AskTog: Interaction Design Solutions for the Real World
Interaction Design Section   Living Section   About Bruce Tognazzini
NN/g Home > AskTog > ReaderMail > December, 1999 Ask Tog, December, 1999

Ask Tog Reader Mail

 Cursor Keys Cursed
 Separating the Men from the Toys Revisited
 Help! Ask Tog Found Crude and Ineffective!
 Help! Ask Tog Found Really Crude and Really Ineffective!
 Retreiving Email Continued

Cursor Keys Cursed


When I’m typing, I use the shift-arrows a lot to select text. Without reaching for the mouse, I can arrow to the beginning of my intended selection, then hold down the shift key and start hitting the right arrow until I get to where I want the selection to end. If I’m feeling adventurous I’ll hold down the option key and select a whole word at a time. And if I’m in a big hurry, I’ll hold the right-arrow key down so the key-repeat does the work instead of my finger muscles.

Of course, sometimes when I’m selecting text that fast, I’ll overshoot my target end-point.

On Windows, if I hit the right arrow too many times, I can just keep all the modifiers held down and hit the left arrow to reverse direction and fix the problem.

On the Mac, if I select too much text I’m hosed. If I try to reverse direction like above I end up _extending_ the selection at the beginning -- back where I started. This is just stupid. I already chose the _start_ of my selection; I left it where I wanted it and now I’m working on the _end_! At this point, hitting either arrow key _extends_ the already too-long extension and you’d might as well give up and reach for the mouse.

I found this very subject described in the Macintosh Human Interface Guidelines (p. 295) and could scarcely believe my eyes: It described both behaviors and then leaves it up to the application! In “many applications” it acts like Windows, but in “some applications” it doesn’t.

(If only that were true! I haven’t found any Mac applications that work like Windows.)

If you’ve got the book handy, look at Figure 10-20 on p. 296 for a perfect illustration of the behavior I wish I had.

Am I missing something? There are no authors named on the Macintosh Human Interface Guidelines”, Tog, and I’m not sure you were still at Apple in ‘92 when it was published. But even if you didn’t write this book, in my mind you wrote “the book” on GUI. Can you explain why Apple chose to implement the more inefficient behavior themselves, and then punt to the developers on the larger issue? Maybe I’m missing something?

-Jeff Kandt

Lets break this down into three subjects:

  1. On the Macintosh, selection sometimes works in a bizarre way.
  2. This way is offered as an option in the Macintosh Human Interface Guidelines
  3. Mice are inconvenient

1. Bizarre Macintosh selection experiences

While I don't doubt for a minute that you are getting the effect you describe, I have not been able to duplicate it. I have tried Word, GoLive, Tex-Edit, Netscape Communicator, and Filemaker Pro, and, in each and every case, they have worked according to the "good and proper" method you described. Might you have some friendly little piece of OS-add-on shareware inserting itself into the stream? Or perhaps you just use different software than I.

2. The Macintosh Guidelines

The 1992 edition of the Guidelines do, as you pointed out, offer an alternative behavior of cursor-key selection. Bad, bad Apple. The 1987 edition I authored offers no such alternative, nor do either my or Chris Espinosa's earlier ones. And no alternative so out of whack should be offered.

The difference between earlier editions and the 1992 one is that the latter was assembled by committee led by writers, rather than being written by a designer. When you assemble such a committee; you often end up reportage instead of decisions. "In many applications [it acts this way]....In some applications [it acts another]" is excellent verbiage for a bug report but really, really lousy for guidelines. Not that there shouldn't often be variability and breadth, but this ain't one of those times since it violates one of the fundamental principles of design: Levels of Consistency.

And speaking of the Macintosh Guidelines, even where the Guidelines are extremely specific, such as calling for the Shift key to be used for continous selection and the Command key for discontinous selection, guess which application has had it wrong for the last fifteen years? Yes, the Finder. And, yes, I did try to do something about the Finder's transgressions while I was there. The engineers just didn't give a damn.

3. The convenience of mice

...If I’m in a big hurry, I’ll hold the right-arrow key down so the key-repeat does the work instead of my finger muscles.

If you are in a big hurry, you should be using the mouse, not the keyboard. Using the keyboard cuts productivity in half. Yes, I know this does not match personal experience, but personal experience is based on your subjective sense of time. If you have a disinterested third party monitor you with a stopwatch, you will see how slow cursor keys make you really go. How can our personal sense of time be so far off from what the stopwatch tells us? Cursor keys are about the most engaging devices known to man, and the time spent twiddling them just fairly flies by, while the hours and hours spent looking for the mouse (<2.5 seconds, actually) seems to drag on forever.

