Why is the Close button in the upper right corner?


Chris wants to know how the close button ended up to the right of the minimize and maximize/restore buttons. "In OS/2, it is on the left, which left the two other buttons in place."

I don't know why the Close button went to the upper right instead of going to the left of the other buttons, but I'm going to guess. (That's what I do around here most of the time anyway; I just don't usually call it out.)

Two words: Fitts's Law.

The corners of the screen are very valuable, because users can target them with very little effort. You just slam the mouse in the direction you want, and the cursor goes into the corner. And since closing a window is a much more common operation than minimizing, maximizing, and restoring it, it seems a natural choice to give the close button the preferred location.

Besides, maximizing and restoring a window already have very large targets, namely the entire caption. You can double-click the caption to maximize, and double-click again to restore. The restore even gets you a little bit of Fitt's Law action because the top of the screen makes the height of the caption bar effectively infinite.

Comments (47)
  1. configurator says:

    And maximizing would hardly ever benefit from Fitts's law, because the window isn't maximized.

  2. William says:

    @configurator:  Well, you still get the entire width of the caption for maximizing.  If the window's not maximized, it's pretty tough to take advantage of Fitts' Law because it's most likely not up against any window edges.

  3. Now, whoever wrote the separator between the message list and the preview pane in Outlook 2010 needs to be smacked by a 23 inch monitor on the head. Anybody tried the resize by mouse?

  4. Jonathan Wilson says:

    I have a great book called "User Interface Design for Programmers" which devotes a section to Fitts' Law

  5. Anonymous Coward says:

    Doesn't Fitts's law make it more surprising that the top screen edge is completely handed over to the titlebar? Because if you think about it for a second, restoring isn't that common an operation (not compared to accessing the menu for instance) and commonly all the titlebar does is display the title of the active document, the same one highlighted on the taskbar. It's of course possible to have a restored active window in front of an inactive window, but does this justify sacrificing the screen edge to this waste of space?

    And don't forget that the top edge originally was intended for the taskbar, and was moved down for technical reasons, rather than Fittsian concerns. I think that given that when Windows 95 was designed Fitts's law wasn't given as much thought as it should have been (note: the start button was actually 2 pixels away from the corner), it's much more likely that the close button was simply added to right of the other buttons because that's the first place where developers tend to put a new button. It didn't go in the top-left because it didn't replace the window menu (which incidentally is useless in almost all cases and also accessible as the titlebar context menu) and a little space between the close button and the others was added possibly as a visual cue to warn Windows 3.1 users.

    With Fitts's law in mind, the menu might have been a better use, possibly with the window buttons on it. On the other hand, in more and more cases people don't actually use the menu any more (for various reasons) to the point where e.g. Chrome's designers opted to but it in a toolbar button, so maybe it would have been better to simply hand (most of) the edge to the application.

  6. RP says:

    I don't agree with AC's statement that the window menu is useless.  You can double-click it to close the window.  That's not useless, because sometimes your cursor is much closer to the lefthand corner than the right, so it's then quicker than the newfangled way of closing windows.

  7. John says:

    Metro – taking the windows out of Windows

  8. Tim says:

    Windows 98 also added some Fitts's Law juice to minimize because you could click the button in the taskbar on the bottom of the screen to minimize.

  9. My shortcuts:  Alt+Space, then "r", "x", "n", or "c".  Much faster if I'm not on a computer with a good pointing device.  Even if I have a good mouse, I still use them since they are fast, and my left hand is usually able to quickly access these keys.

    @Antonio:  It's not just Windows XP.  Windows 7 has the same problem when you plug/unplug an external monitor/projector.  All the maximized windows are resized to completely fit the screen they are on, which exposes the rounded corners again.  After plugging/unplugging the projector, you have to manually restore and maximize each window to hide the rounded corners.  So annoying!  I wish they would fix it, but I guess they don't care; they're too busy Metro-izing the world (which I'm sure will introduce a whole host of new bugs that they won't fix).

