Aug 30, 2025, 7:39 AMSolid colors clearly delineate controls
More importantly, the visual metaphor of a button is universal. Anything you can press (click on) has a bevel around the edge that makes it appear raised. This includes the menus. Every menu item is a button, so you don’t need to learn anything about menus to know that you can click on them. Menus that expand to submenus have an arrow, which you do need to learn, but it’s discoverable as soon as you click on one (which you know you can because of the bevel). This is even carried to window title bars, which are raised in a similar way to indicate that you can drag them.
It’s not visible in this screenshot, but anything that you place data into is sunk, to represent a hole in the UI that you can fill in. Text fields, for example, have the opposite bevel to a button. Colour wells combine these because they are a drop target for colours (you can drop a colour in them to set the colour) but they are also a button that you can press, so they look like a caldera: a raised area with a coloured hole in the middle.
Newer GUIs seem to have forgotten that the purpose of visual design is to convey information in a way that reduces cognitive load. You can look at a NeXT window and see exactly what you can click on, what you can enter data into, and what is purely informative, without thinking. The design is so consistent that you don’t even realise that you’re making the connection.
With OS X 10.0, Apple maintained this (mostly: there were rough corners where Cocoa / OpenStep visual metaphors clashed with Carbon / Classic MacOS ones). Clickable things were both raised and also coloured like boiled sweets. Informative things were on grey backgrounds, things you interacted with were white. It wasn’t quite as good in a few places, but it mostly worked. Modern macOS (even before Liquid Glass, which is a total disaster) has gradually lost a lot of this consistency.
When we designed the Nesedah theme as an updated NeXT look, we used gradients to indicate anything that you would interact with. It worked pretty well. I tried showing a few people mockups and they were able to identify all of the UI components without my explaining them. If I had been doing it for a corporation, I’d have done proper user studies to make sure. It’s quite clear MS and Apple are not doing that anymore.
