Small UI thing I came across: https://blog.repl.it/clui

One of the unique parts about CLUI is that any new UI is automatically generated from the data you give it.

The top comment on hacker news captures a gripe I had about their criticism of complex GUIs: https://news.ycombinator.com/item?id=22499947 To quote the article:

The moment you want to add a feature to a GUI, there’s an immediate question about where to put that feature. Should it be in the top right? Bottom left? Nav bar? Behind a tab? In the sidebar? Revealed on hover? You end up with buttons, panels, and menus competing for space in the UI-- screen real estate. Every pixel becomes more and more valuable for primary features. You don’t need to look very far to see what happens when you need to include dozens of features into a tiny rectangle.

This writing makes it feel like the GUI design process is somehow willy-nilly. "Let's just stuff the feature wherever we can find room." That's not a substantial critique — the problem, if any, with the design of GUIs isn't that they are sometimes executed poorly.

Professional creative tools like Adobe Premiere are often so packed with features that they become unapproachable, slow, ugly, and unfocused.

• Unapproachable — That's a well-solved problem. • Slow — Show me a CLI-based video editor with live preview of multiple 8k streams and then we'll talk about where this perf is going. • Ugly — That's not measurable. Maybe you'd prefer something.. flat and trendy? http://nenadmilosevic.co/ableton-live-redesign/ • Unfocussed — That's a well-solved problem. 🦃 But, I do like what they're trying to do by bridging the gap between the CLI and the GUI. More exploration of the space is a good thing. I just wish they could justify it by pointing out how the thing they were building was exciting compared to the GUI and CLI at their best, rather than trying to make the GUI and CLI look bad and then putting their thing next to it.

It will be interesting to see how far they get with this approach. My main problem with the article was that they used Adobe Premiere as an example of how GUIs don’t scale, and then claim that their CLUI approach does scale. Yet they only show some trivial examples. They need to show how they can replace Premiere’s GUI with a CLUI to make that case.

People often review advanced products like video editors, and load them up on a single monitor, not realizing that they were intended for at least 2 monitors. Most of the Adobe product line which is only medium grade user interface quality, are assuming double monitors, with all their floating/dockable patents it is natural for them to cling to an idea that was good only at a medium level of complexity, and that they through feature creep are now in a pretty ugly state. Given that the command line tools of Linux have gotten out of control in terms of the number of options over time, i welcome seeing some GUI. The LS command i think has over 80 options. Linux has evolved into a giant mess of command line options, and for some reason people make the unsupportable claim that Linux is easy. I put all my trailing edge friends and relatives onto either Windows 7 or Macintosh because the support costs are lower for me. Command line tools are only suitable for those people with great memories, and for most users past 60 who don't use a piece of software often enough, command line is the kiss of death.

It's true what Edward said about the Linux command line, that it requires a great memory, but that isn't the kind of UI being discussed in the CLUI article. The article begins by quoting Donald Norman that search (eg, google search) is the new CLI. They go on to describe a UI where you use search to find the command you want, and you use autocomplete, drop-down menus and forms to select subcommands and options. This kind of interface is light years away from memorizing bash commands. I had extensive exposure to these kinds of interfaces in the 90's and early 2000's, and it's great experience to use a complex CLI where you don't have to memorize anything.

why not go all the way, and instead of using wimpy google search, which has very little intelligence and uses how many inward links to rank pages which can go completely in the wrong direction, and use Watson to back up your UI? Natural language interfaces are good for low amounts of user input, but in any creative product it would be a hindrance to use a searching based interface. Video editors make tens of thousands of little tweaks, and use $1000 keyboards with jog wheels and tons of dedicated keys to speed it all up. Ultimately the best interface style all depends on how many decisions/actions the user is going to make. Frankly in word processing people loved the old IBM XT keyboard with the 10 function keys on the left. People completely understood those keys did actions, they were in a convenient and special place, and you had little plastic templates for your word processor. People loved it but unfortunately the stupid PC industry standardized on the DEC keyboard which has the function keys way up at the top where nobody can reach them or have muscle memory for them. This idea that users are supposed to stumble around their software and ask for help using text queries is really a strange attitude about UI. Why do commands have to have so many subcommands and options? It is kind of nutty, to use super powerful search engine technology to fix the lack of intuitive workflow and tidy organization of functions. How does one know what is possible in such a system? it is completely undiscoverable, so I say Norman is completely incorrect. And not only that, naive users have no clue as to how to describe in proper terminology what they are trying to accomplish. So their queries will end badly. The worst systems IMHO are those that present a dark screen with no clue was to what commands or functions are available. That reminds me of a survey where programmers asked which text editor was the easiest and some nerds answered vi and emacs. Yeah, right, people love using LISP for their word processing queries. Yikes. MS won the OS game by using wizards on top of a complex dialog box tree. if you have enough wizards people don't have to learn where all the functions are buried. In this sense they were compensating for their intrinsically more complex OS than Apple offered. MS. had a lot more hardware to support; it was a practical compromise and worked rather well for them. Apple took the tack of drastically limiting hardware options which made it possible to offer a much simpler OS interface. That split between the two companies persists, but Apple has never really been able to crack 10% market share even in their best days, due to the tradeoffs this entails.