Re. the Finalist 12 article:
When I read the part about “Mark Miller from Xerox Parc wanted to write Xanadu in Smalltalk,” I had to laugh, because that sounds just like something I would’ve wanted to do. 
Only thing was I had just graduated from high school when that part of the story was happening, and I didn’t know Smalltalk existed. 
The part that didn’t make sense to me was Xanadu being written on “PCs.” I guess they used Smalltalk/V from Digitalk? I was under the impression that a full Smalltalk system needed at least 1 MB of memory, and even then, not much could be accomplished with that. There was a 286 version of Smalltalk/V by 1988. It’d be interesting to know how much memory was added to their PCs. I guess to speed it up, they translated Smalltalk to C++, and compiled that. I had trouble understanding, though, why the C++ compile took a week. The only thing I feel I can compare it to is when I compiled MiNT kernel updates on my 16 Mhz Atari Mega STe, in 1993. I had MiNT and GCC set up, running on a hard drive. The kernel took an afternoon to compile, and I think the binary took up 256-384K of memory. A C++ compile would’ve been slower, for sure, and 286’s were likely running at half the speed of my STe, but it makes me wonder, “How big was this thing?”
I take Gwern’s point that Nelson thought of all this stuff before it had been practically tried out, and so there were bound to be some problems with it. Nelson tried to do “the whole thing,” when it would’ve been better to do trials of pieces of it, to see how well they worked out.
When I read over Nelson’s vision, I saw one big problem was the only way he would’ve been able to make it work was if the entire network, the entire Xanadu system was proprietary, mainly because of the tracking issues with copyrights, and enforcing payments. I’ve tried to think about what would need to be done to make something like it work on the internet. The only thing that came to mind was to use some kind of DRM, and turn Xanadu functions into network protocols, but of course, then the issue would be how to attract people enough to prefer it over the web.
I sort of agree with the criticism about the side-by-side reading scheme Nelson wanted to use. The only time I felt like I would’ve liked that was when I was reading a computer history book called The Dream Machine, by Waldrop. It was rather frustrating to read, especially when I was using it as a research source, because with many of the chapters, Waldrop had you “rewind” the clock back about a decade. The feeling was like, “Okay, I know I’ve gotten you from 1960 to 1970 with this one set of characters and projects, but let’s go back to 1960, because I have this other stuff to tell you.” I saw why he was doing it, because to intermix the stories together in book format would’ve been a mish-mash, distracting the reader, back and forth between different research projects it described that were going on simultaneously, but when I was using it for research, I felt I would’ve preferred to have the parallel stories side-by-side, because I was fine with forming my own narrative out of the source material. I could line up events together, as I felt appropriate for what I wanted to convey, and reading the stories side-by-side, in chronological order, would’ve made that more efficient.
I have sometimes thought that while Ted Nelson, Doug Engelbart, and Alan Kay have deeply wanted their thought processes about content to become mainstream in society, the only places that would feel an excuse to use it–where it would feel practical–are in universities. Though, I was struck that when Google promoted its Wave collaboration platform–a service that showed some glimmers of Doug Engelbart’s vision–that university faculty didn’t understand the first thing about how to use it, and avoided it. The students weren’t much better, just using it like a chat app., which professors felt distracted from their school work. Google spent much of their promotion efforts with Wave at universities, having the same idea that while the general public wouldn’t find it useful, universities were fertile ground. However, that turned out to be a bust, as well. Wave was cancelled in 2010.
It was interesting to read that there were some early prototypes of Xanadu produced, and demonstrated where Gwern could see them, and that the early versions were not practical for use. He said the text was indecipherable, because the display resolution was too low.
I very much like Nelson’s idea of two-way links, particularly for open source texts. I can see where this would be impractical for commercial UIs, though.
