Hacker News new | past | comments | ask | show | jobs | submit login
Project Xanadu (xanadu.net)
103 points by shortlived on Apr 21, 2019 | hide | past | favorite | 32 comments

Always interesting to see Xanadu bubble back up. This Wired profile from 1995 (!) is required reading on the history of the project: https://www.wired.com/1995/06/xanadu/

And this, Ted's response, is also required reading. http://xanadu.com.au/ararat

Related: On April 15, 2019, Ted Nelson published a video on Youtube with the title 'Pre-Final Reply to "The Curse of Xanadu" by Gary Wolf / Gory Jackal' [1]

[1] https://www.youtube.com/watch?v=-_-5cGEU9S0

Awful landing page for what seems to be a rather neat idea. Once I dug in and figured out what Xanadu actually is about, I like it a lot.

"This post cost you $.02 to download and read"

^^ this is the crux of "project xanadu". They wanted a copyright money speedometer to assess every possible transaction. And there was a promise that individuals would have things to say that would go viral as well.

I'm glad we got the 'default free' internet, rather than paywall hell for everything.

it's much more than just micro-payments

Micropayments are bad enough, there’s no need to go into more details.

I think they are an afterthought on the attribution/quoting mechanism. It's very well worth to look at the rest too.

After all it's about the Idea, not the decision to use this specific product (as it's still not in a usable state as far as I can tell).

Previously on Hacker News, only 4 months ago.[1]

[1] https://news.ycombinator.com/item?id=18635123

So to summarise, from what I can gather by browsing the various links, it is essentially the web but with a built in payment system for document publishers, and mechanisms for publishing new documents using sections of existing documents. What did I miss?

I thinks today it's more a wiki-styled CMS with block-based documents for DRM & Payment, plus automatic content-proxys. Though, some of the ideas were around with WWW too, like traceback/pingback for back-communications between documents, or builtin editing in early web-browsers. But the world evolved different.

The links never go bad because you should never delete data. Also old versions are kept forever.

How do you reconcile this with the reality of systems breaking?

To make it work then you just can't rule out by fiat a hard-disk breakage as one of the simplest adversarial conditions to the model.

Presumably by replicating extensively? How this was supposed to work in 1960 tech (and storage prices), I have no idea.

Xanadu is full of missing pieces like this one. There is a reason it never took off.

Links go both ways (is one way to put it) and the historical aspect I guess.

As a resident of New Jersey, I initially thought this was going to be something poking fun at the never-ending construction project formerly known as "Xanadu": https://www.nj.com/bergen/2018/06/american_dream_how_did_a_s...

There are several other options, if we’re discussing what this post isn’t about.


Ted Nelson recently did a series of videos on the important ideas in Xanadu. Start here: https://www.youtube.com/watch?v=hMKy52Intac

Also you can buy his classic 'Computer Lib' book here: https://computerlibbook.com/

The thing which always puzzled me about Xanadu is that all the attempted implementations seem to have been written by other people. Drawing from Wikipedia:

* "In 1972, Cal Daniels completed the first demonstration version of the Xanadu software on a computer Nelson had rented for the purpose

* "While at Autodesk, the group, led by Gregory, completed a version of the software, written in the C programming language"

* "Then a newer group of programmers [at Autodesk] used the problems with this software as justification to rewrite the software in Smalltalk"

From the Xanadu website, regarding their demo XanaduSpace: "The enthusiastic and talented programmer, Rob Smith, of Manchester England, did a beautiful job combining a lot of our ideas [...] John Ohno and Jonathan Kopetz, in Connecticut, spent several years trying to refactor it, but it's beyond fixing-- though as a demo it works just fine."

The related ZigZag project: "The first prototype consisting of two character-graphical views was implemented as a Perl module by Andrew Pam in 1997", and there's a more current implementation by some other people( https://sourceforge.net/p/gzigzag/_members/) called GZigZag

If I had been trying for NEARLY 60 YEARS to get a software project out the door, I would have probably started trying to code it myself after the first decade or so! Every time Xanadu comes up, I'm consistently puzzled at how there's never any sign of Nelson actually doing the grunt work of implementation!

Edit: here's an interesting link from the ZigZag wikipedia page, where a GZigZag implementor talks about their experience (http://lambda-the-ultimate.org/node/233#comment-1715):

> The GZigZag project did not materialize Ted's dream system as fast as he liked, so he withdrew his support. At first it just meant that he wasn't participating in discussions as much as he used to. At some later time he asserted his trademark (he had earlier given his permission to use it), and we changed the name to Gzz. Around the same time I was burning out, partially due to Ted's distrust and a large part due to project internal chemistry problems. I bailed out on the project in the end of 2002.

