Hacker News new | past | comments | ask | show | jobs | submit login
X to close (medium.com)
704 points by viclou on Aug 13, 2014 | hide | past | web | favorite | 195 comments

As an Atari ST user from 1985 to roughly 1993, I wasn't expecting the author would actually mention GEM/TOS. I was pleasantly surprised when I scrolled down and, lo, there it is.

That said, since the "X" in this case is white on a black background, I always interpreted the icon as four arrows pointing inward to indicate a shrinking/disappearing motion. In fact, when you closed a window, GEM would play an (inelegant) animation akin to the Macintosh of the time, composed of a sequence of boxes first shrinking from the size of the window to a small box and then shuffling that off to the top left of the screen.

As bemmu points out, the maximize button (at the top right in a GEM/TOS window) is four arrows pointing outward. Incidentally, GEM did not have a notion of "minimize."

Put another way, although I find the Japanese inspiration argument interesting, I don't think there's a whole lot to it. I think it's a fun coincidence.

In any event, thank you for the trip down memory lane and for the fun screen grabs!

I don't think there is much relevance in the Japanese argument. One funny detail is that Sony actually inverted in their games the meaning of Round and X for western markets -> making X act as "validate" and Round as "Back/Cancel", the exact opposite of what they do in Japan.

As for "X being a true icon", I don't know. For me, it could stand as well as an abbreviation for "eXit" -> X.

The AmigaOS Workbench used (and still uses) a dot instead of a X. It's just a matter of conventions.

Word Perfect 5.1 for DOS, released in 1989, was very popular indeed among Microsoft's customer base. And the first to have a (DOS) menu bar.

Open the File menu and 'x' is highlighted in red in the word 'Exit'.

In other words, prior to Windows 95's release, DOS programs had already set firmly in Microsoft's customers' minds that 'x' was for exit.

RISC OS (1987), used on Acorn computers in schools in the UK also used an 'X' close button on the windows. And look at what word is prominent in the bottom right of the desktop... exit


It seems that as soon as the word "exit" is prominent in your UI, 'x' is going to become associated with exit.

An enormous amount of Microsoft's target market were already using Word Perfect every day and seeing red 'x's for exit. So when they decided to put a button on the window bar, what else were they likely to pick?

Was this true in all of their games? I know early PlayStation games did use O for confirm and X for cancel - even a few years into its lifetime, this was the case as Final Fantasy VII is an obvious example.

It depends on the game and possibly the console region: http://en.wikipedia.org/wiki/PlayStation_(console)#Regional_...

Are you talking about the Japanese version of Final Fantasy VII or the western version ? I have the Japanese version at home, I can check quickly if needed.

Both versions use O to accept and X to cancel. In the european FFVIII hovewer it was switched around. I think only early and rushed ports of japanese titles used O to accept on the playstation.

It's still true today, for instance dark souls on the ps3 uses O to accept on the japanese version and X on the western.

I'm still not sure why Sony did that by the way. While I'm willing to believe that X strongly means "bad/false" in japanese, I don't feel like it really means "accept" in western cultures as far as I know. When the playstation came out I don't think I would have had a lot of trouble accepting O for accept and X for cancel.

I think in FFVIII it was Triangle to cancel. (O was the menu button.) They completely jumbled it all up for some reason.

Triangle was the menu button, O for cancel, X to accept/action, Square to play Triple Triad.

I believe I had to remap 'menu' back to Triangle when I played through it on my Vita last month. Could be misremembering though.

It depends on the version of the console. Japanese Playstation Portable and Vita have the O button as "accept" - but I have noticed that not all games respect that. European games played on a Japanese Vita will use X for accept, but on the Playstation Portable most games will use the system settings, rather than their own(there were exceptions, however).

> I don't think there is much relevance in the Japanese argument.

I wouldn't discount it; one instance where I'm pretty sure it was taken into account is the design of checkboxes (HTML and otherwise). At least in Germany a checkbox on paper would be marked with an x to represent true - but that would be utterly confusing to a Japanese person.

I always saw it as an X, and when I used to tutor people on the ST (it was very popular where I grew up) I would refer to it as an X. It's more than possible that someone who used the ST, who saw it at as X, then suggested using an X as the close icon on Windows 95.

So the Japanese 'connection' may not be valid, but the notion that someone saw the GEMTOS symbol as an X and then influenced Windows 95 is not that far-fetched...

I was also really surprised (and glad) that the ST even got a look-in though =)

Another old example.

WordStar: Used "X" to Exit to system in its main menu (https://www.flickr.com/photos/markgregory/6946218793/?rb=1) - I do not know the revision shown in the screen shot.

According to Wikipedia (http://en.wikipedia.org/wiki/WordStar) WordStar was released in 1978. Which moves the date back to at least 1978 to use X for exit.

However, there is possibly a very simple explanation that the blog posting overlooked. In text menu's, such as WordStar's, which were quite common for a lot of software from that era, using the word "Exit" to mean "leave this program/application" was also common. When one goes looking for a single character memonic for "Exit" to build in as a keystroke to activate the "Exit" command from the menu, one has four choices: [e] [x] [i] [t]

Since [x] is an uncommon letter, while e, i, t, are more common, and therefore more likely to be used for triggering other commands in the menu(s), choosing [x] to mean exit meant that the same character could likely be used as a universal "leave this menu" command key across all the menus.

