I just noticed that in the low light test the S9 shot wasn't focused anywhere on the scene, really? I know there's a failure rate for focus under those conditions but should you really use a sample like that?

The 70D results seem incredibly mushy. Though Canon sensors are almost always 1-2 gens behind Sony it doesn't look like a sensor problem. It looks to me like you used a bad sample of a bad lens wide open, such that it's not only slightly out of focus across much of the frame(you should test sharpness at infinity, not some scene with such difference in foreground and background, wide open) but soft all across the frame. Normalizing the effective signal it's probably still safe to stop down the lens at least a stop, or better just use a decent fast prime and stop down some more.
Also, I see strange patches of smearing that are simply impossible on a half frame in daylight, it's either bad compression parameters during upload or that the output settings weren't right. I don't think this does the 70D justice, you may want to revise the testing protocol.

Great article, thanks! I have had Samsungs and iPhones before HTC U11+ and really love it but would like even slimmer bezels and a dedicated bookeh effect (now I have to rely on the Pixel 2 camera app). Seems this one might deliver so almost wish I had waited with the upgrade a tad longer ;)