    @AC:  "commonly all the titlebar does is display the title of the active document … does this justify sacrificing the screen edge to this waste of space?"

    Google Chrome devotes this space to a more useful list of tabs:  switching/closing tabs is as easy as slamming the mouse to the top of the screen and clicking.  Someone should tell the Internet Explorer team, because IE wastes the space.

  10. Gabe says:

    Unfortunately multimon inverts the applicability to Fitt's law. Not only might your once-inifinte target not be infinite anymore, but going just one pixel past it might make your cursor disappear from the monitor entirely! Overshooting your target means you have to look on another monitor to find your mouse cursor, then figure out how to get it back to your original target.

    Of course, this is exacerbated by the problem of having huge monitors. Makeing the menus stuck to the top of the monitor made lots of sense when you only had one monitor and it was only 340 pixels tall. When the menu bar is thousands of pixels away on a completely different monitor from the app you're working on, it makes somewhat less sense.

  11. Tergiver says:

    I would guess that they chose the right-hand side because there is already a "close button" (double-click the system menu) on the left-hand side.

  12. Tim says:

    Nate: you can't reach that conclusion as if Microsoft consisted of a single individual. There were many people responsible for building Windows 95; undoubtedly some that were aware of Fitts's Law and some that were not.

  13. John says:

    In regards to the title bar being "wasted" space, I would argue this only became a problem relatively recently due to the shift to widescreen displays.  Oh how I miss my vertical space.

  14. Spencer says:

    I think that it's beneficial to have the close button in the upper right hand.  A minor miss when clicking the "File" menu could cause you to close your window.

  15. Evan says:

    @Anonymous: "…commonly all the titlebar does is display the title of the active document, the same one highlighted on the taskbar…"

    Not to say that perhaps rethinking things a bit would be beneficial, but: And also commonly, truncated in the taskbar so as to be useless. Or, in Win 7, perhaps not even visible at all because you're using icons.

    In fact, for similar (truncation) reasons the removal of the page title from the window title bar is a big part of why I hate the Chrome UI and won't use it on Windows.

    @Anonymous: "And don't forget that the top edge originally was intended for the taskbar, and was moved down for technical reasons, rather than Fittsian concerns."

    But also don't forget that the bottom edge of the screen has exactly the same Fittsian benefits as the top. :-)

  16. The Office UX people abandoned Fitt's law for the "File" button in Office 2010 apps. The Office 2007 orb was a major improvement in that area. If people had problems recognizing that the orb was a clickable button, they should have simply called it "File" but still preserved Fitt's law by keeping at the top. Then again, the Office team has been playing "interface games" after Office 2003. Anyone read "The Interface Games"? It's written by the Windows 8 UX team after getting inspired by "The Hunger Games".

    The Close button of Windows Live Photo Gallery on XP also doesn't support Fitt's law (but did on Vista+). I reported it for 2 releases but the team was apparently clueless.

    [I continue to be amused by your view that any decision that doesn't go your way is proof that the person who made the decision is either evil or an idiot. -Raymond]
  17. I don't think they are evil or idiots and it has nothing to do with just making changes that suit me. It's about removing features – those people are cheats. I continue to be amused by how innocent Microsoft acts ("oh hey what did we do?") when it actively deletes functionality right under our nose after we spend our hard earned money on what was supposed to be an "upgrade".

    [You call them "cheats" and "clueless". That's not evil and idiotic? -Raymond]
  18. configurator says:

    William: That was my point. The maximize button can't reap the same benefit that the close button does because of its location.

  19. Scott H. says:

    Somewhat amusingly, if you make use of the Alt+- Move bug in Windows 95 to move the Start button a little farther down to the bottom left, you realize there's a permanent "frame" around the taskbar that still prevents Fitt's Law from working with it.

    This same issue exists in Newshell for NT 3.51 too, for that matter (which also exhibits the alt+- issue).

