That’s not totally true though. React, used correctly, lets you use and share a mental model where you have clearly defined the minimum complete information required to display any state of the app, and separately you have a set of simple functions which reliably convert pieces of that data to correct html.
That’s insanely powerful and massively reduces complexity in highly interactive apps, because it means you no longer need to think about “user did this, update this here now, and here and here” chains, and can just do “user did this, update the state” and have that state update render potentially hundreds of changes across the app without needing to do additional implementation.
But if you aren’t disciplined with maintaining, or don’t understand the difference between pure functions and side effects, and fall back to hacks to try to “make things render” or “stop wasting render cycles” you’ll end up with even more complexity than necessary.
Using react as another version of `getElementById` leads to a horrible mess much worse than writing vanilla javascript, and most people who have that complaint are doing it to themselves by fighting react either unintentionally or on purpose to show how it can be bad when abused.
Raw HTML and CSS is much easier to understand, and mandatory to understand because at the end of the day creating HTML and CSS is the only goal, but higher level abstractions can be super useful to manage complex projects, and the mental models and abstractions are the actual benefit.