It's also important that you don't misconstrue this research about cursor keys to necessarily extend to command keys. Under two circumstances, command keys do very well. First, when they are used in conjunction with a mouse, offering the user two-handed input. An example would be selecting text with the mouse, while acting on the text selected with command keys. Even as you are zeroing in on the text with the mouse, your opposite hand is finding the key you will press.

Second, when the command replaces several separate actions, acting, in effect, as a macro. A counter-example of this is the almost unversal use of Control/Command P to initiate the long, drawn-out process of printing. ("Yes, I only want one copy. Yes, I want to print all pages. No, I don't want to print to either a file or to a telephone answering machine in Kentucky."). What Control/Command P should do is to print the document without further input. That saves time.

Now, here's the deal. I don't want any letters about cursor keys. None. Nada. Rien. Between Larry Tesler and myself, we have methodically performed this research twice. I've written three previous columns on it over the last decade. No question exists that cursor keys are dead-slow when tracked by a stopwatch—although I have received dozens of letters from people insisting that cursor keys are faster, none have offered a shred of evidence beyond "I can just feel it, so it must be true." 'Tain't so, Magee, and I don't want to hear about it. But if you want to hear about it, please feel free to click on the following links. You'll hear more than you ever wanted to.

Previous columns on cursor keys:

Keyboard vs. The Mouse, Part 1, August, 1989

Keyboard vs. The Mouse, Part 2, October, 1989

Keyboard vs. The Mouse, Part 3, September, 1992

Separating the Men from the Toys Revisited

Dear Tog,

I found your interesting site from a link in an email leading to your article “Separating the Men from the Toys.” It was quite interesting and well-written, with some good points made.

However, I objected to the apparently gratuitous gender analysis. I’m well aware of some people’s (and even my) love of arcane complexity, which makes things more complicated than they need to be. I was not aware that it was engaged in mainly to keep women out of fields of endeavor. This contention seems mainly insupportable, and it made me feel tired. This type of gender “analysis” is currently widely accepted, but impresses me as a rather bigoted catchall explanation for many things. I believe it has a net destructive effect and I hope it runs its course and disappears soon.

Otherwise, keep up the good work.

Regards, Mark G

I claimed keeping women out was the "third" reason, not the main reason, for embracing spurious complexity. Nonetheless, whether it be de facto or de jure, a great deal of anti-female bias does exist in the curricula and practices of aviation and engineering.

When you have a field that is filled with a kind of complexity to which one gender is drawn and the other gender repelled, and that complexity has absolutely nothing to do with success or failure in the central task, you are left with only two explanations:

  1. The dominant gender set out to exclude the other and came up with a master plan that seems to be working.
  2. The dominant gender just kind of drifted into adding exclusionary complexity and, having seen the discriminatory result, just doesn't give a damn.

I will let you choose what you believe happened in both flying and engineering. I would put my money on the second explanation, since I haven't met a whole lot of pilots or engineers who were misogynists, but I've met a whole lot of both who don't give a damn that their professions are so unevenly balanced.

It also may well turn out that women eschew both professions because women just don't like flying and programming, even when reduced to their essentials. However, it will be hard to prove that one way or the other as long as we continue to actively exclude them.

Of course, men are not the only ones who practice exclusion. Men have only to walk into the perfume section of a department store or the styling salon at "Sally Mae's Pink Palace of Beauty" to feel the full force of an "I don't belong here and I've got to get out now!" environment. Just be aware that we men do the same thing, and it can be just as repellent.

As a guy, I might not mind the current state of affairs—after all, it increases my odds of employment—except I don't really need any extra complexity, not when I'm in front of my computer, nor when I'm soaring above your house on a rainy night. No, I need maximum simplicity, and I'm not getting it. Think about that the next time you hear a sputtering engine in the dark sky.

Help! AskTog Found Crude and Ineffective!


Answer this question please. If you are a good interaction designer than why is your site radically hideous and completely void of tactility ? Tog, judging from your silly questions you are just a pontificating web idiot asking complete buffoonery, get off-line and go home forever, leave interactive design philosophy to CA and ID.

Greg P

Hi Tog,

Simple question: is it possible to get your new articles delivered straight to my email? I was totally blown away finding your page and wouldn’t want to miss a bit! ..but I’m such a bad surfer I never visit sites unless the links drop to my inbox.

Best regards, Niko N


Tog, Have you considered setting up a mailing list to notify people when a new article is posted? It would be very helpful.

Thanks! David H

I admit it. I have lost control. AskTog has grown, both in content and in readership, beyond my abilities to maintain the kind of quality that I would like. It is missing three fundamental features that would make readers' lives easier:

  1. Search engine
  2. Mailing list
  3. Usable and useful table of contents and index

I simply cannot set up a search engine. I have tried. I even downloaded a trial copy of a search engine "so simple even a child could install it." Whose child? Einstein's? Even though the instructions were filled with a lot of "just"s and "simply"s, I just simply couldn't understand a thing they were saying and, worse, when I finally did install it, but it flat-out didn't work, I had neither the theoretical framework nor the tools to fix it.

So. If anyone out there would like to volunteer to work on any or all of these three items, please contact me. As you can see, the first two items call for some skill in advanced engineering. The third calls for some skills in editing.

I would love to give you, in return, 50% of the profits of AskTog, but that would mean you would have to pay me around $500 per year (since the site loses around $1000), so I suspect that is out. Perhaps you would settle for my undying and very public gratitude?

Help! AskTog Found Really Crude and Really Ineffective!


Answer this question please. If you are a good interaction designer than why is your site radically hideous and completely void of tactility ? Tog, judging from your silly questions you are just a pontificating web idiot asking complete buffoonery, get off-line and go home forever, leave interactive design philosophy to CA and ID.


Gee, and I only thought the site was homely. It looks the way it looks because my ISP charges by the hit and by the megabyte. Both of those force the site owner to make a choice: Make it pretty-and-small or make it plain-and-big. I've chosen to make it plain and big since they wanted to charge me four times the money to double the size. If anyone can come up with a design that will make it pretty while still all but eliminating graphics, I will be eternally grateful.

Retrieving Email Continued

In the latest AskTog you wrote about “horribly broken” email:

Ever wished you could call that email back or cancel it? Why can’t you? Well, at least one reason why is that the postal service doesn’t allow you to “unmail” a letter, and them must have a good reason. They do: If they allow everyone access to the mailbox, people will steal each other’s mail. That’s the sole reason, one that has nothing to do with email.

No, this is not slavish copying. Internet email systems don’t allow canceling of mail for exactly the same reason as the postal service: it would mean allowing everyone access to the outgoing mail stream, and people would steal (or at least cancel) each others email.

POP3/IMAP are authenticated protocols, just as you need a key to open your post office box. The SMTP delivery protocol has no authentication, just as anyone can send mail to your post office box. Those “From: “ lines are just as reliable (ie not all except for convention) as the return address on an envelope.

The deluge of spam is causing a lot of people to rethink these assumptions. There’s a general trend to requiring authentication at the first delivery hop so that only corporate employees/ISP customers/whatever can insert new messages into the outgoing stream. The real world equivalent would be the post office removing all those public mailboxes, so you’d have to post your letters at the local office...and provide some form of ID when you did so. On such a system it would be easy to add a short delay on transmission and the capability to cancel messages.

There’s always a downside: if you are visiting another site, you won’t be able to send mail as yourself while you’re there, since you can’t authenticate yourself to the local server. Frequently a firewall will prevent you from connecting to the remote server where you do have an identity, since (in the minds of sysadmins at least) the whole point of having a single mail server is lost if you let people use any system they like. Some email systems, such as AppleLink, were built with the ability to retrieve email. It worked well and saved a lot of people a lot of embarrassment. It is about time such a feature became a standard part of Internet mail systems.

AppleLink could only do it because everyone, both sending and receiving, was registered with a central server. Microsoft, AOL, and other would-be monopolists love the idea of requiring everyone on the Internet to sign up with their service before being able to contact other members (e.g. the Instant Messaging battle) but so far have been unable to enforce this. Be careful what you ask for—you might get it.


Hugh Fisher

The last thing I would want to see would be a further erosion of anonymity and privacy on the net. However, were alleviating embarrassment a priority, I suspect the great minds of cryptography could figure out a way that I could "retrieve" (destroy) my own email. Perhaps we could supply a unique pair of keys for each email, with the retrieval key held on the sender's machine. Only if a second email shows up with the retrieve instruction and the appropriate key would the email would be deleted.

Should this scheme have flaws, I'm sure another one could be constructed; this is not an insoluble problem. It just, apparently, is not an interesting one.

Email needs to be looked at with a whole fresh perspective. As currently embodied, it is the very manifestation of computing in the 1960s, the decade that launched the Y2K bug.

About the only system I could think of that would be worse would be one where you were expected to memorize completely nonsense addresses consisting of nothing but numbers, and then would only be able to communicate with the other person if they happened to be sitting in front of their receiving terminal at the time. Maybe to make it even worse, you could eliminate all possibility of communicating via the visual medium by restricting it strictly to voice.

Wouldn't that be something!

Don't miss the next action-packed column!
Receive a brief notice when new columns are posted by sending a blank email to

return to top

Contact Us:  Bruce Tognazzini
Copyright Bruce Tognazzini.  All Rights Reserved