  20. Antonio Rodríguez says:

    Yes, the famous Fitts' Law, infamously ignored by the 1px-wide margin around the Start button in Windows 95, and by the rounded window corners of Windows XP, which made you accidentally close a maximized window instead of the almost-but-not-maximized one you had on top :-) .

    With displays getting bigger and bigger, and with features like Aero Snap, it's more usual to minimize than maximize. IMHO, Windows should have an option to *minimize* doing double click in the titlebar, as the Mac has done since, IIRC, System 8 (published in the late 90s). I had a little utility (WinShade) that let me do that in Windows XP, but I have not found anything that works in Windows 7 x64 with the DWM.

  21. Jon says:

    I think the question would be better phrased as why OS/2 placed the close button to the left of the two buttons. Before Warp 4, OS/2 used the same button arrangement as Windows 3.1, obviously from its Presentation Manager heritage. When IBM decided to add a close button in OS/2 Warp 4, they decided to place it left of the minimize button to allow people to maintain the same muscle memory, avoiding complaints about accidentally closing windows.

    Interestingly, OS/2 Warp does preference boxes the same way OS X does, there's no OK/Apply button.

  22. 640k says:

    > I'm going to guess. (That's what I do around here most of the time anyway; I just don't usually call it out.)

    > Two words: Fitts's Law.

    And you would be wrong. As usual, the gui is a *legal* result of "big business". It was licensed from NextStep. No technical aspect went into the decision making. Remember, everyone is a gui expert. Even the board execs.

  23. Nate says:

    Microsoft engineers at the time were unaware of Fitts’ Law. Just take a look at the Start Button in Windows 95 and 98; or the scroll bar of a maximized window, up through at least Windows XP.

    I doubt they were thinking of that when deciding on the “Close” button placement for Windows 95.

    And a maximized window did *not* have an infinitely-sized close button. Slamming the mouse to the corner and clicking doesn’t work, as far as I can tell. I tried it with Notepad in Windows 7’s XP Mode, full screen.

  24. Daniel Neely says:

    xpclient:  Nope.  Neither Google nor Amazon claim to have heard of it.  Would my inner cynic be right for suspecting that such a cheasy sounding parody of a teensploitation title falls under the category of:  "Not to be dismissed lightly, but instead hurled into the nearest waste receptacle with great force."

  25. jcs says:

    Antonio, James: So *that's* how we end up with windows which fill the screen without being maximized! I've been dealing with this problem for *years*,  across multiple Windows releases. I get windows that appears to be maximized, but when I click the close button, I accidentally close the window underneath. What you say about rounded corners make perfect sense.

    Even worse, some applications remember their previous window sizes and positions, and the application persists in its "almost-maximized-but-not-quite" state for weeks until somehow it forgets.

  26. Random832 says:

    "And you would be wrong. As usual, the gui is a *legal* result of "big business". It was licensed from NextStep. No technical aspect went into the decision making. Remember, everyone is a gui expert. Even the board execs."

    Er, what? Yes, Nextstep does also have the close button in the top right corner, but A) there's only so many places you can put it B) there's no indication that MS _actually_ had to pay them for it, or that any UI element was licensed from NeXT. C) Even if all this were true, it doesn't explain the _why_ they chose to license this bit of… technology. They could have chosen to put it somewhere else, getting them out of having to license the "button in the top corner" concept (what, is that patented?)

    Your claim is absurd without any citations to back it up.

  27. I am not entirely sure that "Fitts' Law" was involved at all, even though everyone seems to be quoting it (inappropriately imho) as if it were going out of fashion.

    Fitts' Law simply derives an Index of Difficulty for a target based on it's distance and size.  By that measure the Close gadget would only have been placed in the extreme corner of ANY candidate target area if the intention was to MAXIMIZE the Index of Difficult – i.e. to apply Fitts' Law to calculate the WORST place (on average).

