I find this story odd because IBM was consistent with their keyboard nomenclature across multiple products, and the 3270 series mainframe terminals used the Tab key, located in the same place where you would find a tab key on a modern keyboard, to move the cursor to the next field.
As an aside, it's worth noting that moving between fields was important enough on IBM terminals that they had a dedicated "back tab" key located on the opposite end of the keyboard to the tab key. On the original IBM PC, they decided to combine both functions into a single key. As a result, the tab key on the classic PC keyboard features the symbols for both forwards tab and back tab on the same key, the back tab symbol being on top to indicate that you need to hold down shift to use that function.
EDIT: The 5250 series terminals used the terms "Field Advance" and "Field Backspace" instead of Tab and Back Tab, but otherwise they used the same symbol on the keys, and the keys were located in roughly the same position as the 3270 series. Reference: https://www.bitsavers.org/pdf/ibm/5291/GA21-9409-0_5291_Disp...
Having worked at IBM, I would guess that using the tab key in this way was part of a patent they were pursuing and Microsoft's use would show this to be 'obvious' and thus not patentable. But that is just a guess.
In the 80's IBM had a whole class of high level technical people called "Systems Engineers" whose entire job description was to opine on the merits of any given system. Not write systems, not debug them, and certainly not to explain them, it was simply to opine "you're doing it wrong."
Microsoft is suffering from the lack of such a group today; they're definitely doing it wrong, where "it" is pretty much everything... except pissing off users.
> I would guess that using the tab key in this way was part of a patent they were pursuing and Microsoft's use would show this to be 'obvious' and thus not patentable.
Something that's bothered me about user-facing patents:
Let's assume that the idea of using a keyboard key to move between input fields in a software form is not obvious, and in fact is a brilliant stroke of genius the likes of which the world is not likely to see again. If that one guy hadn't been born, we would have gone thousands of years with no method, keyboard-based, mouse-based, or otherwise, of moving from one input field to another input field. Every piece of software would use nonconfigurable timers, and you'd just have to hope you could type fast enough.
I don't see what the hypothetical benefit of extending patent protection to this brilliant idea is supposed to be.
Say you're the company who comes up with the idea. You can benefit by including it in your product, where all your users can see it. In other words, the benefit you get from coming up with this idea is that you can publish it for the world to see, and that's the only way you can benefit from it. A usability feature that your users cannot use or know about doesn't increase usability.
Even though the idea isn't obvious, the implementation is. If you disclose your brilliant idea, everyone will copy it and your advantage in the marketplace will be transitory.
So... what is the purpose of giving you a patent? That cripples the marketplace, but it fails to realize the benefit of patents, publication. Publication necessarily had to happen anyway.
> A usability feature that your users cannot use or know about doesn't increase usability.
Cannot is maybe doing a lot there. There's plenty of usability features that aren't really obvious or apparent unless you look very closely. Ex: pinball machines have timed shots, but there's almost always a grace period so if you contact the ball with your flipper around when the timer hits zero and it makes the shot, chances are you'll get credit for it even though the timer expired. That's a usability feature most users won't ever notice. At WhatsApp, I would never send an S40 user a verification code where the 4th digit was 8, because if you got a text message with 123-890, s40 would turn -8 into an 8th note emoji; until today, probably 3 people knew that ... but it dramatically improved usability.
> Even though the idea isn't obvious, the implementation is. If you disclose your brilliant idea, everyone will copy it and your advantage in the marketplace will be transitory.
> So... what is the purpose of giving you a patent? That cripples the marketplace, but it fails to realize the benefit of patents, publication. Publication necessarily had to happen anyway.
If I had gotten a patent on the 'avoid -8 in verification codes', then the technique would have been public for everyone to see. So publication for exclusivity / forced licensing is an exchange of value between society and the inventor. Of course, avoid -8 is pretty obvious, when someone testing the s40 client complains about getting an 8th note in their verification code message.
For an invention that must be disclosed to be used, society isn't really getting anything in return for exclusivity. Maybe promotion of progress, theoretically, I guess, in that whoever thinks of it first gets paid; leading more people to think about things?
Here's a real IBM 3270 keyboard.[1] Note the "Next field" key on the left, and the matching "Previous field" key on the right.
