Very cool, if true

A *brilliant* article, and really damn long iirc.

But amazing fucntionality. As with Wave, there's oodles of ways to do this stuff that we've never tried.

From:

Absolutely.

But the very ideas behind Xanadu are actually pretty silly - and make it the kind of thing you'd have to be paid to use.

Wave is designed from the user interface back - what would you like to do, and then how to do it. Whereas Xanadu was working from the back forwards - how should documents work, and then trying to make people fit into that, which was never likely to produce anything people actually wanted to use.

From:

True, but had Xanadu gone where the designer wanted, we might have all ended up assuming it was the default. Like, every time I go from my personal, searchable Gmail account to my work Outlook account, it's... retarded.

Xanadu's transclusion ability isn't wildly different to parts of Wave, and the audit trail is a bit of what wikis and Google Docs now do. But for the longest time, we had little or no trails in Office (not as a default, anyway).

From:

I'd also argue, having looked at certain n00bs, and done lots of usability testing, that interfaces are instructive. That is, Wave will tell us how to have conversations. It isn't just built on how we have conversations and adds some technological sparkle.

From:

AND, what ideas behind Xanadu are silly? Silly from a user POV ('I just want it to do what I want') or from a database developer ('It can't be done')?

From:

From an ontological perspective :-> And also from a technological one.

They wanted to build something that meant that you _couldn't_ copy data, you could only include it. Where I could quote you, quoting Bob, quoting Charlie, quoting the NY Times and the back end would put all those bits together, creating a chain all the way back - and enforce payment for all the bits that were copyright.

It's a massively locked down system - because they focussed on 'correctness' rather than 'utility'. So you couldn't do anything that would violate the integrity of the system. Whereas the systems that actually work best are ones where you've got the ability to put in a mish mash of stuff, tie it all together, and then caretake it a you go along. Which is what the WWW does.

So, for instance, in Xanadu, moving a page means updating all of the links that point to it. Desining _that_ into a system is insanely hard, because you have to know what points at your page. Which means a mechanism for telling every site you link to that you link to it, keeping track of that, updating whenever the list gets out of date, etc. Whereas with the WWW you just move the page - and if you feel like it you put in a redirection page saying "This page is now over here at XXX." - no design necessary, no tech necessary. It's not 'perfect' in that sometimes you'll lose pages, but it's a hell of a lot easier to manage in the first place, and doesn't lock you in to a particular way of working.

And solving the things that go along with the original 17 rules was the reason that Xanadu was _always_ 6 months away from working.

I hear that Google Wave is also having fun problems - because they, again, have lots of little problems they're trying to solve that each require heavyweight architectural solutions. They're not trying to be quite so complex and 'correct' in what they're doing, so I expect they'll eventually get there, but by the looks of things it's the kind of project that will be eternally owned by one company, because the interoperability is just too unwieldy to work well with others.

From:

Oh, from an ontological point of view, it's daft. ;)

But then you'd have no problem with citations and attribution, fewer problems with copyright and plagiarism... from a strictly academic user's perspective, it'd be amazing.

But then it's almost the opposite of small pieces, loosely joined. And that I'd have to cite who said that and where and when before submitting this post would also reduce the poetry, somewhat.

A Xanadish approach would have been interesting during 1993 - 1996, for me, as pages got lost and were deleted a lot, and the spiders couldn't keep track. Doesn't happen so much any more.

From:

It's a massively locked down system - because they focussed on 'correctness' rather than 'utility'.

This is the standard argument that the Web 'works' because of the 404 error. While 404 was something that distinguished the Web from contemporary open hypertext systems (namely Hyper-G and Microcosm) and Xanadu, the Web succeeded more because a) the protocol and data format definitions were freely available, and b) it was initially targeted at a user community rather than being a predominantly research system (as was the case with Hyper-G and Microcosm).

So, for instance, in Xanadu, moving a page means updating all of the links that point to it.

Moving a page? No such thing. Once a document is created with a given ID, it's there for perpetuity. If you want to refer to it by a different identifier, that's what transclusion is for.

From:

a) the protocol and data format definitions were freely available, and b) it was initially targeted at a user community

Also - it was (fairly) simple to write both a client and a server. Xanadu was, from my understanding, insanely complex, because of all the situations it had to handle.

From:

Not clear. The Front End-Back End protocol (from Xanadu Green, the version described in Literary Machines) would have pretty straightforward to implement on the client side, mainly because all of the heavy lifting was being done by the server. The Back End-Back End protocol (which was what I had to ask Ted about), which would have been key to the implementation of the servers, is a different matter.

The other key to the server side is the enfilade data structure. Ted didn't publish anything about enfilades until fairly recently (in the Udanax source release, as source code) because he believed them to be such a good idea that they were worth retaining as a trade secret. I haven't implemented enfilades, but from what I can tell, they should be no harder to implement than any other moderately complex hierarchical data structure.

