12 ms·

This isn't wrong, but the problem in general is that non-programmers have pre-decided that "simple UI" means a clickable GUI, probably with a Paperclip type ass

This isn't wrong, but the problem in general is that non-programmers have pre-decided that "simple UI" means a clickable GUI, probably with a Paperclip type assistant. The benefit of the GUI is to present something that text can't - such as lines between components, or pictures, etc. But what's the (specific) domain use for those thing? If there isn't one then a GUI is just a text UI with a mouse pointer and a bunch of overhead. There's almost zero connection between what seems simple to your boss in a screenshot and a productive tool. If someone really removed unneeded complexity from a burdened UI, then great, but one person's bell and whistle is another's required tool.

humanrebar 10y ago
Agreed. I've seen many simple, bulletproof UIs and APIs that only support a discrete set of workflows, so they ultimately fail as expert tools. Sometimes you don't want an expert tool, you want an easy button. That's fine. But often there's an inherent tension between 'can never fail' and 'lets you open up the throttle and tear through your problem'. Compare the editing and reviewing workflows supported by a share drive and those supported by git. A shared folder is undoubtedly simpler, but good luck maintaining a coherent revision history or editing a document in parallel on different isolated subnets.