The decision that wasn't
Somewhere in the early 2010s, a border radius of sixteen pixels and a drop shadow of roughly four to eight pixels at low opacity became the standard appearance of a card, a modal, a button, a panel — almost any contained element on a screen. Neither value was the result of research. Both travelled because they were copied from one popular interface to the next, copied into production, then copied again into the next toolkit that learned from the first.
The sixteen-pixel radius sits in a specific perceptual pocket: soft enough to read as friendly, small enough to avoid looking like a pill, large enough to register at a glance. That balance is real, which is why the value survived scrutiny when people eventually applied it. But surviving scrutiny and being chosen are different things. Most teams inherited it before they had the chance to question it.
The shadow is subtler and stranger. A box shadow with a vertical offset, a moderate blur, and an opacity somewhere around ten to twenty percent produces the impression of a surface lifted slightly off the page — the core visual logic of Material Design, but also the logic of every card interface that arrived before and after it. The shadow tells the eye: this thing is a discrete object, not a zone. That is useful information. It is also a convention so widespread that removing it now requires a deliberate act, a conscious decision to go flat and accept that flatness will read as a different signal, not as neutral.
The card as universal container depends on both of these values to function visually. Strip the radius and the shadow, and the rectangle stops reading as a container; it reads as a layout region, which carries different implications for how a user approaches what is inside it.
The cost of inherited defaults is not that they are wrong. It is that they arrive before the question.