Which would then lead to the common _F_ile->E_x_it command accelerators in drop down style menus (whether in a GUI or in a text menuing system). [x] was unlikely to have been used for the keyboard accelerator for other entries in the "file" menu, so picking e[x]it was a safe choice.

It is not a far reach from _F_ile->E_x_it using [x] as its accelerator key to labeling the title bar button that performs the same function with an X as well, to take advantage of whatever familiarity users might have with the drop down menu accelerators

In WordStar it's a little more nuanced than that.

First, it's properly ^K X, as the ^K prefix subcommands block/file actions, as written by Rob Barnaby into all the WordStar versions starting with CP/M.

Second, ^KX as 'exit' means to save the latest revisions out to file before quitting, while ^KQ, 'quit', means to abandon the revisions. You might get a confirmation dialog and a chance to change your mind before you're dumped back to the commandline.

Current-convention iconic close-window behavior more closely emulates the latter.

Correct on the commands while editing a document, but I was specifically referring to the WordStar start screen before one begins editing a document (check the flickr link) where it has "X EXIT to system" (the all caps is also in the screen shot).

On WordStar 7.0a for dos, the main start screen menu selection is "X exit WordStar".

I think Turbo Pascal worked that way too

Not to mention the letter X sounds like the word "exit," or "ex it." In this way using X as a keyboard command to represent exit might be distinct from the way OP posits X to represent batsu (false, bad, wrong or attack).

though X already means false or bad in western culture, the opposite of a tick (✓). there's also "ex-partner" (my ex) etc. in a gui, it's an ex-window.

In Ultima IV (1985) the spell for exiting a dungeon was the X-it spell (activated with the letter X).

this helps prove your point. This is from Qbasic that ran in DOS. I use to run this on my windows 3.1 all the time:


Interesting that that screenshot shows other windowing decorations, but that "close" is hidden in a menu option.

ignore the outer part, thats from a newer version of windows. i remember using the short cuts for x to exit so i searched for a screen shot that had the file menu open

The inner part shows scrollbars. MS Edit has alerts and buttons and scrollbars.

Alt-F-S Alt-F-X

Used many times.


NextStep 0.8, '88 vintage.

I came to post the same thing... I had NeXT cube serial number 32 on my desk at Los Alamos as an intern back in '88 running NextStep 0.8 and clearly remember how amazing the machine was. We got the first 50 of the line (I think) because a guy in the Laboratory was a friend of Jobs, and Jobs wanted to promote the box as the perfect research tool thanks to Mathematica and the like.

Those were fun times!

NextStep, yes. The first time I saw Win95 (Chicago beta), I decided it was a rip-off of the NextStep UI.

Good find. That's clearly an [X] in the upper-right corner.

Yeah, but still more recent than Atari's Gem.

The Atari Gem thing is not an X, it's actually supposed to be a gem. It's just confusing because of the low-res image. Take a look at


That's not GEMTOS, that's GEM for DOS... big difference...

I was looking at the wrong end of the window, to boot.


Looks pretty X-like. It seems GEM itself was licensed from Digital Research but their GEM didn't use X.

Edit: Fixed DEC to DR, as pointed out by reply comment below.

I doubt DEC had anything to do with GEM. It was a product of Digital Research (same company that gave us CP/M, MP/M and DR-DOS)

They didn't, I just brainfarted. You know, http://vt100.net/dec/alpha_era_logo_small.png and all that.

Even so, Atari GEM could have influenced NeXT, which then influenced Windows 95... it's a cool bit of retro-chic =)

It's possible, sadly there is no folklore.org for Nextstep. In the end, both the X and things like context-menus were made ubiquitous by Windows '95.

Windows 95 looks quite similar to NeXTSTEP, with very similar gray window decorations and beveled (might be the wrong word) borders and buttons. It could be a design of times kind of thing, but to me it looks very "inspired".

For what it's worth, I remember thinking that Windows95 was 'inspired' by NeXTStep when it was new. The NeXTStep machines themselves were pretty amazing for their late-80's origin. They had (and effectively used) 4-level 1-megapixel displays at a time when PC's were largely stuck at 640x480 or worse.

High screen resolutions were certainly possible on DOS PCs of the era, but since PCs were an open platform, developers would often target their software at the lowest common denominator of hardware, so outside of specific niche markets, they didn't typically produce DOS applications intended for high-resolution displays.

On top of that, the DOS market in that era always preferred to make the tradeoff in favor of higher color depths rather than higher resolutions. At 640x480, a VGA display contemporary to NeXTStep could display 16 colors, and at 320x200, 256 colors.

> On top of that, the DOS market in that era always preferred to make the tradeoff in favor of higher color depths rather than higher resolutions.

I think I'd characterize it a little differently: I remember games favoring 256-color modes (320x200, ModeX, etc.) and productivity applications favoring higher resolution. At the time, having both high bit depths and high resolution at the same time was asking too much of the hardware.

To put it in perspective, for a while, my desktop PC was an older generation 486/33. When I say 'older generation', what I mainly mean is that it was one of the last machines that lacked a local bus of any kind. This put the video board behind an ISA-bus, which was lucky to sustain 8MB/sec shared across all peripherals. Just looking at the bus, it couldn't update a full 1024x768 desktop at more than 10Hz. (320x200 was well over 100Hz).