    If Fitts' Law was being applied to determine the BEST place, then taking as an initial assumption that the calculation be made based on the user having their mouse somewhere within the window to be closed as the dominant starting condition, then the close gadget would be in the center of the window to yield the lowest average ID based on Fitt's Law.

    That obviously would be stupid, so the next best place would be the center of the caption bar, which is similarly impractical.  Having decided to place it in a corner – ANY corner, pushing it to the EXTREME of that corner simply guarantees the WORST Fitts' Law ID.

    Much more likely therefore imho is that it was simply a decision based on practicality and aesthetics.

    More windows are closable than are resizable (in general), so placing it in the corner made for the most consistent appearance of the gadgets in that corner.  It yields the most consistent UI and the most reliably placement of the (likely) most often used gadget.

    This makes far more sense than the notion that there was any mathematics involved in the decision, especially mathematics where the results to support the decision yield results that are the inverse of the presumed intent of the mathematics cited.

    No ?

  28. Evan says:

    @Jolyon: "Fitts' Law simply derives an Index of Difficulty for a target based on it's distance and size. … Having decided to place it in a corner – ANY corner, pushing it to the EXTREME of that corner simply guarantees the WORST Fitts' Law ID."

    In the case of maximized windows, you're ignoring the "size" part of Fitt's law and only paying attention to the distance. An object along a top or bottom edge essentially has infinite height; an object along a left or right edge has infinite width. An object in a corner is infinite in both directions. :-)

    A similar principle comes into play when you've lost the mouse cursor. One of the easiest ways to find it is to just move the mouse a whole bunch towards a corner; after a couple sweeps of the mouse, you know it's there.

    Of course this doesn't help when windows aren't maximized.

  29. Anonymous Coward says:

    Tim, we do know for sure that the people who were responsible for Windows 95's general layout weren't aware of Fitts's Law. See the posts above for more detail.

    John, when I was using Windows 95 the resolution was 640×480 and space was even more precious then than it is now.

    Evan, the underlying problem that causes unwanted truncation is ego. The kind that causes people to title their windows ‘Microsoft Excel – …’ and their websites ‘Google Image r…’ although it's much less of a problem than in 1995. Most applications nowadays put the document in front and most websites the article. I just loaded up some pages in Chromium to see how much of a problem truncation is, and it isn't one at all. Google image search is the only exception I could find.

    And yes, the bottom edge would have been very Fittsy, but at no point did Microsoft have the intention of putting oft-used elements alongside it, or in either corner. In any case, most UI elements in applications that could potentially benefit from Fitts's magic tend to be near the top, so I think handing (most of) the top edge to applications is probably best. Of course, applications can turn off the titlebar when maximised, but there's no unified design guideline about whether the window title and icon should be shown or not and where, and you'd have to reimplement the titlebar buttons.

    And when people roll their own UI, the result is often ugly as sin. Just look at Chrome/-ium.

    Raymond / Xpclient, I don't want to discuss terminology, but the fact of the matter is that I picked the window theme / colours because I like them. That the Office UI designers want to force their preferences upon me is incredibly arrogant, and made even more sour by their complete and utter lack of taste. Oh well, I've switched to what was back then still called OpenOffice precisely because things like that.

  30. If I had a pound for each time I've slammed my mouse into the top left corner on W8CP and double clicked, just to switch window to the last metro app and back when I've tried to close a window…

  31. Neil says:

    [I'm going to guess]

    If the law fi(t)ts, guess it?

  32. Skyborne says:

    Yeah, the Amiga put Close in the top-left, too!

    Top-right were raise/lower buttons.  I missed them for a few years (thank goodness for sloppy focus and tweakui), but then monitors got big enough to juggle position instead of depth.

  33. silly says:

    I like how they glow on a mac if you hover the mouse over them. Gives you time to look, locate, and realise. Coming from a windows background that was kind of neat. Maybe windows does it too. Not sure. Anyways, whatever.

