The CSS I Stopped Writing Once I Learned Flexbox and Grid
Float clearfixes, negative margins, absolute-positioned centering, JavaScript measuring heights — a before/after of the layout hacks I retired the moment Flexbox and Grid clicked.
The short answer: once Flexbox and Grid clicked, I stopped writing float-based columns, clearfix hacks, negative-margin gutters, absolute-positioning-plus-transform centering, and any JavaScript that measured or set element heights for layout — because the engine now does all of that natively if you just ask it directly.
There's a specific kind of CSS I used to write that I now find almost funny to look at. Not bad CSS exactly — it worked, it shipped, people used it — but it was fighting the layout engine instead of asking it for what I wanted. Floats pretending to be columns. Negative margins pretending to be gutters. JavaScript pretending to be a layout system. Flexbox and Grid didn't just give me new tools; they let me delete whole categories of code I'd convinced myself were just "how CSS works." Here's the before/after.
Centering a box
The classic. Vertically centering a thing used to be a genuine interview question, which tells you everything. My go-to was absolute positioning plus a transform, and it was fragile the moment the content's size changed.
Three properties, no positioning context, no transform math, works for any content size. The "after" reads like a description of the goal instead of a workaround for the engine.
Columns
This is the one that still makes me wince. To get a row of equal columns I'd float them, calculate percentage widths by hand, and then — because floated elements collapse their parent — add a clearfix. We all copied the same clearfix snippet from the same Stack Overflow answer for the better part of a decade.
gap alone retired every negative-margin gutter hack I ever wrote. No more .row { margin: 0 -8px } paired with .col { padding: 0 8px } and the inevitable off-by-one at the container edge.
Two-dimensional layouts
Flexbox handles a row or a column. The moment I needed rows and columns to line up — a card grid, a page shell with a header, sidebar, content, and footer — I used to nest flex containers inside flex containers until the markup was a matryoshka doll. Grid is built for exactly this.
The layout is now readable as ASCII art in the stylesheet. A new teammate can see the structure of the whole page in six lines, and rearranging it is editing a string, not untangling nested divs.
Responsive card grids without media queries
The hack I'm least sad to lose: computing how many cards fit per row at each breakpoint with a stack of media queries. Grid does it with one line using auto-fill and minmax.
The JavaScript I deleted
The quiet win is all the layout code that left my JavaScript. I used to measure element heights with getBoundingClientRect to make two panels match, or to pin a footer to the bottom when content was short. Flexbox's flex: 1 and Grid's 1fr do that in the layout engine, synchronously, with no resize listeners and no reflow thrash. Every one of those useEffect-measures-the-DOM hooks was a bug waiting to happen, and modern CSS made them vanish.
The shift
The real change wasn't learning new properties — it was a change in posture. Old CSS was adversarial: I'd picture the layout, then reverse-engineer which hacks would trick the box model into producing it. New CSS is declarative: I describe the layout I want and let the engine solve it. I write a fraction of the code, it's more robust, and it reads like intent. If you're still reaching for a float or a translate(-50%, -50%) out of muscle memory, stop and ask the engine directly. It's been able to do what you want for years now.
If the open question is still which of the two to reach for in a given spot, I laid out the rule in Flexbox vs Grid: when to use which.