max-width is useful. It is also easy to over-credit.
A lot of sites “handle readability” by setting one container width and moving on. That works just well enough to hide the deeper problem: a container is not the same as a reading measure.
A page can need room for metadata, figures, side notes, or supporting controls while the prose itself wants a tighter column. The minute those two needs get forced into one number, someone loses.
What usually loses is the reader.
The cleanest upgrade is to stop thinking in terms of one inherited width token and start thinking in layers:
- page width for composition
- reading measure for prose
- explicit breakouts for blocks that genuinely need more room
That shift is small in code and large in effect.
A lot of readability work is like this. The improvement is not flashy. It is structural. Pages stop feeling careless. Writing gets easier to follow. Supporting blocks stop knocking the prose off balance.
If a layout improvement makes the content feel calmer without making the system more fragile, it is probably the right kind of change.