> Some time after my departure from the project, Ted asserted his ZigZag patent. The project was forced to abandon all ZigZag related stuff; it was reformed as Fenfire by salvaging those innovations that were not inseparably linked to ZigZag and using RDF as the new core. I hear that's not the last bit of trouble from outside parties the project has run to, but I cannot discuss even the little that I know in public. Suffice to say that even Tuomas burned out on the project eventually.

In 1988, Autodesk (makers of AutoCAD) was so impressed by the Xanadu project that they gave them financial backing. After four years, Autodesk gave up:

> […] Come 1992, the “resources of Autodesk” were still funding “talent of the Xanadu team” which had not, as of that date, produced anything remotely like a production prototype—in fact, nothing as impressive as the 88.1x prototype which existed before Autodesk invested in Xanadu. On August 21, 1992 Autodesk decided to pull the plug and give its interest in Xanadu back to the Xanadudes.

(Quoted from footnote linked from this page: https://www.fourmilab.ch/autofile/e5/chapter2_64.html)

The Wikipedia article and a few other sources also says their criticism of the web is its broken links. But if I read his EDL files correctly, transclusion is done with byte offsets within text files served over http, which obviously is extremely brittle without versioning in the protocol. (I guess you could do versioning by appending rather than editing, so that old offsets work, or handle versioning within the URL. Personally I think a feature like git hashes would be helpful, where contents are effectively sealed after publication.)

You omitted the first browser, Silversmith. From Wikipedia, "History of the Browser": "Another early browser, Silversmith, was created by John Bottoms in 1987.[9] The browser, based on SGML tags,[10] used a tag set from the Electronic Document Project of the AAP with minor modifications and was sold to a number of early adopters. At the time SGML was used exclusively for the formatting of printed documents.[11] The use of SGML for electronically displayed documents signaled a shift in electronic publishing and was met with considerable resistance. Silversmith included an integrated indexer, full text searches, hypertext links between images text and sound using SGML tags and a return stack for use with hypertext links. It included features that are still not available in today's browsers. These include capabilities such as the ability to restrict searches within document structures, searches on indexed documents using wild cards and the ability to search on tag attribute values and attribute names."

Silversmith implemented a portion of Ted's vision and I have discussed it with him from time to time. Ted developed Hypertext fixed links for books and we extended that to Hypertext links across the Internet, along with returns to multiple sources which was not done in books.

I was not able to get funding to continue the work. TimBL's innovation was in 1.) soliciting funding from the 10 large companies that funded the Web, and 2.) embedding advertising on web pages and in the content. But along with this was the funder's specification of how the WWW should work.

My descriptions of Silversmith were removed from wikipedia at W3c's request and Silversmith is now conveniently a "precursor" and most people believe Netscape was the first, when in fact it was released about 8 years after Silversmith The government was the early adopter and we released 4 different browsers in the first year, the last of which was a semantic browser for the U.S. Army Material Command. Silversmith included a browse-able index of each document, automatic markup of submitted texts, a parser to automatically check the page composition and it was a distributed system. The 5th browser was Erwise, https://en.wikipedia.org/wiki/Erwise, which Tim looked at before he started on his browser, the W3c implementation (the 6th). He declined to pursue it because the comments in the code were written in Finnish. Our other browsers were for the DoD Computer Aided Logistics Support (CALS) documents, for DoD Tech Manuals, and one for Old Norse documents written in Old Norse.I designed the browser to aid in the development of ontologies for AI. We have been working in the field for 33 years. My hope is that our next browser will include Ted's vision of draft versions and payments.

Are there any links to silvesmith resources? It sounds pretty amazing, and it would be nice if it were written down somewhere. I’d be curious to see what was forgotten along the way to v6.

Also, I see the Erwise Wikipedia screenshot uses the phrase “world wide web” in the authors box, which is kinda funny, since the article goes on to say Tim Berners Lee later invented the world wide web.

It's because Ted is not a programmer himself.

Leo Laporte interviews ted nelson in 2014 https://twit.tv/shows/twit-bits/episodes/199

The longest-lived vaporware in history

I wouldn't necessarily label it as vaporware, but I would call it the most disappointing great idea in the history of computing.

A similar thing happened with Chiron: http://maf.directory/misc/chiron.html

Quote from Donald Knuth for context: "[Floyd] was destined to be disappointed that his dream of a new, near-ideal programming language called Chiron was never to be realized--possibly because he was reluctant to make the sorts of compromises that he saw me making with respect to TEX. Chiron was "an attempt to provide a programming environment in which, to the largest extent possible, one designs a program by designing the process which the program carries out." His plans for the Chiron compiler included novel methods of compile-time error recovery and type matching that have never been published."

The Copernican revolution had the same problem

Along with Augment

I wonder if Herzog's list of 'sane people in computing' would have jumped to 2 had he met Terry Davis.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact