Here’s a design system failure I’ve seen more than once: Everything looks similar, but nothing actually connects. Six months in, the team realizes the button component doesn’t share spacing tokens with the form inputs. The card component uses a slightly different border radius than the modal. The mobile nav was built by a different squad and uses a different grid.
LEGO solved this problem in 1958 and never looked back.
Every brick, across every product line, every theme, every decade, uses the same stud spacing, and the same base geometry. It’s not magic. It’s a single constraint that was decided once and never negotiated away. Because of that one decision, a brick from 1978 connects to something from last year’s Technic set.
That constraint didn’t limit what LEGO could make. Pirates. Space stations. Dinosaurs. Medieval castles. Entire universes. They’re all built from the same underlying system. The foundation never changed. Everything built on top of it did.
The design systems that hold up over time do the same thing. It’s not because they have more components, or better documentation, or a more thorough governance process. They hold up because someone decided what the non-negotiables were early, and protected them.
Base grid. Spacing scale. Interaction model. Those don’t move.
Everything else can flex. Color themes, component variants, density options. Fine. But the studs stay put.
The drift never announces itself. It starts with a reasonable exception. A component built under deadline that uses a slightly different spacing value. A nav rebuilt by a new squad that inherited a different grid. Nobody flags it. Nobody feels like they have to. It looks fine. It’s only later, when you try to connect the pieces, that you realize the foundation moved.
Most systems I’ve seen fail didn’t collapse. They drifted. And nobody noticed until the day someone tried to connect two components that should have been identical and found they weren’t quite.
(Did I use making a LinkedIn post to an excuse to pull out the LEGO? Yes. Yes I did. But the question underneath it is real: what are the non-negotiables in your design system?)