Mouse Abuse

Question: Should we engineers go postal when we see a civilian take their hands off the keyboard, grab the mouse, and slowly do something that <Tab> or <Enter> could have done 750 times faster?

Keyboard "shortcuts" like <Tab> to go to next field or <Alt>+<Tab> to switch windows are interaction idioms (yet another useful AlanCooper coinage). Like idiomatic expressions in ordinary language, they do not really make sense but have acquired meaning through conventional association that has little or no relationship to their literal meaning (if any). They are not in any reasonable sense intuitable.

Supporting <Tab> and <Shift>+<Tab> is an absolute requirement for all Windows apps and Web forms not only because it’s the standard but because many (though certainly not all) users expect it. The ones who don't know about or expect it are unaffected but the ones who routinely rely on it are deeply affected if you don't do it or do it right. Getting it right is more than just supporting field-to-field tabbing but supporting it in a way that enhances usability, which means getting the right tab order that makes sense with the likely use of fields, sequencing vertically or horizontally depending on the workflow and how user thinks of the fields, planning the right field behavior onfocus and onblur, etc. For some users (heavy keyboarders, production staff, etc.) in some applications, these details can make as much as 40-50% difference in end-to-end productivity.

Back to Jeff's original post, the preference for keyboard or mouse is not only an individual trait but it is heavily conditioned by context. Some evidence suggests that, aside from habit and manual skills, users shift between mousing and keyboarding to fit how they see the immediate task or the upcoming one. (For example, changing from one idiom to the other when they reach the bottom of a screen with content "below the fold.")

Contrary to the claims of keyboard fanatics and the now largely discredited early Apple research, one technique is not necessarily or universally faster than the other. In general, most people can type successive keystrokes faster than a move-to mouse, saccade*-and-select, move-back, but this is not always the case nor is it the whole story. For example, all but highly trained clerks doing repetitive heads-down data entry make a distinct (sometimes fairly long) pause at the end of entering each field anyway during which they typically reread what they have typed, confirm its content, form the intent to move on, mentally select the method (interaction idiom), and execute it. In most work contexts where there is not high volume input, the impact of the user choosing the "wrong" idiom is minimal.

In any case, as InteractionDesigners (or developers subbing in as interaction designers) we have a responsibility to understand our users and the context of use. Responding to the Phlip remark about expecting keyboard competence and weaning users off help, we always have to see the "TimeAndMotion?" thing in the larger context. In a PointOfSale? app, the total number of fields and user steps per customer is typically relatively modest. The savings of seconds here or there through keyboard rather than mouse operation will be unlikely to have much impact on the length of the queue. Far more important is presentation and interaction design that reduces the chance of error and that promotes rapid learning and acquisition of skill (high turnover means many users will be still ramping up). Just one need for a supervisor void can cancel out the gains from thousands of keyboard-only steps. Or consider the user typing-and-tabbing at a snail's pace because they are trying to figure things out on a poorly conceived screen. Or the clerk flummoxed by a message that makes no sense....

As we consultants say, it all depends. By exquisite attention to what nurses actually do when charting and managing patient care, right down to shaving milliseconds here and there, we and the crack team at McKesson? were able to speed these tasks by almost 50% while cutting training time to a fraction of what it was.

-- LarryConstantine [from the agile-usability mailing list]

* note Larry uses the word "saccade" for both eye movements and mousepointer movements...

Keyboard shortcuts are necessary for accessibility requirements, and may be legally required in some jurisdictions. All functions must be accessible by keyed input. -- SunirShah


Most of that is very reasonable, but:

from above: "In most work contexts where there is not high volume input, the impact of the user choosing the "wrong" idiom is minimal."

from above: "Keyboard "shortcuts" like <Tab> to go to next field or <Alt>+<Tab> to switch windows are interaction idioms ... do not really make sense but have acquired meaning through conventional association that has little or no relationship to their literal meaning (if any). They are not in any reasonable sense intuitable."

Beg to differ!

Tab was invented for typewriters, and never meant anything but "move to start of next field!"

The meaning for <Alt>+<Tab> is somewhat more arbitrary, but still far from completely arbitrary.

Begin with the modifier <Shift>, actually. Originally it meant to modify a letter to an alternate form. As children we learn that each letter in English has two forms, upper and lower case, both of which are supported on typewriters. If <Shift>+<Letter> means anything at all, it is in fact intuitive that it would mean a capital version of the letter. It's that or nothing (well, we might guess that it means cursive or italic, I suppose, if we forget that we learned capital letters first in school, but exactly one experiment sets us straight, if so).

What, then, might <Shift>+<Tab> mean? Hmm, some modified kind of tab. I don't know about you guys, but I learned <Shift>+<Tab> because I needed to move backwards in a form, wished there was a way to do it, eventually decided that whoever was responsible was an idiot for not allowing me to move back a field via <Shift>+<Tab>, idly tried it, and was astonished to find that it worked as I had wished! Well, in retrospect, what on earth else would it mean to modify a tab?

Before the invention of the PC, the MIT and Stanford researchers decided that the world needed more modifiers, yielding Control-Alt-Shift-Meta-Bucky-Cokebottle, descendents of which we find in <Alt>, <Windows>, <Cloverleaf>, etc.

To the uninitiated, who knows what any such modifier does? But if one of those exotic modifiers is used with <Tab>, my intuition is that it should do some kind of movement as <Tab> does, but bigger somehow, since forward and backward are already taken care of.

Admittedly that might be several kinds of "big movement", but again, a single trial will show me which kind of "big movement" happens; switching windows is in fact a big movement.

The point is that this, therefore, is absolutely not a purely symbolic idiom, with no connection between the symbol and its referent other than convention. There is in fact sufficient iconic connection that someone actually might guess ahead of time both that <Alt>+<Shift> exists, and also what it does.

I am not claiming that this is quite natural enough that everyone should be expected to immediately intuit this; obviously they won't. But it certainly is too transparent to claim "little or no relationship to their literal meaning (if any)". The relationship is obviously large enough to have strong mnemonic value for anyone who already understands <Tab> and <Shift>+<Tab>. There is no mnemonic value in completely arbitrary connections.

The real problem here seems to be that LarryConstantine has forgotten the meaning of <Tab> in the first place, forgotten typewriters, so if <Tab> is merely some random symbol with real inherent meaning, obviously anything combined with <Tab> will be as well. But 'taint so.

The above reasoning is beyond stupid. Or it would be beyond stupid if it wasn't deliberate sophistry.

You're arguing that <TAB> is not unintuitive because you know its meaning from having learned it. What the bloody hell kind of argument is that? You've conceded everything! If you had to learn the meaning of <TAB> then it was never intuitive in the first place. And who gives a damn about typewriters anymore?

Furthermore, the meaning of <shift>+<tab> is NOT obvious or in any way deducible from the separate meanings of <shift> and <tab>, it is purely idiomatic. The whole thing about modified movement and bigger or smaller movement is just so much rationalization and useless scholasticism.

The iconic meaning of <shift> to me is not in fact "modified" but BIGGER and LOUDER since capital letters are invariably bigger than lowercase letters. So the idea that <shift>+<tab> would go backwards is entirely counter-intuitive and I find the entire discussion about the supposed iconic meanings of shift and tab to be idiotic in the extreme.

Finally, regularity and simplicity are the only things necessary for mnemonic value. And none of those imply intuitiveness, obviousness or even predictability. -- RK


Ok, comment sunsetted. Too obscure, you know.


We should go postal when mouse misuse is injuring the civilian. That person needs to do any of several things:


CategoryInteractionDesign


EditText of this page (last edited September 6, 2005) or FindPage with title or text search