice, opinions and preferences are totally subjective and welcome and needed. But liking something more doesn't necessarily mean that it is better and in this case it seems certainly to not be true that enjoyment of the legacy feature equates to a better feature set.