I actually had a 486 DX50 in the early '90s, probably around the same time as you had your 486. It had an EISA bus - a 32-bit extended version of ISA - which was roughly contemporary with NeXTStep, and that supported about 20 MB/sec of usable bandwidth. I had a Diamond Speedstar video card which allowed Windows 3.1 to run very smoothly at 1024x768, 256 colors. It could, IIRC, work very well at 1280x1024 in 16 color mode, which would definitely have been comparable to anything NeXTStep had to offer, at least in the graphics department.

But when it came to productivity applications, they were probably more conservative than anything else on the PC platform. Even though I had this great video card and could run Windows 3.1 at high resolutions, I still exited to DOS to run WordPerfect in 80x25 text mode.

I actually recently came across a box of floppies that I used in middle school. Schoolwork that I did in 7th grade, in 1992, was all saved in WordPerfect 5.1 format. Stuff from 8th grade, 1993, was in Ami Pro format. So I'm fairly certain that I didn't start using any productivity software in Windows until about 1993.

The only graphical productivity application I remember using prior to that, under DOS, was Ventura Publisher, which, IIRC, actually used a custom version of the GEM GUI. But I ran that on my 8088 XT-clone in EGA mode.

There was a store in the basement of my college dorm that sold Intel-based NeXTStep machines. As befitting the OS and the time, they were all very high end machines: 486DX2/66, EISA/VLB, SCSI Disk, 32+MB, and all priced well over $5K. I remember drooling over the machines, but never had the cash to actually buy one. In 1995, I ultimately skipped from the ISA 486DX/33 to a PCI-based P5/100. The P5 was in a completely different league, of course.

Regarding software, my family and I went straight over to Windows productivity software with the release of Windows 3.0. The back story behind that was that our first printer (ca. 1986-7) was a Toshiba P321. This was a relatively new, 24-pin printer that was almost completely unsupported by the DOS software that we had. Windows 3.0 solved the support problem by giving us a single, good printer driver that worked for anything that could print through the Windows API. Between that, the protected mode memory manager, and support for 16 color 800x600, Windows was a compelling enough package that we immediately switched over almost entirely in 1990.

Those screenshots are strikingly modern compared to the windows screenshots of the era: http://en.wikipedia.org/wiki/Windows_2.1x#mediaviewer/File:W...

The NextStep was way ahead on the curve for it's time (which is one of the reasons it failed).

I read an awesome piece a while back (and for the life of me I can't remember where) that said if you want to design for the future, estimate what computing power will be available in 10-15 years then buy that for your developers.

In that project I think they spent 186,000 per machine per user on average but they where designing software light years ahead of its time.

I don't think it worked out for them? The systems were way to expensive for most users.

The NextStep didn't work out for a bunch of reasons cost been one of the major ones.

The other project wasn't trying to create a product they where trying to capture a sense of what hardware would be on every desk in 10 years and design for that, in effect skating to where the puck will be rather than where it is.

I wish I could remember where I read it :|.

Interesting, but the connection to symbols from Japan seems a bit dubious (or at least not very recent). The term "cross out", and hence the use of an "x" to indicate negating something, seems to have been in common use in English since at least the 1920s: https://books.google.com/ngrams/graph?content=cross+out

I was going to say the same. Surely people have been crossing things out in writing and art for a very long time.

Edit: According to etymonline.com, crossing things out dates to at lease mid 15th century: http://www.etymonline.com/index.php?search=Cross

Also, it could have come from 'exit'. I've seen some text based programs use an 'x' key-press to activate a program 'exit', although I'm not sure of the chronology of its use.


Everyone remembers being told at school from a very early age to "cross out" things you dont want any more (such as writing the wrong word/spelling, getting the sum wrong etc). At least in western cultures anyway, this seems fairly universal.

Seems to me to be a very easy semantic jump to go from "disregard this mistake" on paper to "disregard this thing I am looking at" on computer screens.

A few counterpoints.

"X marks the spot"

Checking a box to indictate your selection on a form or ballot.

I can't think of any more off the top of my head.

Checkmarks (positive, accepted item) are quite often opposed to X marks (negative, undesired item).

Where you put a tick, you can put a cross?

Yeah, I could not imagine a Japanese symbol like 近く which according to Google Translate means "close" taking over.

It can also be thought of as a pun -- when you want to "exit" an application, you "X it".

I'd describe that as "verbing" it rather than "punning" it. The verb makes sense too, when you take keyboard short cuts into account. A very common short cut to exit a program (in Windows at least) is File->Exit, which translates to `Alt+F, X` so you do, in fact, "X it".

Wow, I've been saying "X it" for years but never even caught that it sounds like "exit".

No, you don't use a space: "xit name" (http://ex-vi.sourceforge.net/ex.html)

If ex is not your favorite editor, use EX<ESC><ESC> (at least, that is what I think Wikipedia claims. I'm sure any decent TECO user never exits their editor, browsing the web in a browser macro)

[I also checked ed and edlin. They had two commands for exiting: w writes and exits, q quits]

Wow, I had always thought the "X" was like an elevator close button. Sort of like a greater-than sign and a less-than sign put together: ><.

