By localizing basic style information and reducing the amount of weird CSS you have to write, you have thus reduced your keystrokes per style. We're talking about the same thing.
By localizing basic style information and reducing the amount of weird CSS you have to write, you have thus reduced your keystrokes per style. We're talking about the same thing.
No, the semantics are different. The goal has never been "type less", which changes how it interacts with your premise. The goal has been reduce cognitive load and increase semantic density and discoverability. And, counter to what you said ("that's not as useful in the age of AI"), those are very useful in the current age of AI.
Increasing locality increases semantic density. As a utility class library, I'd say it's main goals have healthy overlap with other utility class libraries..
A very large percentage of the Tailwind classes also seem to be more or less 1:1 with CSS, or like 1:2.. So it's the locality and brevity IMHO that contributes most to reducing cognitive load.. Or at least in making it faster..
LLMs get along well with high locality..
I'm not sure if I'm disagreeing or agreeing with you now LOL.