The IBM 3270 was a device for filling up forms. The mainframe sent the terminal a form with blanks, and the terminal let the user fill in the blanks. The terminal hardware prevented the user from overwriting the static parts of the form, and could apply some other form constraints, such as numeric fields. That was all done by the terminal. When the form was filled in, the user pressed ENTER, and the completed form was sent to the mainframe as one transaction. This approach let one mainframe service huge numbers of terminals. The user never experienced delays while typing and could type at full speed, often without looking.
PCs didn't have that usage model. The PC crowd was thinking "typewriter". One of the first terminals for home computers was called the "TV Typewriter".
Web forms do have that model, but with less consistency.
Nitpick: The terminology used by IBM on the 3270 family (including the 3277 whose keyboard you shared) was "Tab" and "Back tab", not "Next field" and "Previous field".
From what I remember, there were two "Enter/Return" keys on IBM 3270 terminals. One was the regular "Return" key we have today, which just advanced to the next field, and didn't submit the form. There was also another "Enter" key where the Right Ctrl key is today, and that submitted the form. So, I presume, instead of being against Tab key, IBM might be against "Return" to be the form submit key as people who use 3270 would expect it to advance to the next field.
And that was true for many DOS programs. Pressing Return would just go to the next field, not submit the form. That was one of the things that needed some getting used to on Windows.
Your memory is correct, and it's interesting to note that on the IBM terminal keyboards, the Enter key was marked "Enter", and the return/new line key was marked "↵". On the classic IBM PC keyboards such as the Model M, the Enter key is marked "↵ Enter". I believe IBM chose this to convey that the Enter key on the PC was both an "Enter" _and_ "Return" key in one. As you say though - individual applications got to chose what that meant in practice, leading to inconsistent behavior.
As someone who prefers tabs (I'm not looking to argue), I once asked Brendan Eich on Twitter why he prefers spaces. His answer was more thoughtful than I'd expected.
The tab key itself is hijacked by modern OS/UI behavior. It makes it complicated to actually type literal tab characters in certain contexts, particularly in the browser.
I still prefer tabs (and I'm a Go developer), but he is absolutely correct about that being a pain in the butt. For instance, try getting a tab character into the text area on Hacker News
It might be reasonable to have a separate key for "tab" vs "next field", as well as separate key for "line break" vs "send". (But, tab and line break are not applicable for all contexts.)
However, it might also be reasonable to have a key or key combination (some programs use ^V) to enter control characters as data rather than as commands.
It might also be a consideration when designing a new computer (which does not have to be the same as existing ones); I had thought about such things and may make such a consideration.
As far as I'm concerned, it's a feature when the viewer controls the number of spaces per tab, it's a bug when someone else does (as is often the case online via css `tab-size`).
Non english speaker trying to wrap my head around what Brendan said to you.
I was once told that the tab key can be represented in different ways on different systems, and that's why spaces are safer because they're always represented the same.
Literally try entering a "tab" character in a Hacker News reply. For me in Safari, it changes the focus to the "reply" button.
It's literally difficult to enter a "tab" character into many text entry fields and dialogs and applications, because it's used so often as a navigation key.
> I was once told that the tab key can be represented in different ways on different systems, and that's why spaces are safer because they're always represented the same.
The main counter argument is that users have different preferences for the amount of indentation, so giving them control over that, just as they (nowadays) have control over the font used and window width, is a good idea.
The tongue-in-cheek counter argument is that fixed-width spaces are preferable over ‘normal’ spaces. They also give you more control over indentation, allowing, for example, mixing usage of THREE-PER-EM SPACE (https://unicode-explorer.com/c/2004) for indentation with FIGURE SPACE (https://unicode-explorer.com/c/2007) for right-aligning numbers.
> The tab key itself is hijacked by modern OS/UI behavior. It makes it complicated to actually type literal tab characters in certain contexts, particularly in the browser.
Tab key is both a control character moving the cursor to the next input and also an input representing the tab character as you see in a text editor.
Now that im thinking about it I’m convinced capslock would have been superior next field key and alt+capslock to be used for toggling capslock. But it’s not obvious to me capslock is even seen by the OS. It could be changed on the keyboards themselves.
Brendan Eich was making a point about the tab key also being used to switch to different text fields and buttons. This makes it difficult to type in certain applications. A space doesn't have that issue.
About your point of tab being represented different on different systems: It will always be ascii 9, how it's draw does differ between text editors but I consider that one of it's strengths for programming. Everyone can configure what an indent looks like for them, it makes reading code easier.
No, the point was that pressing the Tab key often does something else than inserts the tab character. Moves focus to the next input field, for example.
