[Caveat, this is about the US centred mainstream development. There might have happened way more, prior or at other places and around he world (*1)]
I don't have any hard point here, but I would imagine it to be somewhen between 1968, 1973, 1976 and 1979 and hard-linked to the development of the GUI.
1968 NLS - No Vertical Bar
Doug Engelbart's Mother of all Demos of 1968 (*2) shows clearly that the NLS, which premiered so many firsts of the GUI/Desktop/Network age did not feature a vertical cursor - in fact, it did not feature a text cursor as we would expect at all, but:
- Typing was done without a cursor, like on a (glass) TTY (*3)
- The mouse could be used to mark elements,
- not like today as text or field marking, but
- placing a 'mark' in the middle of the element (word, paragraph, letter).
- The mark was a half height circle, like the lower case letter 'o'
- The Mark was drawn at the center and on top of the element marked.
- What was to be marked (letter, word, phrase, ...) was to be selected first on a chorded keyboard (*4)
- Likewise a keystroke, after marking, told what should happen with that item.
[All to be seen here]
Sounds strange? Then consider that this system, despite being a gigantic for its time (96 MiB Memory in 1968), not even supported mixed case. Only uppercase letters were displayed - with 'true' upper case being emphasised by having a top bar. All despite this being a high resolution graphics system.
A great pointer abut how the environment is a (maybe a bit overemphasising) statement by Andy van Dam, that attendance was blown by the Demo, but went back to work with their paper TTY - as even glass TTY was still a thing of tomorrow in 1968.
I always think of this demonstration as a perfect example of how much evolution was needed from the first steps, which already include all ideas, to the implementation we use today and of which we think it's the only way it ever was and ever could be. It isn't.
1973, 1976 and 1980 Xerox Got Stroked
A good part of the NLS team went to Xerox and developed the well known Alto system which at some point led to the Star software as well as Apples copy thereof. The first units were finished in 1973, while production units only became available as late as 1979. The Alto boots into a Text-UI, which features a vertical text cursor as seen in this video clip (*5).
Likewise does Smalltalk-76 running on the Alto use that vertical bar, as seen by Dan Ingalls typing in this demonstration.
1983 Smalltalk-80, One Step Ahead
Interestingly, Smalltalk-80, still on Xerox machines, did not use a vertical bar, but a caret, a small upward pointing chevron at the position where an insert would happen, as seen in this 1983 demo.
It was as well Smalltalk-80 that finalized the names for either marker to cursor meaning the mouse pointer, while the text pointer was now the caret.
Smalltalk 80, as Version 2, is also what became more widely known outside Xerox due to its ports to various systems, including the IBM-PC, eventually cementing the impression that Smalltalk always used the chevron style caret.
Coincidentally this hints to the origin of Apple's GUI(s), which used a vertical bar (*6) being a descendant of the Xerox GUI before the chevron style caret was introduced. This fits the timeline as the Lisa project was started in 1978, following the Smalltalk-76 based one.
The Lisa and later Mac OS became the over all guideline for mainstream (PC) GUI development, so we got stuck with continuing developing advanced concepts from an already outdated starting point.
*1 - For example that the mouse was developed (presumably independent) in 1965 by Rainer Mallebrein. Unlike Engelbart's mouse which rubbed two wheels across the set, it was already based on a ball - it also featured a single button, much like Steve Jobs "invented" many years later for the Lisa :)
By the time Engelbart demoed his prototype mice, the Telefunken Rollkugelsteuerung RKS 100-86 could already be ordered as production model for the Telefunken TR440 mainframe system.
Mallebrein BTW joined Zuse in having his patent application rejected as not sufficiently high enough to be worth one given.
*2 - 9th of December 1968 at the Fall Joint Computer Conference in San Francisco.
*3 - Seemingly people in the 1960s did not see a cursor as something indisputable - or needed at all - one knows where one is typing, right? A great reminder that we have long evolution here. The addition of a cursor as part of the generated display was an unbelievable step ahead making input and later documents fluid, leading to the way we manipulate information today.
*4 - The basic idea was that one hand is on the chorded keyboard selecting functions and the other on the mouse pointing where that function was to be applied. NLS relied on that while the Star systems still show it in more refined but restricted way with their left hand function keys - something I still miss. Heck, it was even great on text only to have function keys to the left, like on the first IBM-PC. Stupid AT Keyboard moving them up :(
*5 - The video also shows another forgotten evolutionary step made by Smalltalk, as the scroll bar does move the text go up and down by clicking above or below the positional box, but depending on the side of the bar one clicks at: left for down, right for up :))
*6 - And a similar shaped muse cursor when over editable text.