This might be a "the grass is greener on the other side" situation because I do a lot more web than native dev, but in my experience native while just as quirky as web will usually give you low level APIs to work around design flaws. On web it too often feels like you can either accept a slightly janky result or throw everything away and use canvas or webgl. Here are some recent examples I stumbled across: - try putting a semi transparent element on part of an image with rounded corners and you will observe unfixable anti-alias issues in those corners - try animating together an input with the on screen keyboard - try doing a JS driven animation in a real app (never the main thread feels hopeless and houdini animation worklets never materialized)

I don't think it's that native has nothing to offer. I think that developing (in case of desktop) for 3 different platforms all with own complication of what is native UI is a nightmare. macos has swiftui (incomplete), uikit and appkit, linux in practice gtk/qt, windows winui 3 (fundamentally broken) with WPF and WinForms still hanging around .