No, it's that it's extremely challenging to insert a literal tab character into anything resembling a web form.
You can configure your editor so that pressing the tab key inserts a tab character. Good luck with a web browser, or a UI where tab moves focus to the next control. The space bar basically always inserts a space character.
No he's saying if you're trying to write code in an edit box in a web browser (for example) and you press tab, instead of inserting a tab character it will move the focus to the next input field.
Bullshit reasoning though, because even people who use spaces for indentation don't do it by hammering the space bar - they also press the tab key. And of course in modern browsers we can give tab the expected behaviour.
The only logic I've ever heard for using spaces for indentation that actually makes sense and I vaguely agree with, is that lots of programmers do not give a shit about formatting code and even using spaces properly is often a bit too much for them.
That doesn't matter so much if you have an autoformatter though, so with the exception of Go I don't know why modern languages with widely used autoformatters don't use tabs.
Control-shift-u, 9, space/enter, works on most Linux systems.
To explain: control-shift-u allows entering a Unicode character by its hexadecimal code. This presumably depends on the Input Method Editor (IME) in use, which is something I've never fully understood, but this seems to work widely across different desktop environments (Xfce, KDE) and display servers (Xorg, Wayland).
A great read, although I'd still like to know what IBM's reasoning for opposing this use of the Tab key was.
Is it because they didn't want Tab to be both an input and a control character? I.e. there are some cases where you can type a Tab into an input field, and there are other cases where you can't, and it's not immediately obvious which ones are which?
All the way in 2026, I would still be sympathetic to this view.
Personally I don't like the tab key for field change.
Firstly it was a breaking change from dos. Dos programs used Enter. And enter meant you could capture numeric data using 1 hand, since the numeric keypad has an enter key.
That means left hand can stay on the (paper) source. Right hand types. People got fast at this. (Really fast). And this pattern lives on in some programs kline Excel).
Lots of people (ie my customers) hated needing both hands on the keyboard. Lots of our programs allowed mapping of enter=tab.
I should be clear. It's not the "name" of the key that matters, it's the location.
The dual-use of the key is just an annoyance we live with. Sometimes the key behaves as a navigator, but in other cases it behaves as a spacer. Daft. (Enter would have the same problem. )
The best solution (by far) would gave been to add another key to the keyboard. Preferably in the numeric key pad. We got lots of new keys in that era. Hindsight says we should have added a "move on" key at that time.
If you're designing a text editor that's intended to run within a browser you need to fully break user expectations about the functionality of the tab key - there are two completely logical and intuitive roles it can serve in such an environment that are in conflict.
The same issue arises (with much higher frequency) with Enter and even in the modern world I'm sure we all have a pretty complex ruleset we've committed of when ctrl+crlf triggers a newline, or a message send and what the corresponding behaviors of a bare crlf and shift+crlf are.
In HN's editor shift+crlf and a bare crlf cause a newline creation and ctrl+crlf does nothing - but often times ctrl+crlf will trigger a form/message/whatever submission, shift+crlf often causes a raw newline insertion (even when in a form context) and then a bare crlf might do one - might do the other or might even do neither! Those are the common bindings but I have seen exceptions and inversions of bindings with shift+crlf causing form submission while requiring ctrl+crlf for raw newline insertion.
All this stuff is just super annoying and causes a lot of user friction (and for a long time MSFT's style guide was considered a seminal reference for best practices as ironic as they may seem to folks these days).
The way I interpreted it was that there was no complete "IBM" reason. I read it as "one person in the IBM bureaucracy stepped in and brought things to a halt, which highlights the cultural differences" considering it is a post about said differences.
I assume so. Obviously there are very different concerns if you're managing an organization with countless moving parts versus you want to build something for the user quickly.
That's exactly my memory too, using a tab was second nature and it irked me when I went to gui apps with a mouse and the tab order was wrong, mainly with visual basic apps.
It's been ages since I used an IBM green screen but I seem to recall that in our application Tab did function to move between fields, but I think there were also function keys reserved for this purpose.
I think we should connect game controllers to all machines so the arrow buttons move you between fields, the 'A' key takes you up a level (in hierarchical menus), the 'B' key takes you into a subordinate menu. So to move between fields, you type some data, then take your hands off the keyboard, pick up the game controller, hit the right or left buttons, then put your hands back on the keyboard. It should make you SO much more productive!
I have such a long history of thinking of Microsoft as the Big Bad in tech, that it's funny to think of them as the young upstart that's just coming into their own and standing up to the big guys for the first time. If it was early enough for folks to be arguing about what keys to use for functions, it must have been 1985, which means Microsoft was just coming to the end of their time needing to satisfy IBM in order to survive/thrive.
They still depended on IBM to some degree. If IBM stopped shipping Microsoft products on their PCs, it would hurt Microsoft quite a lot. But, clones had just begun to break out. Compaq and a few dozen other clone makers were exploding in popularity. I imagine Gates must have seen their orders from clone makers growing exponentially, and much faster than sales to IBM, and realized they didn't really have to kowtow to IBM, anymore.
The thing I find funny here is that MicroSoft in the 80s and early 90s was a scrappy bunch of hackers trying to get something out the door. But over time they've sort of become IBM. Based on the way things have progressed...
Microsoft has become IBM.
IBM has become CA.
Apple has become Microsoft.
Oracle has become DEC (if DEC had a few more lawyers.)
Amazon has become Oracle.
Possibly, OS/2 co-development started in 1985, which is the same year IBM released the a keyboard with arrow keys.
Of course, that assumes it came from a place of corporate strategy rather than individual habit, which could have been learned from other older systems.
I'd never really thought about it before, but Enter to advance to the next field field and Ctrl + Enter to submit the whole form (which is the typical keyboard shortcut for submitting the form while a multi-line text input control has focus) does have a certain appeal to it.
The overloading of return to either send a message or add a newline has become really annoying since chat apps (and then now AI) have become popular.
You have to keep a mental context of whether you need to hold shift before you press return. See also: every message I've ever sent that ended with I' because I fat-fingered the ' key while typing a contraction.
Shift+Enter will usually enter a newline in a message without triggering send... At least that's the convention used most of the time. No guarantees on specific applications, just my own experience with this.
Enter/return on the 'Submit' button, I suppose. The rationale may have been "Start at the beginning of the form, keep hitting Enter after filling in each field, and it will submit itself when you're done."
Some terminal software would use a function key that would be labelled "Execute". You'd usually have a template to put over the function keys to tell you what does what.
Interesting, I wonder if the "TAB" argument was IBM at the time wanting screen input to work just like they did on mainframes ?
Well before DOS was a thing, the mini I programed on was using Tabs to move between the TUI fields. Once you were happy you would press RETURN to process the data. At the time, seems IBM was trying to avoid doing anything similar to any of its competition.
I don't think even modern-day Microsoft is anywhere near as bureaucratic as IBM, but yeah - it seems almost inevitable that as a software company grows it'll lose an "undisciplined hacker" culture and become stuffier, for better or worse.
I find this story odd because IBM was consistent with their keyboard nomenclature across multiple products, and the 3270 series mainframe terminals used the Tab key, located in the same place where you would find a tab key on a modern keyboard, to move the cursor to the next field.
https://www.bitsavers.org/pdf/ibm/3278/GA27-2890-4_3278_Disp... (Page 73 of the PDF)
As an aside, it's worth noting that moving between fields was important enough on IBM terminals that they had a dedicated "back tab" key located on the opposite end of the keyboard to the tab key. On the original IBM PC, they decided to combine both functions into a single key. As a result, the tab key on the classic PC keyboard features the symbols for both forwards tab and back tab on the same key, the back tab symbol being on top to indicate that you need to hold down shift to use that function.
EDIT: The 5250 series terminals used the terms "Field Advance" and "Field Backspace" instead of Tab and Back Tab, but otherwise they used the same symbol on the keys, and the keys were located in roughly the same position as the 3270 series. Reference: https://www.bitsavers.org/pdf/ibm/5291/GA21-9409-0_5291_Disp...
Having worked at IBM, I would guess that using the tab key in this way was part of a patent they were pursuing and Microsoft's use would show this to be 'obvious' and thus not patentable. But that is just a guess.
In the 80's IBM had a whole class of high level technical people called "Systems Engineers" whose entire job description was to opine on the merits of any given system. Not write systems, not debug them, and certainly not to explain them, it was simply to opine "you're doing it wrong."
Microsoft is suffering from the lack of such a group today; they're definitely doing it wrong, where "it" is pretty much everything... except pissing off users.
