a companion discussion area for blog.codinghorror.com

The Opposite of Fitts' Law


Which is why Chrome’s tabs own Firefox’s tabs when the window is maximized: just swing the mouse to the top of the screen and you’ll hit a tab.

And also a reason the Ribbon interface on Office works so well, everything that’s really important is also really large.


And this is why HTML needs a “weak” tag to counterpoint it’s “strong” tag.


@Tiago Almeida Also, in some of the dialogs you don’t have a delete option so you end with lots of “New Folders” everywhere. Pretty annoying.


“… frequently on web sites, I want to find a “dead spot” I can click on just so I can give the browser tab focus”


The neurons in the agreement center of my brain are firing at maximum rate and will soon wear themselves out.

It is surprisingly hard to find those dead spots on many pages, and that goes for regular apps to, not just web page apps.


Another funny example is the Cartoon Network short about Birdman’s Coffee Break:


I tend to cope with tiny, poorly-placed buttons by learning the keyboard shortcuts. For instance, remember Microsoft PowerPoint 2003 with the button to play the slideshow? The button was about 10x10 pixels, in the bottom left corner. Fortunately, F5 and Shift-F5 is an easier way to play slides from beginning or current slide, respectively. Unfortunately, function keys themselves are tiny or require the Fn key on some laptops.


That Birdman snippet makes the point in an excellent way!


Sosh Sosh: As far as I know, all available research suggests that the vast majority of people (other than a subset of SuperGeeks) run everything maximized whenever possible, and essentially page-flip for their application/task changes.

Apart from modals, and things like chat clients that don’t maximize, UI can usually be placed at screen edges, for a very large number of applications.

Back on the main topic, I notice that a multi-monitor setup makes Fitt’s Law significantly less useful in both Windows and OSX; neither does the taskbar or dock/menu on the second screen without hackery, and you end up with windows, typically, maximized with one edge floating in space; your mouse won’t stop on it.

Awkward, huh?


It’s amazing how often this error pops up – I list some more examples here: http://www.brool.com/index.php/implications-of-fitts-law – but the one that cost me the most time was having “Mark All as Read” and “Mark (One Message) Unread” right next to each other on the Outlook menu. Arrrgh!


A very well written article outlining the importance of design.

Kudos to the Wordpress designers for getting this right!


Oooh, a chance to rant about bad ideas:

Control-W vs. Control-Q in firefox. The difference between close tab and close Firefox? Quite large. Which is why they’re next to each other on the keyboard, of course. Not to mention, how often does one even want C-Q? Why does it even have a shortcut, isn’t Alt-F, Q enough?

One of my requirements for an e16 theme (an X11 window manager) is that close window be on the opposite side as minimize/maximize/etc. Mac OS did this right, through 9.x; no idea (other than brain damage) they changed it in OS X. Oh, and made them color coded only (except on hover), just to show what Apple thinks of the colorblind. Microsoft hasn’t managed to get this one right yet (but at least theirs aren’t color only).

Putting close buttons on each tab, that was a great Firefox idea. Especially when tabs get small. Thankfully, can be disabled (is it off by default now? not sure).

The caps lock key, a feature never used except accidentally, gets a better spot on the keyboard than Control (and maybe even better than shift…). Fixable in software, thankfully.

Braindead keyboards with extra-large backspace or return keys, eating into where the other one and the pipe key should be.

And finally… Many UPSes (many APC UPSes, also many other brands) that use the power button as a test button. Release it too quickly, OOPS, just turned off the power. Press it longer for the test. Exactly the opposite of a PC power button (short press = software action, like safe shutdown or hibernate, long = bios override to cut power) and exactly the opposite of sanity. With UPSes you press the eject seat button for 4s to turn on the radio.


I like to think of buttons like “Delete” as those little shields that are over buttons that launch missiles. The ones you have to flip up before pressing the switch or button and launching the missiles. The confirmation dialog is what should have the real button that does the launching.


Ever notice on your typical QWERTY keyboard how ampersand and asterisk are right next to each other?

Back in my university sysadmin days I was removing quarterly student accounts from my department’s VAX 750 running 4.2 BSD (yes, I am that old). I knew it would take a long time, so I ran the following commands:

cd /
rm -rf foo &

After a few seconds of wondering why the prompt didn’t come back, I turned to my student assistant and said “Get the backup tapes.”


Can we then have some sort of standard for the order dialog buttons go in? It’s bad enough pulling the ejector seat handle by accident, but when people swap the ejector seat handle for the joystick it’s a totally different problem.

All GUIs except Windows 3 have the fatal problem of placing the “Kill window” button right next to the “minimise/maximise” buttons. Even worse, the “kill window” button is in the corner, so on a maximised window it’s really really easy to hit.

I’m forever accidentally shutting tabs in Chrome because of the tiny “x” on every tab. Evidently lots of people at Google do this because there’s a “recover closed tab” option.


This sounds a lot like security through obscurity. If you don’t want users to click on a button then don’t give them a button, rather than give them a button that’s slightly more difficult to hit. It won’t matter how small the button is, or how far from the current mouse position, it will still be hit by accident. If you need a self destruct option, force the user to invoke it explicitly and ask for as much confirmation as possible.


Thanks for this excellent post.

I intended to put Larson’s “Wings Fall Off” cartoon in the first edition of About Face way back in 1995, but Larson wanted a couple thousand bucks for rights to it.

Alan Cooper


Google’s services in general have a problem with this. Take a look around. GMail, Reader, Voice, Wave, and many others. It’s pretty much a constant theme of the Google (non-search, of course) interface.


That’s why I thought Fitt’s Law applies well to the recent runaway Toyotas (I talk about it here: http://www.rioleo.org/the-prius-syndrome.php). Braking and accelerating are two opposite tasks, and yet they are so close to each other that driver error, though rare, may actually be explained.

Oh and by the way, the Gmail Undo feature has been a lifesaver for so many occasions!


Fitt’s law always seems to be used to justify tiny little click targets at the edges of the screen. Sure, they work great if your pointing device is a mouse or trackball. They’re not so good when you try to use the same application with a stylus on a tablet PC.


“Good thing we can easily undo a sent mail! Oh wait, we totally can’t. Consider my seat, or at least that particular rash email, ejected.”

  1. Go to GMail.com (logged in).
  2. Top right corner, there is a green flask, click on it (OR click on the Labs tab in the Settings page).
  3. Scroll down until you see Undo Send, enable it.

Congratulations for the spot check. This is why I won’t pay for news.


@Elwin: See http://stackoverflow.com/questions/1495708/clear-form-button-in-html-do-we-really-need-this - the reset button definitely has its uses.

@Papa Smurf: Sure, using keyboard shortcuts stops you mis-clicking. Now all you gotta do is stop mis-typing.

@Fabricio: While that is true, it has the unfortunately side-effect of not letting you see the full page title. I’m glad I can turn the title bar on in Ubuntu. Also, I rarely click tabs, I just use the scroll-wheel over them; it’s a fairly large area.