I think Internet is much better than Xanadu since it not only shipped, but is at 2.0 already. Release date is indeed feature, and I almost never use software that fail to include it.
"There is almost no reliable centralized form of identity on the internet…"
domain registration authority have control over internet…
“Links are the fundamental building blocks of the web.”
I stopped reading right there.
OpenID tries to fix the “no universal identity whatsoever” problem. It might not be here YET but it definitely does come soon. Advances in strong online identity and privacy also bring along strong anonymous and pseudonymous capabilities, which do not exist either. Yet.
Every new technology has its eccentrics who see things differently - Tesla springs to mind for electricity. Nelson seems to fit that mould. If you can find his books, read them with that in mind, and enjoy them.
WOW!!! great article man. Thx
Someone once said ‘I get my best ideas from misunderstanding Ted.’ That’s not a bad legacy, really.
Ironically, he has several storage lockers full of paper notes.
My favorite Ted Nelson quote: “Of course it [Xanadu] is real. You know it’s real because it crashes.”
Not to pick on Andrew Dalke’s criticisms, but I do want to defend Xanadu a bit because it is fairly well thought through, and it’s unfair to attack it as if it were something that someone was trying to create today rather than 20+ years ago.
It’s all too easy to superficially criticize superficial and general descriptions rather than really grapple with the specific technical and design details of some proposed software.
@Andrew Dalke on October 12, 2009 7:52 AM
The current web may be “fundamentally broken” but the ideas of Xanadu are far more broken."
The fact that a bullet list of a few goals or guidelines for the system are overly general makes them broken?
One thing to realize is that there are some assumptions about the software industry/environment that seemed reasonable when Xanadu was first defined that don’t really apply to the web and the open Internet today. The basic one is that Xanadu would be one self contained proprietary system offered and controlled by one or several cooperating companies over dedicated connections (probably via modem) which made perfect sense as one of with many existing online service of the 70s and 80s.
The fact that the Internet is now pervasive and the only realistic (and desirable) way to deliver remote computer services does mean that if you were to redesign Xanadu today, you’d need to use completely different assumptions about how people would be actually accessing Xanadu.
Think of it this way, the Xanadu design is for a particular application that today you might be able to build on top of the Internet. Don’t conflate the particular goals of Xanadu with the way the overall Internet and web work today, that doesn’t make sense.
- “Every Xanadu server is uniquely and securely identified.” Please define “server.” Does it include multiple machines used for load balancing? If I bring a backup online is it the same as the original server? Sharding?
These aren’t hard problems to figure out… use some imagination.
- “Every user is uniquely and securely identified.” Different groups have different security models. Each person has different roles, with different security requirements. And since before the days of the Federalist Paper’s there’s a recognized benefit to anonymity. That’s not saying what we have is good, I’m saying that this rule does not reflect reality.
Yes, this is simplistic and problematic. You would need to think through how to ensure anonymity if you wanted it. But nobody thought about anonymity in the Web either. You have a sort of pseudo anonymity because most of the time, nobody wants to bother tracking your IP address, tracing it through your ISP etc. But that doesn’t stop the government from trying (and ISPs from cooperating) when they want to try. Any true anonymity (Tor etc.) on the Internet are special add-ons to the system that require extra steps and restrictions on users.
- “Every document can contain a royalty mechanism at any desired degree of granularity …” How can that even work?
It works because of the way text in Xanadu is structured. There isn’t actually any singular “document” or “page” the way there is on the web.
Also, it says “can”, not “must”. (Literary Machines and papers on Xanadu have some good arguments for how granular royalties in Xanadu are much fairer for the reader than the way current royalties, licenses and other payment systems work.)
And if the local laws (“fair use”, “copyright term extension”) disagree, is there any workaround?
Well, it does rely on the publisher following the law, and allowing fair use. Maybe that’s an idealistic view?
And if as a publisher you overly restrict fair use of your text, or set too high a fee for access, nobody is going to ever use your text, and you won’t get paid, anyway.
- “Every document is uniquely and securely identified.” I’ve been trying to understand this one, especially in the face of versioning.
Again, the way “documents” are defined in Xanadu solves this.
An identifier for any portion/span of text (of whatever version) can be constructed. This is in fact the main innovation of Xanadu.