From:

As an aside, I'm increasingly struck that in the Web of the early 1990s (pre-Netscape) it was much easier to write browsers and servers than it is now.

HTTP/1.0? Doddle. HTML 2? Still pretty easy (and easier yet if you took the lazy route and didn't try to parse it as SGML first). No CSS, SVG, Javascript/ECMAscript, Flash. By 1995, I'd written special-purpose standalone servers and simple clients. I wouldn't want to try that now.

From:

Oh, absolutely.

Which reminds me of another quote - that every successful complex system started off as a successful simple system.

If the original spec had been HTML5+CSS+Javascript then I doubt it would have got far, the initial barriers would have been too high.

From:

Don't get me started on HTML5 (and its standards process(.

From:

Now there's a post you should make!

From:

It verges on ad hominem, so is best delivered in person.

From:

Well, next time you're in Edinburgh I shall happily buy you a pint in order to hear it!

From:

But the very ideas behind Xanadu are actually pretty silly - and make it the kind of thing you'd have to be paid to use.

Why so? You're saying this from the perspective of someone who has been using the Web for at least a decade, possibly a decade and a half. If that had not been your default view of a global online document system, Xanadu might not seem so strange.

From:

Because it's so incredibly locked down. The Web works because you can do anything with it, and it doesn't get in the way - Xanadu is pretty much the opposite of "small pieces, loosely joined", which gets in the way of innovation.

From:

True enough. In that sense, Xanadu would have been more like the French Minitel.

From:

"Xanadu is ... the opposite of "small pieces, loosely joined""

See, if we'd been using Xanadu, I could have charged you for quoting me. Quoting David Weinberger.

From:

And thankfully we're spared that lunacy!

From:

But what I would like to do may not fit with the programmer - as a common criticism of open source - or with the market - as a criticism of capitalism.

Also there is the whole issue of consciousness, or false consciousness, here - what do I want?

From:

Ted really, really, *really* dislikes that article and Gary Wolf. I once made the mistake of mentioning to him that I'd read it.

From:

Doesn't surprise me. Was it because it wasn't accurate enough, or because it was too accurate? :->

Oh - and I was hoping you'd show up, my knowledge of Xanadu is entirely third-hand, and I think of you as rather an expert in this area. Thanks for correcting the stupider things I day :->

From:

Some from column A, some from column B. Wolf may have misrepresented some of the Xanadu team's work during the Autodesk years, but I suspect that it was mostly Ted's pride being needled.

I like Ted - he's a personable guy, and he has made a significant contribution to computer science with the thought experiment that Xanadu has been. He's also deeply bitter (partly with reasonable justification) and irascible. I've often been struck by the thought that there are strong parallels between Ted and Charles Babbage.

From:

martling.livejournal.com

GNU Hurd fits somewhere between the two.

From:

I hear that's coming in 6 months!

From:

Yes

What is lacking here - http://www.gnu.org/software/hurd/

Is a download, try and use burn ISO

Too much forward thinking, too little pragmatism.

From:

lionsphil.livejournal.com

There is a Debian/HURD distribution with a LiveCD out there somewhere—I've run it under a VM before. It's fairly pointless.

Frankly, I wish Minix gained some more (competent) attention, because that's actually a good design step that solves some hard problems, and that hard bit is largely done AIUI. It just needs more drivers.

From:

supergee.livejournal.com

Ted Nelson, like Bob Wills, is still the king.

From:

You missed the third option: the Analytical Engine.

After that, it would have to be Ted's dream, despite the fact that there have been source releases of two partial implementations (Xanadu Green and Xanadu Gold). I talked with him about some of the background details when I was writing my thesis (there are some bold assertions in Literary Machines regarding distribution), and he admitted that a) that stuff had never been written and b) the Xanadu team had never been able to work out how they could effectively distributed open hypermedia.

From:

Oooh, yes. I wonder if anyone will ever produce a working Analytical Engine. Maybe once we have a post-scarcity society there'll be enough free time :->

From:

It would be a hard task, especially given that he'd never finished the design. The credit for the Difference Engines (No.2 in particular) lies as much with Clements (and latterly Doron Swade's team at the Science Museum) as it does with Babbage; while Babbage's basic idea is sound, there were a lot of detailed engineering problems that needed to be resolved before it would work.

For this reason, we're more likely to see a mechanical computer fashioned after Babbage's Analytic Engine plans than we are to see a working Analytical Engine per se.

If such a machine were ever built, it would most likely be made from silicon rod logic at nanoscale.

From:

henriksdal.livejournal.com

oh.. I thought we were talking about.. never mind.

From:

The Analytical Engine?

From: