Firsty. despite the title, I love writing UI. It's one of the best open ended challenges programming has to offer.
I thought this discussion focused to me on the implementation of UI and lacked discussion of the connection between the program and the UI. When people say that UI is difficult, they mean that its difficult to control. Writing buttons and edit fields, layout etc. takes effort but it shouldn't be difficult by itself.
What matters is how you get data into and out of the UI. How it responds to or controls program state. For example, I might be using a program on a device with a touchscreen, and I'm doing a two-finger pinch-zoom. But I also have a mouse, and I'm rolling the mouse wheel at the same time. Perhaps I also have a pen and want to pinch-zoom while drawing a line, while using voice recognition to say the colour of line I want. I rotated the device and then the phone rang when a file arrived from some cloud storage and the battery power dropped below 10%.
This is obviously a complex problem. It has little to do with buttons and sliders but everything to do with user interface. I think the reason the conversation focused that way it did is because the DOM has turned UI implementation into a challenge of its own. All these feature are made irrelevant by the browser to a large degree, it handles some of them and restricts others. In theory immediate mode UI should help with this but the podcast didn't cover that aspect.