Microsoft could implement the "Am I doing it wrong?" check via the shell script `/bin/true`
> I would guess that using the tab key in this way was part of a patent they were pursuing and Microsoft's use would show this to be 'obvious' and thus not patentable.
Something that's bothered me about user-facing patents:
Let's assume that the idea of using a keyboard key to move between input fields in a software form is not obvious, and in fact is a brilliant stroke of genius the likes of which the world is not likely to see again. If that one guy hadn't been born, we would have gone thousands of years with no method, keyboard-based, mouse-based, or otherwise, of moving from one input field to another input field. Every piece of software would use nonconfigurable timers, and you'd just have to hope you could type fast enough.
I don't see what the hypothetical benefit of extending patent protection to this brilliant idea is supposed to be.
Say you're the company who comes up with the idea. You can benefit by including it in your product, where all your users can see it. In other words, the benefit you get from coming up with this idea is that you can publish it for the world to see, and that's the only way you can benefit from it. A usability feature that your users cannot use or know about doesn't increase usability.
Even though the idea isn't obvious, the implementation is. If you disclose your brilliant idea, everyone will copy it and your advantage in the marketplace will be transitory.
So... what is the purpose of giving you a patent? That cripples the marketplace, but it fails to realize the benefit of patents, publication. Publication necessarily had to happen anyway.
> A usability feature that your users cannot use or know about doesn't increase usability.
Cannot is maybe doing a lot there. There's plenty of usability features that aren't really obvious or apparent unless you look very closely. Ex: pinball machines have timed shots, but there's almost always a grace period so if you contact the ball with your flipper around when the timer hits zero and it makes the shot, chances are you'll get credit for it even though the timer expired. That's a usability feature most users won't ever notice. At WhatsApp, I would never send an S40 user a verification code where the 4th digit was 8, because if you got a text message with 123-890, s40 would turn -8 into an 8th note emoji; until today, probably 3 people knew that ... but it dramatically improved usability.
> Even though the idea isn't obvious, the implementation is. If you disclose your brilliant idea, everyone will copy it and your advantage in the marketplace will be transitory.
> So... what is the purpose of giving you a patent? That cripples the marketplace, but it fails to realize the benefit of patents, publication. Publication necessarily had to happen anyway.
If I had gotten a patent on the 'avoid -8 in verification codes', then the technique would have been public for everyone to see. So publication for exclusivity / forced licensing is an exchange of value between society and the inventor. Of course, avoid -8 is pretty obvious, when someone testing the s40 client complains about getting an 8th note in their verification code message.
For an invention that must be disclosed to be used, society isn't really getting anything in return for exclusivity. Maybe promotion of progress, theoretically, I guess, in that whoever thinks of it first gets paid; leading more people to think about things?
Presumably it's to give you an advantage for putting in the work to develop it for a period of time.
You can say the same about swipe to unlock and that had been litigated to death.
I did say the same about swipe-to-unlock:
>> Something that's bothered me about user-facing patents
Here's a real IBM 3270 keyboard.[1] Note the "Next field" key on the left, and the matching "Previous field" key on the right.
The IBM 3270 was a device for filling up forms. The mainframe sent the terminal a form with blanks, and the terminal let the user fill in the blanks. The terminal hardware prevented the user from overwriting the static parts of the form, and could apply some other form constraints, such as numeric fields. That was all done by the terminal. When the form was filled in, the user pressed ENTER, and the completed form was sent to the mainframe as one transaction. This approach let one mainframe service huge numbers of terminals. The user never experienced delays while typing and could type at full speed, often without looking.
PCs didn't have that usage model. The PC crowd was thinking "typewriter". One of the first terminals for home computers was called the "TV Typewriter".
Web forms do have that model, but with less consistency.
[1] https://sharktastica.co.uk/resources/images/model_bs/themk_1...
Nitpick: The terminology used by IBM on the 3270 family (including the 3277 whose keyboard you shared) was "Tab" and "Back tab", not "Next field" and "Previous field".
Funnily enough, IBM had already published this. CUA explicitly says tab and backtab move between fields.
so they spent seven layers of management escalating against their own standard: https://archive.org/details/ibmsj2703E/page/n13/mode/2up
From what I remember, there were two "Enter/Return" keys on IBM 3270 terminals. One was the regular "Return" key we have today, which just advanced to the next field, and didn't submit the form. There was also another "Enter" key where the Right Ctrl key is today, and that submitted the form. So, I presume, instead of being against Tab key, IBM might be against "Return" to be the form submit key as people who use 3270 would expect it to advance to the next field.