That's interesting... is the latin alphabet your native script?

Yes, English is my first language. I don't know why I internalized that symbol the way I did. And looking at the Atari TOS screenshot, I would have said the symbol in the upper right corner looked like an elevator "open" symbol, something like <>. But thinking more deeply, >< and <> are in only one dimension. They are good for expressing the idea of "close" and "open" respectively, but they fail to acknowledge the two-dimensional character of a window. So, I like what some of the other posters have said about four arrowheads pointing in and out respectively. That said, the symbol of two diagonal, outward-pointing arrows in the upper right corner of Mac OS X windows now strikes me as brilliant; in a minimal way, the idea of maximizing in two dimensions is expressed.

That's a really interesting idea, that had never occured to me.

Probably because I'm pretty aware of the close-doors button actually looking like >|< on typical elevators. The vertical line, I guess, represents the opening.

See for instance https://www.google.com/search?q=elevator+close+button&tbm=is....

Actually there are buttons with a line through them, and ones without. They maddeningly mean different things. See http://petesguide.com/symbols/elevator-button-symbols/

The Acorn Arthur operating system, a precursor to Risc OS used a sort of fat X icon to close windows in 1987


I never used Acorn Arthur but I did use RiscOS 3 whilst at primary school circa '91, and this had an X in the corner: http://www.mjpye.org.uk/images/screens/riscos-02.gif

I'm sure the GUI design of RiscOS 3 helped to inspire Windows 95.

I don't remember where the [X] came from specifically but there were a bunch of Brits on the Windows team at the time (myself included) and we had a number of Archimedes machines around. A lot of Lemmings was played.

I remember there was nearly endless debate about where the [X] should go so people wouldn't accidentally. I think the desire to make it visually distinct was a big factor.

Interesting. That surprises me - I'd always assumed that the UK scene didn't have much influence on the US technology world.

At the time Microsoft's Languages Division was run by a Welsh guy (David Jones) and he liked to hire fellow Brits.

*so people wouldn't accidentally click it

RiscOS had the x as well in the late 1980s

edit - Here's Arthur, the precursor to RiscOS in ~ 1986 - http://www.rougol.jellybaby.net/meetings/2012/PaulFellows/10... - It has nice x icons.

Here's a cleaner shot: http://upload.wikimedia.org/wikipedia/en/f/f7/Risc_OS_311_De...

Until the NewLook sprite set, it looks more like a weird flower shape than an X.

Here it is with NewLook: http://www.guidebookgallery.org/pics/gui/desktop/full/riscos...

Clearly now an X.

That's RISC OS 3, which is several years later. Arthur already had the X in 1987, see for example http://mobile.osnews.com/story.php/18941/mobile-opt-out.php . (Later than the 1985 find in Atari OS of the original article, but still interesting.)

As far as I know, Acorn's RISC OS also pioneered the icon bar and the context menu.

I see that, but it's the same symbol as pre-NewLook RO3, and is only charitably described as an "X". It's more of a splat.

In this early demo (Codename: Chicago), the minimize and maximize buttons have been redesigned, but the close button remains the same, and to the left as before.

I wonder where the author got the idea that the [-] button at the top-left was a close icon. It was the "Control Box", a menu icon. AFAIK it's still there, just invisible -- hit alt+space to open it.

Disclaimer: I'm currently unable to test that.

If you double-click it, it closes the application. It was converted from a picture of a spacebar to the application's icon, but still functions the same way.

Some time ago I was stunned discovering how many Windows users had no idea that double-clicking [-] closes the window. I bet that was the main reason for introducing separate close button.

As someone who used Windows 3.1 for years and now uses Powershell on Windows 7 every day at work, I'm ashamed to admit I had no idea.

> It was the "Control Box", a menu icon. AFAIK it's still there, just invisible -- hit alt+space to open it.

Still there in Windows 7. It's visible in some applications (e.g., PowerShell) but not others; ALT+SPACE seems to open it consistently whether it's visible or not.

My hazy memory is that the control box was retained for accessibility and for the many users trained to use it on Windows 3.1.

Control box was replaced by window icon itself and even on Windows 7 seems to work the same way, double clicking on window's icon closes the window (I had to test that).

I worked at Atari, on the Atari ST (writing a bunch of systems-level code). My cow-orkers were working closely with DRI to port GEM to the ST hardware. GEM wasn't done yet, and much of the engineering effort there was helping DRI finish it up. A lot of stuff was done on the Atari side of the fence that never made it back to the DRI sources.

I can categorically state that there wasn't any Japanese influence on that "X".

If anything, it was programmer art. We Atari folks were mostly video-game programmers, with some sense of design, and a lot of the stuff that was coming out of DRI was pretty ugly. So it probably got tweaked late a night until it "looked pretty" and wasn't revisited (the ST was started and shipped in about 10 months, so we were in kind of a hurry).

As the article shows, the close button on MacOS classic was basically an empty box, but on mousing down on that box, it transformed into something that looks a bit like an x. I'm basing this on what I can see from using [1], but from my possibly inaccurate recollection of using the real thing in the 80s and 90s, some versions of MacOS had an even more "x like" mouse down image on the close button.