…
Nelson is primarily a scholar, he’s only a software designer by accident really.
As a scholar, he sees a liberal writing/quoting/editing/reuse ecosystem and economy where monetary payment is automatic as fair and efficient. It’s really a completely efficient, free market approach to writing.
Whether it would work in today’s environment we don’t know, nobody’s really tried it. A lot of people call Xanadu a failure, but it didn’t really fail on its merits, but through its mismanagement as a project.
The thing I like about Gabriel’s Internet Fuckwad Theory is that it’s been cited in scholarly works about the Internet and anonymity.
Interestingly, I think the document system DOORS achieves a certain amount of the Xanadu two-way link concepts.
Looking over the Wired article, the response letter, and the 10-year reflection by Wired, I think Xanandu seems like a Vaporware project. It has great ideas, lousy implementation, monolithic design, and a few mad geniuses.
Nelson’s Computer Lib/Dream Machines and Literary Machines are both great books and still well worth reading. Nelson’s a fascinating writer; his books bubble over with ideas; unlike, say, most of the comments on this post.
Remember, the ideas for Xanadu predated personal computers by 15 years, the Internet as we know it now by 25.
I find it ironic that you linked to the Wikipedia article on Xanadu but not the actual Project Xanadu site.
I think Internet is much better than Xanadu since it not only shipped, but is at 2.0 already. Release date is indeed feature, and I almost never use software that fail to include it.
Thank you for your reply @Reed. My own posting was in part a reaction to the uncritical comments from Jeff, but it does also reflect my views of Xanadu as being both overhyped and not reflective of how the world actually works. I have quite admittedly not done much study of Xanadu, but then again I don’t need to fully understand all of Greek mythology to make the statement that the Greek gods living on Mt. Olympus don’t actually exist.
Regarding the counterpoints you mentioned, in order. 1) the best identification service we have for the web is SSL, but I can tell you the number of times where there’s been some SSL error message that I’ve ignored. (With ssh I’m a little more suspicious, but I only ssh to a handful of different machines.) There is a large problem with phishing and other means of fraud on the internet based on not being able to identify the server machine. My imagination is not good enough to solve this problem, so I doubt I’m imaginative enough to solve the problem for Xanadu. Well, other than through draconian methods which often start with “create a new internet but this time …”
With #3 I’ll repeat what I wrote - “I’m not saying what we have is good, I’m saying this rule does not reflect reality.” The idea that every user is uniquely and securely identified does not work. Period. If I borrow someone’s cell phone to browse the web, how can anyone know? Replace “is uniquely” with “can be uniquely” then it’s a bit more attainable.
With #9, I ask how if someone chooses to have an embedded royalty mechanism then how can that even work? That is, I recognize that not all documents will embed that scheme. But if enabled, the first requirement is solving the micropayment problem, which hasn’t been done yet after some decades of trying. I note that no such scheme can prevent some forms of copyright infringement, such as public performance.
I also listed two cases - fair use and copyright term extension - which would be hard to incorporate into a document. You responded “it does rely on the publisher following the law, and allowing fair use” but it’s not a question of the publisher being fair or not. How does any piece of software confirm that a given request falls within fair use? Even now there are court case to handle ambiguities, and that’s after centuries of experience, assuming Gyles v Wilcox (1740) counts as the start. How do you expect someone in Canada, which has a more restricted “fair dealings” concept to include the broader US “fair use” laws if the document is read in the US? And if some time later the US extends its copyright terms by another 10 years, how do you update all of those electronic royalty statements accordingly?
As for #10, I don’t know how documents are defined in Xanadu. I pointed out some concepts that I think would be hard to incorporate in what I’ve heard about Xanadu, like including something which displays the current date, or embedded advertising which is localized based on the language of the incoming reader. At the very least, how are all those documents versioned, and who pays for all the versioning of every timestamp every displayed on a page?
You mentioned that Nelson is a scholar, but I don’t know why that knowledge is relevant. He proposes something that seems quite unworkable. I’ll grant that he started in the days where among other things we thought True AI would be here within a decade, the vast experience since then, including in more controlled spaces like Minitel, gives plenty of evidence against his ideas.
Thanx
Hi
Hello there
Hello there
Hello there