These are excellent points. One huge advantage of GUIs is that someone unfamiliar with the system can still figure their way around with relative ease. That is much harder with a text based system. Making structured language a part of their culture would be a great step.

CLI is better just for the tasks for which it is better. IDE, CAD, imaging, document edition, mapping applications,... Mostly everything a human can do with a computer is easier in a GUI.

@DiegoSánchez Many professional programmers use advanced text editors(Emacs, vim) with additional packages and they like them better than any IDE (I am one of them). I spend all the time in CLI, using my Emacs and I'd never trade it for any other IDE. For me it's vastly superior. This comes at a price(you need to learn it), but now I can change it so that it suits my needs. CLI is very bad for things which aren't really text - you can't really edit video or image in CLI. Editing text is perfectly fine.

@MatthewRock that's my point: programmers are a small subset of all computer users, and even among those only a relatively small percentage prefer CLI over GUI. Those and sysadmins are the only groups that really make use of CLI this days. For almost anything else, mouse or touch are vastly superior.

@DiegoSánchez: They are not "vastly superior", they are "vastly easier to use".

What would prevent other languages & dialects from popping up? Every developer writing a tool will have to appropriate The Language and make assumptions/invent new words for actions that have not yet been considered. A real life example are the French, they have a central committee/institute that invents new words for things that have recently come up, but these things need to first come up, so that won't work without delaying technological progress

@dot_Sp0T That's exactly the challenge. But as in your example, if you assume that once a word is invented it's a standard and everyone uses just that, then problem solved.

@DotanReis what I try to imply is that it's wishful thinking at best to assume everyone will comply to that one single standard; especially when you need to progress in technology..

@Kys I'm pretty sure that someone placed in front of, say, an early IBM PC which auto-starts Norton Commander with automenus enabled, can find their way around using the basic functionality of the computer with minimal training (on the order of "use these keys to move the selection, use this key to make a selection, and use this key to stop doing something"). Just because it isn't a GUI doesn't have to mean it is based purely on a command line and manually entered commands.

@MichaelKjörling I don't think you can get very far if you allow Norton Commander as "not GUI". Sure, it's text-based with easy access to CLI, but it's fundamentally a GUI. It even uses the mouse, a pointer, keyboard shortcuts, menus... What exactly would you consider a GUI if Norton Commander isn't a GUI?

@Luaan - I am surprised how many people remember Norton Commander (I still use it daily as Midnight Commander - try it!) but NC is perfect example how some interface can be visual without being graphic. NC works perfectly on alphanumeric terminal.

@Kys: Absolutely and totally untrue that "someone unfamiliar with the system can still figure their way around with relative ease". I certainly can't, and I've been making a decent living working on the bleeding edge of tech for quite a while now. Put me in front of a Windoze machine, and I'm clueless. Even with e.g. Open Office on Linux (which I'm forced to use sometimes when people send me Word documents), I could never manage to produce a decent paper with graphs & equations, something I can do with ease with text-based LaTeX.

@jamesqf It's not as simple as "graphical" vs "CLI". Which system do you think has better discoverability - one where you type "help" to get a nice helpful screen to get you started, or one where you need to type "man"? One where you press the button with a question mark icon, or one where you press F1? One where you type "exit" to exit, or where you type ":qw!"? TeX is not discoverable. You get good resources if you know where to look, and it's quite powerful; but I never needed equations in Word, and it took me 10 seconds to find the function. Insert -> equation. Or search for "equation".

@MichaelKjörling Well, that's exactly what I'm saying :) Though I have to say that MS-DOS wasn't entirely single-tasked - it would be possible to use a TSR to provide contextual help anytime, anywhere. It just wasn't worth the trouble.

@Luaan: Why on earth should I think a button with a question mark on it means "help"? (If I preferred to type "help" instead of "man" in a CLI, it's trivially simple to create an alias.) If you want "discoverable", what could be more so than looking up things in an index, or RTFM? But with (most) GUI problems, your supposed to proceed by trial & error. And if you want "exit" to leave your editor of choice (which for me is not vi(m), define it that way. I prefer the F10 (save) and F3 (quit) of my editor, but I could just as easily type the full commands.

@jamesqf On the other hand, you completely missed the point. Congratulations, that must have taken considerable effort.

@Luaan: Perhaps that means your point wasn't discoverable? Or maybe you missed my point, which is that complex systems like computers SHOULDN'T have discoverability through trial & error as a primary design criterion. That's not even considering the problems faced by dysiconics like me, to whom those supposedly 'intuitive' icons are absolutely meaningless. I'll ask you again to explain exactly why I should be expected to deduce that a button with a picture of a question mark on it is supposed to mean 'help'.

@jamesqf I don't think there's any point. There just seems to be an inherent inability of the two of us to communicate with each other, as many other "discussions" over the SE network clearly show. It doesn't seem our thinking is compatible, and it's not getting better over time :) And you know, you called it "a picture of a question mark"; do you really not get the association that it means "question"? That's very interesting. I've never known a "dysiconic" before, do you have others in your social circles with the same "disability"? That might be a great reason to keep information textual.

@jamesqf As for whether the discoverability should or shouldn't be there, tough luck. The guy who "invents" discoverability will win over whatever you design and spread anyway. People want discoverability - because computer is a tool, something you use to achieve some other task. You don't care about the tool, you only care about the task. People do the same thing with all their other tools, mostly ignorant of any dangers. If the tool doesn't help you, you get a different tool. Look at people using Excel - they could save so much time with a bit of investment, but they still don't invest.

Good luck making music without a GUI. Good luck editing video without a GUI. CLI Games? (Zork? yeah, it was good, but....). AutoCAD or other 3D design, project management, data analysis, report generation, ladder logic, logistics, scheduling, flight planning and management,... seriously, without a GUI the modern computer would be woefully limited. The CLI is great for sysadmins and (some) developers - for the rest of the world who are using computers to actually do things beyond configuring a computer for someone else who will do things, the GUI is absolutely indispensable.

Does anyone really use actual text console for work any more? Terminal windows, sure. But those terminal windows are in GUI desktop windows, so they can be seen at the same time. Trying to seriously do something complex with (for example) Linux virtual terminals (the text ones) is rather tedious, unless you know the task well enough that you could almost do it eyes closed.