[1] http://jamesfriend.com.au/pce-js/

Yeah, I was remembering this too. Of course, it also kind of felt like you'd "selected" the box. In fact, I think early checkboxes had x in them on Mac, didn't they? So it might have been a bit of a coincidence ... or an inspiration.

Yeah, weird. I remembered it being just an X but it had lines in all directions.

The delete/rubout key on many old terminals had an X on it. Like this: http://www.cosam.org/images/vt220/keyboard.jpg

The backspace key on my Lenovo keyboard is identical. And if I remember correctly, it was the same key on an electrical typewriter I used in the 90s.

On typewriters you would backspace and then overtype an X. Those keyboards tend to just use the word "backspace" or a back arrow symbol.

So typewriters use this symbol when they automate the overstrike or they moved to correction tape.

So that's where they took that Backspace icon on smartphone keyboards from...

I always found it interesting that Sony swapped the X and O buttons for the western Playstation market. In Japan X (batsu) does mean "back" or "no", whereas elsewhere it is reversed.

Reminds me of how I use [x] to tick boxes in paper forms in Europe, but I was told in Asia (Taiwan) that I should use a check-mark instead. x is no.

I think check-for-yes, x-for-no is a fairly common convention in English as well, just not in tick-box forms. You do see it in feature-comparisons grids a lot, often with the check-mark colored green and the x colored red, to mix in another convention. Example: http://prezi.com/pricing/

I always though it was for the color. Blue for OK, red for No

They mention X and O on the PS controller but usually in games O is for no and X is for yes. Completely opposite of the batsu/maru, incorrent/correct they were discussing.

It seems like Sony, being a Japanese company, originally intended for O to represent yes, and X to represent no. If you look at a lot of early PlayStation games, or most modern games released in Japan, that convention is apparent. The O and X buttons on the PlayStation controller even match the placement of A and B buttons on Nintendo's controllers, providing a clear analogue between the two.

I'd be curious to know what caused that convention to change in the west.

> I'd be curious to know what caused that convention to change in the west.

The XBox controller has a swapped A/B pair and uses A confirm B cancel.

The XBox controller is a clone of the Dreamcast controller. If you go back even further to the Megadrive, it had 6 buttons (on advanced controllers) with two rows of buttons - xyz and ABC arrayed from left to right.

When the Playstation 1 was released, the two consoles consumers could have been familiar with in the west were Super Nintendo and Sega Megadrive. SNES had the familiar ABXY, but mirrored with AX on the right. It used A for yes and B for no. The Megadrive had the aforementioned ABCxyz and also used A for yes and B for no. Meaning, one console used the bottom button for yes and the one on its right for no (Megadrive), while the other had the bottom button for no and the one on the right for yes (SNES).

So I doubt SONY copied the competition for their decision to swap the yes/no buttons.

>SNES had the familiar ABXY, but mirrored with AX on the right.

Those of us that grew up with Nintendo consoles would say that B on the left of A is the natural order of things ;)

>Meaning, one console used the bottom button for yes and the one on its right for no (Megadrive)

Actually, almost every Megadrive game I've played lets you use both A and C for accept in menus, so you could use whichever orientation you were more comfortable with.

I am just quoting another instance from around that time, besides this is all kind of moot since:

> I doubt SONY copied the competition

IIRC this wasn't decided by SONY, each game used its own variation until a standard was created organically.

Makes sense to me. We have scan sheets and forms with empty bubbles and boxes meaning, "not this one," and we put a check mark, X, or other mark in them to indicate our selection.

In Japanese variants, he is correct. When Sony westernized the playstation controller, the O and X functionality was flipped. Sony is yet to comment on the reasons why.

It might be because of Sega's consoles, which were more popular in the US than they were in Japan, where Nintendo ruled.

The Genesis' button layout was A,B,C arranged in a diagonal from bottom-left to top-right. A was usually 'accept'. The Dreamcast had a diamond with A at the bottom and B to the right.

Nintendo's Famicom buttons read A,B from right-to-left, and that trend continued with the Super Famicom's diamond, which had A to the right and B at te bottom. The N64 had a weird layout, but again B was to the left of A.

Sony probably focus tested the pad in the US and found that players were more used to Sega's layout.

As I said in another comment, almost every Megadrive game I've played lets you use both A and C for accept in menus, so you could use whichever orientation you were more comfortable with. I think anyone who started with Nintendo consoles would instinctually rest their thumb between B and C.

My guess is that Sony thought that X and O wouldn't have as obvious connotations outside of Japan, and figured that people would assume the button closest to the player (X) would be the OK button. In practice, I have found that people with very little exposure to Japanese culture still have the same association with X and O in their heads and get confused when using Playstations ("you press X to accept???"), so I'll curse Sony forever for this stupid regional change.

This is almost certainly a result of intense focus testing, so I don't think it would be a design choice by Sony based on cultural differences so much as an observation of user comfort and expectations.

That's not to say that those expectations weren't due to cultural differences.

I always thought the reason for the switch was down to driving games, where X is a more natural fit for 'accelerate'. Also, X to me is a more 'definitive action' symbol compared to O, make of that what you will.

SNES was also more popular than the Genesis in the west.

AFAIK it came close, but the SNES launched later in the US and the Genesis sold more units. As opposed to Japan where the Genesis didn't really get a foothold.