And that was true for many DOS programs. Pressing Return would just go to the next field, not submit the form. That was one of the things that needed some getting used to on Windows.
Your memory is correct, and it's interesting to note that on the IBM terminal keyboards, the Enter key was marked "Enter", and the return/new line key was marked "↵". On the classic IBM PC keyboards such as the Model M, the Enter key is marked "↵ Enter". I believe IBM chose this to convey that the Enter key on the PC was both an "Enter" _and_ "Return" key in one. As you say though - individual applications got to chose what that meant in practice, leading to inconsistent behavior.
Also conspicuously missing from the story is what key IBM DID want to use. I mean... that's the first question you'd ask!
Lame.
Ctrl-F6 ?
Shift Ctrl F6
And this is a great example of why I read the comments on HN stories... thanks for the info!
RIP /.
As someone who prefers tabs (I'm not looking to argue), I once asked Brendan Eich on Twitter why he prefers spaces. His answer was more thoughtful than I'd expected.
The tab key itself is hijacked by modern OS/UI behavior. It makes it complicated to actually type literal tab characters in certain contexts, particularly in the browser.
I still prefer tabs (and I'm a Go developer), but he is absolutely correct about that being a pain in the butt. For instance, try getting a tab character into the text area on Hacker News
The fact that most people think the Tab key is the correct choice is a perfect example of why it was not.
It had a purpose, and it got hijacked and made its actual purpose more difficult to use.
It's not dissimilar to Apples initial Touch Bar and then removing the Escape key.
Average user might never use that key; average developer doesn't got long without using that key for its purpose.
It might be reasonable to have a separate key for "tab" vs "next field", as well as separate key for "line break" vs "send". (But, tab and line break are not applicable for all contexts.)
However, it might also be reasonable to have a key or key combination (some programs use ^V) to enter control characters as data rather than as commands.
It might also be a consideration when designing a new computer (which does not have to be the same as existing ones); I had thought about such things and may make such a consideration.
There's also the fact that no one seems to have tab stops set the same way.
That's a feature, not a bug.
As far as I'm concerned, it's a feature when the viewer controls the number of spaces per tab, it's a bug when someone else does (as is often the case online via css `tab-size`).
Non english speaker trying to wrap my head around what Brendan said to you.
I was once told that the tab key can be represented in different ways on different systems, and that's why spaces are safer because they're always represented the same.
Is that what Brendan was trying to say?
Literally try entering a "tab" character in a Hacker News reply. For me in Safari, it changes the focus to the "reply" button.
It's literally difficult to enter a "tab" character into many text entry fields and dialogs and applications, because it's used so often as a navigation key.
> I was once told that the tab key can be represented in different ways on different systems, and that's why spaces are safer because they're always represented the same.
The main counter argument is that users have different preferences for the amount of indentation, so giving them control over that, just as they (nowadays) have control over the font used and window width, is a good idea.
The tongue-in-cheek counter argument is that fixed-width spaces are preferable over ‘normal’ spaces. They also give you more control over indentation, allowing, for example, mixing usage of THREE-PER-EM SPACE (https://unicode-explorer.com/c/2004) for indentation with FIGURE SPACE (https://unicode-explorer.com/c/2007) for right-aligning numbers.
No. What Brendan said was:
> The tab key itself is hijacked by modern OS/UI behavior. It makes it complicated to actually type literal tab characters in certain contexts, particularly in the browser.
Tab key is both a control character moving the cursor to the next input and also an input representing the tab character as you see in a text editor.
Now that im thinking about it I’m convinced capslock would have been superior next field key and alt+capslock to be used for toggling capslock. But it’s not obvious to me capslock is even seen by the OS. It could be changed on the keyboards themselves.
It definitely is visible, OS login screens usually warn you if caps lock is on when typing password
Brendan Eich was making a point about the tab key also being used to switch to different text fields and buttons. This makes it difficult to type in certain applications. A space doesn't have that issue.
About your point of tab being represented different on different systems: It will always be ascii 9, how it's draw does differ between text editors but I consider that one of it's strengths for programming. Everyone can configure what an indent looks like for them, it makes reading code easier.
Little irritates me more than logging into a new system, opening up code in vim, and witnessing the insanity of tabs-as-8-spaces
No, the point was that pressing the Tab key often does something else than inserts the tab character. Moves focus to the next input field, for example.