  34. Evan says:

    @Anon: "I just loaded up some pages in Chromium to see how much of a problem truncation is, and it isn't one at all. Google image search is the only exception I could find."

    That's funny, because all of my tabs are truncated to at most 3 letters now. Of course, I have a couple dozen of them open, but then again, that's a pretty typical state. Opera gives me a nice ctrl-tab view that shows the full titles of all.

  35. Mark S says:

    Ha, after reading Nate's thing about maximized windows' scrollbars, I just noticed that I'm now annoyed that this is a problem with IE8 on W7.  Le sigh.  

    As for W8, the less said the better.

  36. Tim says:

    "Tim, we do know for sure that the people who were responsible for Windows 95's general layout weren't aware of Fitts's Law. See the posts above for more detail."

    The only way you would know for sure is if you had proof that the same person who designed the task bar also determined the position of the close button on windows. Since you know for sure, can you share this proof with the rest of us?

    [I remember when the hit-testing for the Close button was specifically tweaked to accommodate Fitts's Law (though it wasn't cited by name). So the number of people who were aware of the issue is at least nonzero. -Raymond]
  37. John Hensley says:

    @Nate If you can't close a window by clicking the upper right corner of the screen in Windows 7's XP mode, it might very well be a bug in the mode. In original Windows XP chrome, the sliver between the close button, and the border or screen edge, is part of the close button's target area. First time I've noticed that, really.

  38. ABCDSchuetze says:

    "[…] because the top of the screen makes the height of the caption bar effectively infinite. "

    This argument is nonsense. I usually hear it from MacOS-fans. It might be infinite if you put it into some formula (and if you do, the result won't be useful), but effectively, how far do you ever move your cursor 'infinitely far away' along the imaginary above-the-top of your screen? I tried yesterday, and I am still trying to move it back into the infinitely small visible rectangle of the screen plane.

    While we are at the topic, Raymond, maybe you can answer some question:

    When I click the start button with the cursor at the screen border, the cursor jumps a few pixels into the screen, at least on Windows 7 with classic theme. Why is this so?

    Btw, it does so on every edge of the taskbar, but nowhere else.

  39. John Hensley says:

    looks like ABCDSchuetze had no trouble slamming his cursor into the nitpicker's corner

  40. Anonymous Coward says:

    Evan, I think your use case is atypical, but I feel for you. Something should be done, although what exactly is impossible to determine without knowing why you have 20+ tabs open in a single window.

    Raymond, you specifically stated that you *guessed* that the close button was in the top right corner because of Fitts's Law. While it is theoretically possible that someone very late in the design process tweaked the button, while somehow ignoring the start menu, or the taskbar buttons, or the clock, I think it's at least as likely that your ‘memory’ is wrong. Whatever the case may be, it has been proven beyond all reasonable doubt that the people responsible for the design of Windows 95 didn't give good old Fitts a thought when deciding what things to put where, and that includes the location of the close button, whether it was tweaked at the final hour or not.

    Schütze, one centimetre of desk space seems like a sufficiently good approximation of infinity in the case at hand. If the mouse is moved suddenly, rather than slowly, the mouse cursor accelerates and hits the top screen edge quite quickly, where it will stay since the mouse cursor won't move beyond the screen edge. So given that menus tend to be rather wide so you don't need very accurate horizontal aim, and you can hence get away with moving the mouse quickly, the top screen edge is indeed a good location for the menu bar, as anyone who has ever used a Mac or IIgs will testify. (Of course Apple's operating systems have their own problems, I know, but we were talking about the menu bar.)

    [You're assuming that the people responsible for the Close button are the same people responsible for the Start button. -Raymond]
  41. Evan says:

    @ABCDSchuetze: ""[…] because the top of the screen makes the height of the caption bar effectively infinite. " This argument is nonsense. I usually hear it from MacOS-fans."