I thought the popular rumor was this was to help people with Nintendo 64 muscle memory, where the X corresponded (loosely) with where the A button was positioned, and the O corresponded to where the B was positioned.

You're wrong on both counts. B is still to the left of A on N64 controllers, and the PlayStation launched a year before the N64 in America.

My SNES muscle memory prefers the Japanese version. :)

IMO one of the most short-sighted decision made Sony Entertainment EU/US. They had a standard and decided to change it just for the sake of changing it.

In Japan they're sometimes switched. That is to say, I've worked on PS3 games where the US SKU had X=accept, O=back but the Japanese SKU had X=back, O=accept.


In Japan the actual convention is X=cancel/back and O=select/validate. Most japanese games follow that, because the Batsu/Maru meaning is obvious in Japanese with these symbols.

Try playing some Japanese games on PS. It's the other way round in Japanese games

On the PS1, Final Fantasy VII was O=yes, X=no, and so were a bunch of the less mainstream JRPGS; also, Metal Gear Solid games were O=yes, X=no at least into the PS2 era.

I actually distinctly remember getting a copy of Metal Gear Solid 2 and being confused why I couldn't start the game. Hitting any button on the splash screen would take you to the menu, and then hitting X from there would take you back to the splash screen. Having primarily played western games up to that point, it didn't even occur to me that another button could be used for 'confirm'.

If you buy a PSP or PS Vita in Japan it will use O as accept and X for cancel in the system menu. Not sure about the home consoles.

not always. western games tend to be bound that way, but Japanese games, espescially RPGs , were not.

example : final fantasy 7 control bindings. http://www.cavesofnarshe.com/ff7/buttons.php

Interestingly the positioning of the two matches the common use of the nintendo buttons, A for confirm and B for cancel. The western Playstation games are really the exception to the rule.

No 'x' to close vi? Was that not always there? I've certainly been using it as long as I can remember; that's not to say it's always been there though - does anyone know when it was first available?

Edit: seems Wordstar used X too, probably starting in 1978.

Well, that's different. You are talking about an "x" that you type in, as opposed to an "x" you can click on.

To answer your question, "x" (short for eXit) was available in vi from the beginning:

ZZ Exits the editor. (Same as :xCR)

Source: Bill Joy's "An Introduction to Display Editing with Vi" http://www.verticalsysadmin.com/vi/vi_editor__bill_joy.pdf

Yes, it's different, but the article was trying argue that since "X" wasn't used in that context at the time it was introduced as a GUI element, the GUI element couldn't possibly be referencing the letter as a way of closing a program. While I agree that it's unlikely that the GUI "X" refers to a letter "X", that's not a valid argument for that position.

Edit: and thanks for the vi ref.

Understood, and you're welcome!

Nano also uses Ctrl+X to exit.

nano is from 1999, so hardly relevant.

But nano is just a pico clone, and pico started out as pine's text editor. pine as a project was initiated in 1989.

ex uses x to exit as well (1976).

Too bad, that popular Windows applications like Skype and Spotify have gone against this and made "X to minimize". And their making of Alt+F4 also to minimize drives me nuts.

X closes the window not the application. The only intuition you need is to realise that window !== application.

That's not standard. If it's a one-window app, such as Skype, it should close the app. BareTorrent is also sort of a daemon process (where you often want it to live in the systray), and it follows standards. Minimize will put it in the system tray and close actually closes it. It's standard and feels intuitive.

I don't think so - I find this very subjective - I've always assumed that X just closes the window, even if it's a single window app. This is consistent, just some applications happen to also run in the background.

>I've always assumed that X just closes the window

That was one of the old differences between Windows and OS X behavior (or app-centric vs window-centric).

If you're reffering to Windows, then X usually closed the app too.

> If you're reffering to Windows, then X usually closed the app too.

I think his point is the X closes the window, and many, but not all, Windows applications also choose to quit when their (last) window is closed.

In case of Spotify, that window is the application. And the only window that the application has. Hitting X in that window should quit the application IMO.

Far from exclusive to them though, and it seems to be (/have been?) a very common thing for IM clients to do. I can live with it as long as the devs put in the appropriate options for the behavior you want, but I'm not happy about it.

I sort of understand the point for IM client, that people won't close it unintentionally. But that is solved by showing a dialog asking do you want to close. Breaking Alt+F4's functionality is such an abomination that it is impossible to understand the thought process behind the decision.

Huh? I thought the point for an IM client is that you don't need a window anywhere, if you have the icon down there. So you can close the main window, or you can separately quit the application.

Well, in Skype (and Windows applications in general) there is a MINIMIZE button for that, next to X and maximize. Why make X also act as minimize?

Minimize puts your window in the taskbar whereas X closes the window (not in the taskbar anymore), but whether it also closes the program or not at the same time is up to the program (and ideally should be configurable). That's how I understand it at least.

BTW: In Skype there is an option for "Keep Skype in the taskbar while I'm signed in." When it is unchecked, minimize and X behaves as you said.

Not to mention web sites that use a graphical icon in the background instead of a textual one or lack a fallback X on pop up windows. I'm forever groping around trying to shut these overlays in the dark.

A large number [citation needed] of Windows apps that have done this are things that most users would want to run in the background, primarily things like instant messaging clients, music players, and daemon-y programs (torrents come to mind).

