I took online courses on PLATO and wrote simple games for it. The hardware was laughable by today's standards - plasma screens that glowed orange text (and lines!) from a dark blue background with touch input provided on about a 1/2" grid coupled with hideously clunky keyboards having their own special function keys - but it was reasonably reliable and allowed some of the first really large scale research on CHI.
Not that anyone other than researchers actually gave a crap about that last part.
But the system was fun to write programs for. It had a pretty OK language for the day, called TUTOR, that contained necessary primitives to make it Turing complete along with others to let you write onto the screen in a variety of ways. Again, pretty primitive by today's standards, but enough to teach programming with - they were debugging the interpreter (I think Fortran) and I played with it once. Pretty advanced for the time with breakpoints being highlighted.
And of course this is back in the late 1970's. Before the PC was a gleam in IBM's eye. The whole thing ran as on a huge CDC 6600 running a custom OS (as many were, in those days). Odd instruction set by an even odder designer you may have heard of - guy named Seymour Cray. Quirkier than hell with 60-bit words, 18-bit address space, and 6-bit bytes (yes, we spoke octal). But that was back in the day when minicomputers were eating the lunch of the mainframe boys. CDC, whom the University of Illinois partnered with to productize the system, couldn't muster the resources or talent to market this system while swirling down the toilet.
And, like so many things in computing, we see progress, good ideas thwarted by, well, nothing but the fact that people are short-sighted and, if something doesn't make a buck for someone, we drop it on the floor. So it goes...