No, it's that it's extremely challenging to insert a literal tab character into anything resembling a web form.
You can configure your editor so that pressing the tab key inserts a tab character. Good luck with a web browser, or a UI where tab moves focus to the next control. The space bar basically always inserts a space character.
BTW, Douglas Crockford had an interesting argument in favour of spaces, see https://www.youtube.com/watch?v=En8Ubs2k1O8
No he's saying if you're trying to write code in an edit box in a web browser (for example) and you press tab, instead of inserting a tab character it will move the focus to the next input field.
Bullshit reasoning though, because even people who use spaces for indentation don't do it by hammering the space bar - they also press the tab key. And of course in modern browsers we can give tab the expected behaviour.
The only logic I've ever heard for using spaces for indentation that actually makes sense and I vaguely agree with, is that lots of programmers do not give a shit about formatting code and even using spaces properly is often a bit too much for them.
That doesn't matter so much if you have an autoformatter though, so with the exception of Go I don't know why modern languages with widely used autoformatters don't use tabs.
Control-shift-u, 9, space/enter, works on most Linux systems.
To explain: control-shift-u allows entering a Unicode character by its hexadecimal code. This presumably depends on the Input Method Editor (IME) in use, which is something I've never fully understood, but this seems to work widely across different desktop environments (Xfce, KDE) and display servers (Xorg, Wayland).
A great read, although I'd still like to know what IBM's reasoning for opposing this use of the Tab key was.
Is it because they didn't want Tab to be both an input and a control character? I.e. there are some cases where you can type a Tab into an input field, and there are other cases where you can't, and it's not immediately obvious which ones are which?
All the way in 2026, I would still be sympathetic to this view.
Personally I don't like the tab key for field change.
Firstly it was a breaking change from dos. Dos programs used Enter. And enter meant you could capture numeric data using 1 hand, since the numeric keypad has an enter key.
That means left hand can stay on the (paper) source. Right hand types. People got fast at this. (Really fast). And this pattern lives on in some programs kline Excel).
Lots of people (ie my customers) hated needing both hands on the keyboard. Lots of our programs allowed mapping of enter=tab.
I should be clear. It's not the "name" of the key that matters, it's the location.
The dual-use of the key is just an annoyance we live with. Sometimes the key behaves as a navigator, but in other cases it behaves as a spacer. Daft. (Enter would have the same problem. )
The best solution (by far) would gave been to add another key to the keyboard. Preferably in the numeric key pad. We got lots of new keys in that era. Hindsight says we should have added a "move on" key at that time.
If you're designing a text editor that's intended to run within a browser you need to fully break user expectations about the functionality of the tab key - there are two completely logical and intuitive roles it can serve in such an environment that are in conflict.
The same issue arises (with much higher frequency) with Enter and even in the modern world I'm sure we all have a pretty complex ruleset we've committed of when ctrl+crlf triggers a newline, or a message send and what the corresponding behaviors of a bare crlf and shift+crlf are.
In HN's editor shift+crlf and a bare crlf cause a newline creation and ctrl+crlf does nothing - but often times ctrl+crlf will trigger a form/message/whatever submission, shift+crlf often causes a raw newline insertion (even when in a form context) and then a bare crlf might do one - might do the other or might even do neither! Those are the common bindings but I have seen exceptions and inversions of bindings with shift+crlf causing form submission while requiring ctrl+crlf for raw newline insertion.
All this stuff is just super annoying and causes a lot of user friction (and for a long time MSFT's style guide was considered a seminal reference for best practices as ironic as they may seem to folks these days).
The way I interpreted it was that there was no complete "IBM" reason. I read it as "one person in the IBM bureaucracy stepped in and brought things to a halt, which highlights the cultural differences" considering it is a post about said differences.
I now imagine a world where CAPSLOCK is used as "select the next input" and TAB as the character.
You mean the Control key to the left of the 'A' key? That's already the control key.
I assume so. Obviously there are very different concerns if you're managing an organization with countless moving parts versus you want to build something for the user quickly.
Maybe because it would compete with IBM mainframes / 3270 terminal emulators?
Weird because on the mainframe 3270 the tab key was used to move between fields, at least that is my recollection.
+1, always thought it was an IBM standard, move with TAB and commit the form with ENTER.
I wonder about the midrange systems. I never used AS/400's, but I know they have the concept of a a "Field Exit" key separate from <ENTER>.