There's frequently a setting to configure if it truly closes or not, but it's not usually there for more mass-market targeted apps (death by settings and configuration options to support being the likely cause).

That's because Microsoft's guidelines say not to leave icons in the notification area any more for running apps. For skype and friends, the behaviour has always been to keep running even after the last window is closed, but keep an icon in the notification area (a.k.a. systray).

It does mess with everybody's expectations of closing the last window freeing space on the taskbar though.

What about crossing out dates or tasks?


Or crossing-out an item to "delete" it on the page?

NeXT had X buttons to close windows before windows 95, with a very similar look to to win 95 window button styles too. I think NeXTStep 1.0 was in 1989 or thereabouts.

It's not X, it's ❌. (90° angle of the two lines) I always hate it when someone actually uses an X. Looks ugly.

I recall being mildly shocked when Windows 95 came out with the the [x] button. I don't know why, but I thought that it was somewhat dangerous to allow users to quickly exit an application like this.

Maybe it's because I was used to Windows 3.11, where you had to actually double-click the [-] button to exit an application.

Agreed. But more to the point, I'd like to know what idiot decided it would be great to put a close button right next to the maximise and minimise buttons. It's a disaster just waiting for a mis-click. Since Win95, everyone else has copied this particular feature.

First thing I do whenever I do a new Linux install is put the close button on its own on the left where it belongs.

On Linux I always eliminate window decoration. I don't even like to have borders, let alone buttons I never use.

I really wish I could do the same on Mac.

That's actually not such a bad idea. I have just a single pixel border around the left/right/bottom at the moment, just so I can see the edge of overlapping terminal windows. I have to say I wouldn't miss the titlebar much either.

If you notice, there's some space left between the maximise and close, so they're more like _[]...X

Plus, you can maximise/restore by double-clicking the titlebar which is a larger target and I expect most users learn that and almost never touch the actual button.

I still occasionally double click where the [-] used to be, which still works. Even longtime Windows users have given me weird looks at this. They are surprised it does something.

In the Atari TOS screenshots, other icons such as arrows are black on white background.

If the icons in upper left and right are also like that, then the upper left icon is actually four little triangles pointing inwards and not an X. The one on the right is four little triangles pointing outwards.

(Or it could be an X)

Atari ST users seemed to see it as an X -- so whether or not it was intentional on the part of the designer, it's not that hard to see how one of these users could have then translated that misconception to the Windows 95 GUI. It's all about perception...


So this little article has travelled pretty far! There were a lot of good tips, comments and insights into the origin of [x] but none as good as this email that I received from Windows 95 team member Daniel Oran.

“Hi Lauren,

A friend forwarded me your Medium piece, “X to Close.” He remembered that I had worked on Windows 95 at Microsoft — I created the Start Button and Taskbar — and thought I’d be amused. I was! :-)

It’s fun to see how history gets written when you actually lived those long-ago events. I joined Microsoft in 1992 as a program manager for the user interface of “Chicago,” which was the code name for what eventually became Windows 95. 

So, who was responsible for this last minute change? As far as I can tell, this person is responsible for the proliferation and widespread use of [x] in UI design today.

It wasn’t a last-minute change. During 1993, we considered many variations of the close-button design. And the source wasn’t Atari. It was NeXT, which had an X close button in the upper right, along with the grayscale faux-3D look that we borrowed for Windows 95.

I wanted to put the Windows X close button in the upper left, but that conflicted with the existing Windows Alt-spacebar menu and also a new program icon, which we borrowed fromOS/2, on which Microsoft had originally partnered with IBM.

Attached is the earliest Chicago bitmap I could find that includes an X close button. It’s dated 9/22/1993. (In attaching the file to this email, I just realized that it’s so old that it has only an eight-character name. Before Windows 95, that was the limit.)

Thanks for your very entertaining essay!



I guess you could say case [x]ed.

Thanks again to everyone who helped track down earlier examples of GUIs and early text editors that used [x] to close as well. Fascinating!

Windows 95 was the first time I remember using it, and I have been using PC's since TRS model 80. It makes sense, X means "stop" in most cases and stop essentially means close or terminate a process / app.

> It makes sense, X means "stop" in most cases

I'm trying to think of any cases where this is true. Stop signs aren't crosses, they have a special shape.

Windows versions prior to Windows 95 lacked an "X" button, but double clicking on the left menu icon would close the window.

A behavior still present in modern versions.

The use of the X symbol to mean "cancel, close" isn't nearly so mysterious as the author claims. "Cross off" and "cross out" are common phrases in English, and traditionally denoted by an X symbol (the "cross").

There is no reason to suppose that the GUI usage was inspired in any way by exotic Japan. The X as "cancel symbol" has been quite common in the west and indeed worldwide for millenia.

If I recall, clicking on the X on old Macs added an X inside the square, so I think there's a step missing from this article.

It wasn't really an X, more "stretch marks" to visually indicate the mouse-down event.

Odd they didn't use such stretch marks anywhere else.... except perhaps check boxes.

"Vi, vim, emacs or edlin?

No [x] to close these 1980's text editors either. X was commonly used to delete characters in-line, but not to close the program."

Hmm... I've used :x to write+quit in Vim for years. And, :X is to encrypt+quit. Don't have a year when that was added though. Could be fun to try and dig that up.

>Hmm... I've used :x to write+quit in Vim for years. And, :X is to encrypt+quit. Don't have a year when that was added though. Could be fun to try and dig that up.

:x (short for :xit) was in the original ex written by Bill Joy in 1976. According to Joy[1], ex pulled together ideas from a few different places:

1. em from QMC, written by George Coulouris[2]

2. a modified version of ed from UCLA

3. An early version of ex written by Charles Haley based on the em source in 1976

4. Bill Joy himself

em uses 'x' for its interactive find-and-replace mode ("exchange"), so it didn't originate there. That leaves 2, 3, and 4 as possible origins. I can't find anything on the UCLA ed. If the origins are in 3 or 4, :x is from 1976.

[1] http://roguelife.org/~fujita/COOKIES/HISTORY/1BSD/exrefm.pdf

[2] http://www.eecs.qmul.ac.uk/~gc/history/

Not sure why you got downvoted so much.

It would be fascinating if there was a text mode [x] somewhere, especially if it was in common software.

EDIT: It's not in Norton Commander.

Wow great article. I don't agree with his conclusion that it came from Japan. But it's as good a reason as any I suppose.

One quick thing, IIR Windows 2.0 and 3.0, the '-' button in the upper left wasn't "close". It was a small menu that happened to have close as an option.

You could double-click it to close, though. And of course the menu (and the double-click-to-close functionality) is still there, it's just that the [-] icon was replaced with the application's own icon. So contrary to the article, the close button was added in Win95, all the other elements are still there.

If you want to see the UI of GEM for yourself, here's an in browser emulator of an Atari ST with GEM:


This is a screenshot from OS/2 Warp 4, which was released after Windows 95 (source: http://www.os2museum.com/wp/?page_id=132). OS/2 Warp 3 that came before Windows 95 had the following set of Window buttons:


Looks like `[x]/2 == [/]`.

IIRC on windows 3.1 keyboard navigation X was always the key for exiting, as E was already used for other things.

I clearly remember that for closing windows one could do alt+f4 (which was itself a shortcut to Close) or open the file menu (Alt+F) and select eXit.

I can't check but I believe it was the same for Write and notepad as well and any other programs that had the Exit option.

So maybe that's where the windows 95 developer took inspiration for the X icon


I had a non-technical friend who insisted it was an x because you used it to "x it"

Someone in our office insists on calling closing a window 'crossing it off'.

When we have completed a todo task we "cross it" to mark it done. i would say the x to close is intended to represent a "crossing out" not the letter x. It is pressed to signify a task has been completed.

My thoughts exactly. Not being from exactly a Western culture (I'm Russian) I always saw the X not as a letter but rather as a cross (Russians say "click the cross to close the window"), a sign of deleting or cancelling something by crossing it out. It made perfect sense so I never even thought there could be other explanations. All because in our language there is no X-exit connection.

Lots of banner ads make the close symbol e.g. the second from right (swap maximize and close) or swap the functions... thus exploiting muscle memory of people to open the ad :/

I think I remember that hitting the "close" button on early, black-and-white macs would make a star appear in the square, signifying the press. Almost like the X...

Perhaps it's not an icon, and was meant to indicate eXit. I know must use 'q' for quit, but I've seen a few programs that use 'x'

I'm not icon designer, but I just finished the hackdesign.org course (I recommend it) and now I understand a little bit of it and now I always try to think as one.

The [X] icon in graphic windows software (not in WordStar, Vim, etc), and not thinking as a letter of the alphabet (remember that maximize and minimize don't are also) but just as picture, it remembers me something collapsing. Like something bigger in a normal state with the borders collapsing to a center till disappear. As when you turn off and old CRT television (or an Android powered cell phone).

I'm still thinking on this... in design you don't think in one part alone, all the context is important. The minimize represents the future state of the window in the bottom bar, and the maximize represents the window occupying all the available area.

I conclude these three icons are really good and well designed.

This is a great story, and I enjoyed the look back at all the different OS. sadly if it happened today the x to close would have been patented.

I'd always thought of it like the "crossing out" kind of gesture such as drawing an X over something on paper.

What about using CTRL-X to exit DOS programs?

'X' always seemed fitting for another, more poetic, reason: The kiss of death (X also represents a kiss). I wonder if it was in the designers mind.

Well what you know about `windows` in GUI? What is the first appearence of the `windows` based-GUI?

I misinterpreted the title as "X Windows consortium to close."

I'm glad I'm not the only person to think that (mind you this was over breakfast and before much coffee).

I did think very briefly that it was something to do with X, then thought X was a variable as in "$X to close".

me too

Mouse-wheel to scroll (to intro).

For me, the pinnacle of Windows UI design has always been Windows 95.

The best windows UI imho was Win2000. Best UI overall would still be OSX, also imho.

Mind telling us why this is so important !

Because it is a design artifact that has survived decades, and here on HN there are a significant number of people who are designing artifacts that they hope will survive decades. Context is everything - for those people, this article provides a little context. Yesterday's "X" to close is today's 'swipe to close', and tomorrow's [___?___]. Solve for [___?___].

Does it have to be important to be interesting?

Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact