Wimp Is Good Enough

Morpheus: The current interfaces are a system, Neo. That system is our enemy. But when you're using these interfaces, you look around. What do you see? Icons, file names, modes, hidden interface elements, and the people attempting to use them. The very minds of the people we are trying to save. But until we do, these people are still a part of that system, and that makes them our enemy. You have to understand, most of these people are not ready to change interfaces. And many of them are so inert, so hopelessly dependent on the system that they will fight to protect it.

I don't see it like CLIs is broken, rather I see it like there are good CLIs and bad CLIs.

When you can provide an argument in favour of it, then I'll care about that.

Wimp isn't pretty good, nor is it good enough. Wimp is WorseIsBetter.

Not a very impressive argument, because in the sense of WorseIsBetter, worse really is better most of the time. Wimp (which I, too, happen to detest) does not deserve the name of WorseIsBetter; it is dressing upon a rotten cake.

You'll need a lot more ammo than that to support any argument against Wimp. Tens of millions of users would like to know of something better; show us.

[Can you wait for the BlueAbyss? This is not one of those all or nothing things that demagogues like to make it. The interfaces we have now are not that bad. People get a lot done with them. At the same time they could indeed be a lot better. There are lots of people working hard on these issues in the academic and business worlds. But future interfaces are very likely going to include all the components W,I,M,P still, to some degree. (Hopefully not so much "I") At least until shortly before cybernetic implants become commonplace (even then, we will probably have windows like objects projected directly onto our retinal nerves)]

Future interfaces hopefully won't include WIMP at all. There's already tons of research on interfaces that don't include any element of WIMP (check out http://www.nooface.net for some examples). Most video games don't include WIMP, so WIMPless interfaces are mainstream. It's only the operating systems market that stubbornly resists any advances that could benefit all end users.

For example, the game BlackAndWhite has no windows, has no icons, and has no menus. And for crying out loud, it's many years old already!

What else? Many FirstPersonShooter games get along fine without pointers of any kind; no, the barrel of your gun doesn't count. Of course, they have very limited interaction with the environment.

I believe most modern RPGs are far more complicated (interface-wise) than strategy games. The latter are mostly standardized on one interface now but the former are much more diverse.

Where it really matters nobody uses WIMP anymore!

Evidence to back this claim up, please. Perhaps I'm the "nobody" in question, or my work doesn't "really matter," but I still use Windows every single day to do productive work, just like millions of other people around the world.

(I'd be interested in other examples of pointer-less interfaces.)

[How much of this "tons of research on interfaces that don't include any element of WIMP" is for general purpose information handling tasks? Action games are obviously a different sort of case. Can you please provide references to some of this literature? I am already familiar with JefRaskin's stuff, and I think one can argue that insofar as it abandons WIMP features it is flawed, even if it contains lots of interesting ideas, such as loosening up on the concept of file, insisting the computers should start up right away, etc., (Again, except for the "I" -- the "I" is the weakest part, so abandoning that is nowhere near so big a deal)]

I just checked out TheHumaneInterface's demo. Not bad at all. I mean, it's just 2 1/2 dimensions instead of 3+ D, and it doesn't seem to have any organization to it at all, and generally has lots of problems, but there're lots of details that seem useful.

So about the UI literature. You can start with ZoomableUserInterfaces, MorphicInterface from SelfLanguage, and NakedObjects. Once you've done that you can always go trawling through HumanComputerInteraction on CiteSeer.

[Yes, I knew about zoomable interfaces, and looked briefly at these other things. They all seem to involve a lot of components of the WIMP model. Let's see, a few points: zoomable interfaces are useful in some cases, but when you are looking at a lot of text there is a limit to that. There are other cases where such interfaces are more useful. Probably zooming will be used in subwindows of a larger overall task a person is doing that may require other windows. As for "morphs", they appear to be windows, potentially very lightweight ones, that can be reconfigured and nested. This is all in the WIMP paradigm, as far as I can see. It certainly is a potential advance from how that paradigm is currently employed. I would also mention work of some of Schneiderman's [sp?] other students, such as the stuff on TightlyCoupledWindows? and SnapTogetherVisualization?. All of this stuff is potentially good, but to me it all seems like it is inside a WIMP paradigm: Windows -- regions where stuff happens, that can be moved, resized, put aside, brought back, used with other windows simultaneously. Icons -- well, little pictures that represent things and which provide affordances for direct action on those things (the problem here being "what does that picture mean"? Menus -- lists of possible things one can do on an item or in a context, provided to a user as a set of things they can choose. Pointers -- well, just the mouse here. All of these things have all of those things. The "NakedObjects" talk about how they have a lot of right click menus, as opposed to menus stuck onto a window. That seems to be a minor point. There probably should be both although often enough the latter isn't really necessary. But it provides, if done well, two levels of affordance. Always having a list of possible topics available as affordances for further action could be quite helpful]

zoomable interfaces are useful in some cases, but when you are looking at a lot of text there is a limit to that.

Care to explain what you're even talking about?

blah blah windows blah windows blah windows

Zooming eliminates the need for windows. And you can't know very much if you're still talking about these windows. Windows concentrate the user's attention on the interface instead of on their task. Windows are the work of megalomanic programmers pissing on users. Windows are the tools of Satan.

blah blah windows blah == blah regions blah blah

Oh no! You don't get away with taking everything that's quintessential about windows out of them and calling what's left over "windows". That's like trashing a Ferrari until it runs like a Lada and then saying it's still a Ferrari because it performs the same task ("grouping content").

The AlternateRealityUserInterface (Self UI) specifically had NO windows. Every object was a live object. There was only ever the live object. There could not ever be more than one representation of the object, even on separate machines. There was no interface to the interface (no "close the window" "minimize the window" "resize the window"). There were no windows period.

I hate windows. I really, really hate windows.

And? This still does not address the page's statement, that WIMP is good enough. Many of us despise Microsoft on general purposes, but continue to use their products to produce professional work for which we are paid. That makes those tools good enough, wouldn't you say?

He said "windows", not "Windows". If you despise Microsoft, don't buy into their embrace and extend of terminology. And no, that doesn't make those tools good enough. It makes them better than nothing, which is different.

<ahem> To repeat, many of us use a windowing interface OS daily to produce professional work for which we are paid. That makes those tools good enough, wouldn't you say?

windows, yes, Windows, no. :-) I do sympathize with the point of view that there could be something better, on the other hand. There have been very few workable new ideas in GUI since the 1970s. There's tooltips/hover-help, that's ok. Rooms never quite made it, don't know why (I still wonder if they may become important). No one cares about PieMenus.

This "new" stuff about EEG-controlled interfaces also dates to the 1970s, BTW, but is cheaper and more noise immune and has faster DSP processors, is all. And it's still futuristic rather than helpful.

View edit of June 20, 2005 or FindPage with title or text search