Bit-level skeuomorphism! And since NUL is zero, does that mean the program ends wherever you stop punching? I've never used punch cards so I don't know how things were organized.

ASCII was started in 1960. A terminal then would have been a mostly-mechanical teletype (keyboard and printer, possibly with paper tape reader/punch), without much by way of "circuit logic". Think of it more as a bit caused a physical shift of a linkage to do something like hit the upper or lower part of a hammer, or a separate set of hammers for the same remaining bits.

Look at the Teletype ASR-33, introduced in 1963.

I still wonder if it wouldn't have been better to let each digit be represented by its exact value, and then use the high end of the scale rather than the low end for the control characters. I suppose by 1970 they were already dealing with the legacy of backwards-compatibility, and people were already accustomed to 0x0 meaning something akin to null?

And this is exactly why I find the usual 16x8 at least as insightful as this proposed 32x4 (well, 4x32, but that's just a rotation).

I guess it's an age thing, but I thought this was really basic CS knowledge. But I can see why this may be much less relevant nowadays.

Yes, the diagram just shows the ASCII table for the old teletype 6-bit code (and 5-bit code before), with the two most significant bits spread over 4 columns to show the extension that happened while going 5→6→7 bits. It makes obvious what was very simple bit operations on very limited hardware 70–100 years ago.

(I assume everybody knows that on mechanical typewriters and teletypes the "shift" key physically shifted the caret position upwards, so that a different glyph would be printed when hit by a typebar.)

Tangentially related, there is much insight about Unix idioms to be gained from understanding the key layout of the terminal Bill Joy used to create vi

https://news.ycombinator.com/item?id=21586980

Not 'man ascii'?

A similar bit-flipping trick was used to swap between numeric row + symbol keys on the keyboard, and the shifted symbols on the same keys. These bit-flips made it easier to construct the circuits for keyboards that output ASCII.

I believe the layout of the shifted symbols on the numeric row were based on an early IBM Selectric typewriter for the US market. Then IBM went and changed it, and the latter is the origin of the ANSI keyboard layout we have now.

xor should toggle?

Modern keyboards = some keyboards. In the Nordic Countries modern keyboards have parantheses on 8 and 9.

What happened to this block and the keyboard key arrangement?

  ESC  [  {  11011
  FS   \  |  11100
  GS   ]  }  11101

Also curious why the keys open and close braces, but ... the single and double curly quotes don't open and close, but are stacked. Seems nuts every time I type Option-{ and Option-Shift-{ …

  | ASCII      | US keyboard |
  |------------+-------------|
  | 041/0x21 ! | 1 !         |
  | 042/0x22 " | 2 @         |
  | 043/0x23 # | 3 #         |
  | 044/0x24 $ | 4 $         |
  | 045/0x25 % | 5 %         |
  |            | 6 ^         |
  | 046/0x26 & | 7 &         |

It's more that shift flips that bit. Also I'd call them bit 0 and 1 and not 5 and 6 as 'normally' you count bits from the right (least significant to most significant). But there are lots of differences for 'normal' of course ('middle endian' :-P )

I guess in this system, you'd also type lowercase letters by holding shift?

Going off the timelines on Wikipedia, the first version of ASCII was published (1963) before the 0-9,A-F hex notation became widely used (>=1966):

- https://en.wikipedia.org/wiki/ASCII#History

- https://en.wikipedia.org/wiki/Hexadecimal#Cultural_history

    x x x x 0 0 0 0
    x x x x 0 0 0 1
    x x x x 0 0 1 0
    ....
    x x x x 1 1 1 1

Currently 'A' is 0x41 and 0101, 'a' is 0x61 and 0141, and '0' is 0x30 and 060. These are fairly simple to remember for converting between alphanumerics and their codepoint. Seems more advantageous, especially if you might be reasonably looking at punchcards.

[0-9A-Z] doesn't fit in 5 bits, which impedes shift/ctrl bits.

I'm not sure if our convention for hexadecimal notation is old enough to have been a consideration.

EDIT: it would need to predate the 6-bit teletype codes that preceded ASCII.

They put : ; immediately after the digits because they were considered the least used of the major punctuation, so that they could be replaced by ‘digits’ 10 and 11 where desired.

(I'm almost reluctant to to spoil the fun for the kids these days, but https://en.wikipedia.org/wiki/%C2%A3sd )

What are you trying to achieve, none of those characters are printable, and definetly not going to show up on the web.

    for x in range(0x0,0x20): print(f'({chr(x)})', end =' ')
    (0|) (1|) (2|) (3|) (4|) (5|) (6|) (7|) (8) (9| ) (10|
    ) (11|
          ) (12|
    ) (14|) (15|) (16|) (17|) (18|) (19|) (20|) (21|) (22|) (23|) (24|) (25|)    (26|␦) (27|8|) (29|) (30|) (31|)

Ok this set does not even show on Android, just some boxes. Very strange.

Smaller, 6-bit code pages existed before and after that. They did not even have space for upper and lower case letters, but had control characters. Those codes directly moved the paper, switched to next punch card or cut the punched tape on the receiving end, so you would want them if you ever had to send more than a single line of text (or a block of data), which most users did.

Even smaller 5-bit Baudot code had already had special characters to shift between two sets and discard the previous character. Murray code, used for typewriter-based devices, introduced CR and LF, so they were quite frequently needed in way more than few years.

Maybe 32 was a bit much, but even fitting a useful set of control characters into, say, 16, would be tricky for me. For example, ^S and ^Q are still useful when text is scrolling by too fast.

On top of the control symbols being useful, providing those symbols would have reduced the motivation for Unicode, right?

ASCII did us all the favor of hitting a good stopping point and leaving the “infinity” solution to the future.

I started using the separator symbols (file, group, record, unit separator, ascii 60-63 ... though mostly the last two) for CSV like data to store in a database. Not looking back!

It is interesting that, as a guess, we waste an average of ~5% of storage capacity for text (12.5% of Unicode's first byte, but many languages regularly use higher bytes of course).

I don't fault the creators of ASCII - those control characters were probably needed at the time. The fault is ours for not moving on from the legacy technology. I think some non-ASCII/Unicode encodings did reuse the control character bytes. Why didn't Unicode implement that? I assume they were trying to be be compatible with some existing encodings, but couldn't they have chosen the encodings that made use of the control character code points?

If Unicode were to change it now (probably not happening, but imagine ...), what would they do with those 32 code points? We couldn't move other common characters over to them - those already have well-known, heavily used code points in Unicode and also iirc Unicode promises backward compability with prior versions.

There still are scripts and glyphs not in Unicode, but those are mostly quite rare and effectively would continue to waste the space. Is there some set of characters that would be used and be a good fit? Duplicate the most commonly used codepoints above 8 bits, as a form of compression? Duplicate combining characters? Have a contest? Make it a private area - I imagine we could do that anyway, because I doubt most systems interpret those bytes now.

Also, how much old data, which legitimately uses the ASCII control characters, would become unreadable?

only that would have broken the whole thing back in the days ;)