That's exactly my memory too, using a tab was second nature and it irked me when I went to gui apps with a mouse and the tab order was wrong, mainly with visual basic apps.
It's been ages since I used an IBM green screen but I seem to recall that in our application Tab did function to move between fields, but I think there were also function keys reserved for this purpose.
I think we should connect game controllers to all machines so the arrow buttons move you between fields, the 'A' key takes you up a level (in hierarchical menus), the 'B' key takes you into a subordinate menu. So to move between fields, you type some data, then take your hands off the keyboard, pick up the game controller, hit the right or left buttons, then put your hands back on the keyboard. It should make you SO much more productive!
Check this out :) https://github.com/madewokherd/xalia#default-gamepad-control...
(It's intended to allow video games that use standard OS controls to be controlled sanely with a gamepad.)
What a terrific idea! You totally should file a patent. Meanwhile, we'll have to use the next best thing: the mouse.
I have such a long history of thinking of Microsoft as the Big Bad in tech, that it's funny to think of them as the young upstart that's just coming into their own and standing up to the big guys for the first time. If it was early enough for folks to be arguing about what keys to use for functions, it must have been 1985, which means Microsoft was just coming to the end of their time needing to satisfy IBM in order to survive/thrive.
They still depended on IBM to some degree. If IBM stopped shipping Microsoft products on their PCs, it would hurt Microsoft quite a lot. But, clones had just begun to break out. Compaq and a few dozen other clone makers were exploding in popularity. I imagine Gates must have seen their orders from clone makers growing exponentially, and much faster than sales to IBM, and realized they didn't really have to kowtow to IBM, anymore.
A real shame about OS/2, though.
The thing I find funny here is that MicroSoft in the 80s and early 90s was a scrappy bunch of hackers trying to get something out the door. But over time they've sort of become IBM. Based on the way things have progressed...
Microsoft has become IBM. IBM has become CA. Apple has become Microsoft. Oracle has become DEC (if DEC had a few more lawyers.) Amazon has become Oracle.
What did IBM want? Arrow keys?
Possibly, OS/2 co-development started in 1985, which is the same year IBM released the a keyboard with arrow keys.
Of course, that assumes it came from a place of corporate strategy rather than individual habit, which could have been learned from other older systems.
https://en.wikipedia.org/wiki/IBM_PC_keyboard
Enter/return commonly used elsewhere.
I'd never really thought about it before, but Enter to advance to the next field field and Ctrl + Enter to submit the whole form (which is the typical keyboard shortcut for submitting the form while a multi-line text input control has focus) does have a certain appeal to it.
The overloading of return to either send a message or add a newline has become really annoying since chat apps (and then now AI) have become popular.
You have to keep a mental context of whether you need to hold shift before you press return. See also: every message I've ever sent that ended with I' because I fat-fingered the ' key while typing a contraction.
Shift+Enter will usually enter a newline in a message without triggering send... At least that's the convention used most of the time. No guarantees on specific applications, just my own experience with this.
What would be 'submit' then?
Enter/return on the 'Submit' button, I suppose. The rationale may have been "Start at the beginning of the form, keep hitting Enter after filling in each field, and it will submit itself when you're done."
Some terminal software would use a function key that would be labelled "Execute". You'd usually have a template to put over the function keys to tell you what does what.
To me that sounds like the way MacOS avoids Home/End with alternative solutions that kinda work but are not great
(And yes I do miss those - with an external keyboard these get less painful but still don't work 100% like on a PC)
The latest in a long line of articles where "Raymond Chen" is my first thought before even opening it, just based on the title
Great read
This article seems to just be a dig at IBM without bringing any receipts or adding any substance. "A colleague told me that they said..."
Honestly, why should I even believe it?
So, what did IBM want to use instead?!?!
Interesting, I wonder if the "TAB" argument was IBM at the time wanting screen input to work just like they did on mainframes ?
Well before DOS was a thing, the mini I programed on was using Tabs to move between the TUI fields. Once you were happy you would press RETURN to process the data. At the time, seems IBM was trying to avoid doing anything similar to any of its competition.
IBM CUA FTW LOL
Interesting to think how Microsoft is today's Ibm having adopted that beurocratic culture and deep hierarchical org.
I don't think even modern-day Microsoft is anywhere near as bureaucratic as IBM, but yeah - it seems almost inevitable that as a software company grows it'll lose an "undisciplined hacker" culture and become stuffier, for better or worse.