The AppsLab

Command lines gave way to GUIs, and it seems like touch computing is the next O/S paradigm. I think your point about a common understanding is valid, and ideally, touch computing focuses on commonality and not on replicating the mouse experience for fingers.

I think it is not GUI as such, but the agreement on what the visual concepts mean. This is different to text terminals, because you need to remember what that command was called.

Have you tried using voice commands on Vista? I did. What threw me was that I do not think in terms of "switch to notepad", as the tutorial suggests. I think in terms of "click that over there" or "alt-tab, watch with alt pressed, then press tab a few times to select the mozilla icon". That is why I cannot use the voice commands very well.

The point I am making is that with GUI people start using things without having common understanding of how it works, so it can be used in cases when this understanding is not important. Try to share knowledge like that.

I think we can agree that GUIs are good for the new/uninitiated users and for inherently visual tasks. Whereas, keyboard actions are better for tasks that require a lot of typing, e.g. coding, writing, etc. The cross-over apps are the hard ones to figure.

I would like to disagree about "learning the concepts faster". Perhaps, it is more of being able to use the concepts without learning them.

GUI is good when you don't know how to get what you want, you can stick more controls and emphasize them differently using GUI, unlike when using a text terminal. You can use the mouse to tell "I want *that*" without knowing the concept. (But you still need to learn the concepts of click, double-click and drag-and-drop, the concepts of button, menu, drop-down list boxes and text scroll area, and loads of other concepts which seem to us, GUI users, natural)

I think that GUI comes easier to the people because the developers of the interface silently came to a standard of what interface concepts mean what. (Go standardize shell commands!.. But buttons come across as activators of an action across all GUI)


A keyboard has 101 dimensions of freedom, whereas a mouse has 4 or slightly more, if you have three buttons or a wheel.

However, in the dev world, what do you need a GUI editor of the source code for? The project is built using ant, there is a lot of disk navigation and file inspection, when debugging a problem. Code highlighting is available in non-GUI editors too.

I have tried GUI editors, where they attempt to tell me what the method name is. I am sorry, if it didn't stop to think about the method name, I would have typed it long ago.

I have seen developers using XTerm to navigate somewhere, and trying to copy-paste a path from a different XTerm windows. Poor soul struggled to select the right part of the string with the mouse. This is an example of GUI getting in the way. Tab-completion of the path would have worked faster to type the whole path from scratch.

Oh yes, I do have the Internet browser window open to read javadocs. :-)

Bring back the PF keys!

Matt,
I'm a big keyboard only guy, especially for repetitive tasks. It saves a lot of time. From my days writing code in notepad, I still use Alt-F-S to save all kinds of files. Back when Win 95/98 were new, only the shiny new keyboards had the Start key, so I used Ctrl+Esc a lot.

I'm not against GUI by any stretch, just remembering the old days.

GUI is much easier to use and new users learn the concepts faster because of it. We expect more from modern apps because they deliver more.

However, you're right when you say "Mousing wastes time." In Oracle Apps, I can hit F11 and start my query much faster than clicking View - Query by Example - Enter from the toolbar; 1 keystroke > 3 mouse clicks. Or if I want to start Firefox I hit Ctrl + Esc, then the F key. If I want to open that spreadsheet from yesterday I hit Ctrl + Esc, the D key, and the letter of the file name to open it from My Recent Documents. Try it, you'll like it!