What night mode actually does

Night Shift on Apple devices, Night Light on Windows and Android, and the equivalent features elsewhere all work the same way: they shift the display's white point toward the warm end, reducing the blue component of what the panel emits, usually on a schedule tied to sunset.

Two honest observations about it. First, on a strong setting the effect is substantial — considerably more than a near-clear lens does. Second, the evidence that it improves sleep is no stronger than the evidence for the glasses; studies of screen colour shifting have produced mixed results, and screen brightness and how engaging the content is may matter more than its colour.

What glasses do differently

A lens sits between your eyes and everything in your field of view, so it applies to the television, the phone, the laptop and the lamp you happen to be facing, all at once. That breadth is the genuine advantage over a per-device setting.

The limitation is geometry. Light entering around the sides and top of a frame is unfiltered, which is why the safety-style wraparound designs used in some sleep research look the way they do. An ordinary frame leaks.

Device night modeBlue-blocking glasses
CostFree, already installed$15 upwards
CoversOne screen at a timeEverything you look through the lens at
LeakageNone for that screenLight enters around the frame
StrengthAdjustable, can be strongDepends entirely on tint; clear lenses do very little
Affects other peopleNoNo
Colour workToggle it offTake them off
Evidence of sleep benefitMixed, unresolvedMixed, unresolved

What neither one addresses

Room lighting. A bright ceiling light in the evening delivers far more light to your eyes than a phone at arm's length, and neither of these interventions touches it. If the circadian argument is what motivates you, dimming the room is the larger lever and it costs nothing.

Nor does either address what the screen is doing to your attention. An argument in a comment thread at midnight is stimulating regardless of its colour temperature, and the alerting effect of engaging content is not a lighting problem.

Related: the circadian evidence in full.