    "Infinite" is of course a practical exaggeration, but it's definitely no exaggeration to say that it's effectively a *very large* target. If you don't believe me, time how long it takes to hit the close button of a maximized window vs one that isn't maximized.

    (Also don't accuse me of being a MacOS fan. :-) They do a number of things well, but so does every other system… and they do a number of things badly. I don't own any products from that company and never have.)

    @Anon: "If the mouse is moved suddenly, rather than slowly, the mouse cursor accelerates"

    Speaking of me being atypical (though I'm not convinced it's that weird in the browser case), I feel compelled to add "unless mouse acceleration is off." :-)

  42. Medinoc says:

    Fitts's law is the reason I always advise people with multiple monitors to put the rightmost monitor slightly lower than the left one (including in the monitor settings). This allows the user to still use the close button's infinite² area.

    Also, I can personnally confirm that a recent web browser (Chrome, Firefox 10) in mode "tabs on top" IS harder to unmaximize since you can no longer use the title bar for that.

  43. @Evan – no I'm not ignoring it at all.  Rather you are misapplying it.

    Yes, maximising the window increases the distance (makes the ID higher – i.e. worse).  The size stays the same.  The size does not become infinite because on my system the task bar is up the right hand side of the screen.  Since this is always a potential configuration the only "true" size of the gadget is it's actual size, if/when applying Fitt's Law, because if you are going to use an arithmetic principle to determine your design decisions you simply cannot ignore variables that are ALSO part of your design – i.e. allowing the user to configure their taskbar in a way that breaks your (supposed) Fitt's law calculations w.r.t the close gadget.

  44. ender says:

    Fitts's law is the reason I always advise people with multiple monitors to put the rightmost monitor slightly lower than the left one (including in the monitor settings). This allows the user to still use the close button's infinite² area.

    Depends on how low you put it – if you move your pointer fast enough along the top edge of monitor, Windows will helpfully move the pointer to the next monitor, even if that one has a lower origin (how fast you need to move it depends on how much lower the next monitor is).

  45. Medinoc says:

    @Jolyon Smith: I think this is more of an "edge case" because while you're operating within allowed parameters, you're outside *recommended* ones.

    To you, I recommend using the old-fashioned way to close a window: Double-clicking the system menu, which is still of infinite width and height (good luck if you're using Firefox, which no longer has a system menu in the upper-left corner.

  46. @Medinoc:  "To you, I recommend using the old-fashioned way to close a window: Double-clicking the system menu, which is still of infinite width and height (good luck if you're using Firefox, which no longer has a system menu in the upper-left corner."

    Which no longer works in Windows 8 for any application, as I quickly discovered.  All the corners seem to be tied to doing something with Metro now.  (Clicking the "X" in upper-right pixel still works though).

  47. Evan says:

    @Joylon: "Yes, maximising the window increases the distance (makes the ID higher – i.e. worse).  The size stays the same."

    No, it doesn't.

    Remember, you're not trying to hit a part of the screen. You're trying to hit *a portion of your mousepad* (/desk) that corresponds to a screen region. If the upper right corner of the screen is a part of the hit box of a control, the portion of your mousepad that corresponds to it is very large.

    > "…you simply cannot ignore variables that are ALSO part of your design – i.e. allowing the user to configure their taskbar in a way that breaks your (supposed) Fitt's law calculations w.r.t the close gadget"

    No, you don't ignore them. What *can* do is say 99% of users (actually this is probably a significant underestimate) will not put their taskbar at the top or right of the screen, and optimize for them.

    For those 99% of users, Fitt's law says to put stuff in the corners. For you, the corner is only a bit worse than the center of the title bar from a Fitt's law perspective, and also has a number of other advantages (e.g. aesthetic, not limiting or interrupting the title bar text, etc.).

    Or if you really say that Fitt's law doesn't apply to corners, then call it something else. "Fitt's law enhanced with additional real-world attributes" or something.

Comments are closed.