I love the Sage so far. The one I ordered was a used one from Amazon warehouse, a "like new" unit. I figured it's better to order this if I am not sure about the device as it's already been used in case I want to return it. The unit itself has a really annoying dead pixel which is why it was returned I suspect in the first place. I'll have to return it and get a replacement.
The Sage feels premium. The screen is large at 8" and I can somehow notice the extra 0.2 inches of real estate compared to a 7.8" despite it being only about 5% larger in terms of square area. The contrast and clarity are excellent. You get a sense that the screen/text is right there in front of of you. Sharp and crisp. I cannot see a layer in between. Although the background paper is slightly whiter on my old PW4, the PW4 is really no match to the Sage in terms of clarity and even blackness of the text. The Sage is definitely superior. But it just makes me wonder how Amazon is able to create that white paper effect. Interestingly, the name "paperwhite" must have been used for a reason. Someone linked this video
https://youtu.be/jm-YBQMj8Jw and although it's outdated, it does confirm that PWs have really good background paper whiteness. But yes the Sage is superior in clarity and even in contrast to my PW4 (as the blackness of the text more than makes up for the slight decrease in background whiteness)
I am not even going to test the Scribe given its bigger size and heavier weight. Going to stick with the Sage.
One thing about the battery. It drained from 70% to 1% over 2 days of light use (2-3 hours a day).
Interestingly, once it got to 1% I did put it on autoturn and left it with the frontlight on. It's been nearly an hour now of it turning pages every 10 seconds on 1% with the light at 30% (on koreader).
This does fit with the sentiment I got from this forum that battery isn't really that bad.
Perhaps it's just poor battery software calibration/poor battery hardware by Kobo that puts off many users initially. I know batteries need their cycles to calibrate and there is a software component. However, one hour with the light on (and counting) tells me there are fundamental issues with the battery level reporting (whether on a software front or hardware front). I've never had a new device that lasts an hour on 1% with the frontlight on and page turning every 10 seconds handling a heavy academic PDF. Hopefully the battery reporting will get better and recalibrate in a few cycles but the scale of the inaccuracy in battery reporting is bizarre to say the least and it may be why this device has managed to amass the poor battery reputation rather quickly.
Performance wise, the it is snappy. You won't appreciate how snappy it is until you try it. Feels more like a tablet to me. And handles a very large academic PDF with pictures and drawings on Koreader better than my laptop. Page turning and rendering is instantaneous compared to other readers. It is rather ironic as one of the things that put me off buying it when i had a look at user reviews was someone reporting how slow it was... it turned out to be quite the opposite.
Koreader is a must for academic textbooks. With the text reflow option and also the column reading option, I feel like there is no need for a larger device.
Overall, very happy. Thank you for the suggestion. I'll report on koreader battery life in a few days time. But battery doesn't seem too bad at all so far.