Co-author of article here. Some notes:

* Light levels matter a lot, and we are trying to move the discussion to objective (and impartial) measures rather than talking about colors and subjective impressions

* Yes, our surveys of viewing distance (5000 people) were done in imperial, so we describe it that way for an English-speaking audience

* Yes, metamers are real - two spectra can stimulate the cones identically but be different to the non-visual system. We are comparing screens with identical luminance and 2-degree observer chromaticity matches, and the melanopic response does in fact vary

* The SCN has a logarithmic dose-response to light, so indeed 20% isn't that much. But 2x is an important change to talk about

* Our goals do not include making screens redder, but to improve outcomes for most people. We think the way to do this is by improving the contrast between day and night, and that this means that many of the the lights in the world should use timers that people can adjust based on the outcome they want

* The biggest surprise here (and the impetus for the article) is "wow that's more than double" when you first measure the phone, because the iPhone X doesn't feel that much bigger than an older phone

* Defaults certainly matter, especially when a product implies it will improve your sleep