More Sophisticated Referencing

OK... I'm (still) trying to play with my Kwiki [http://havoc.gtf.gatech.edu/cgi-bin/austin/kwiki], and this is my next big thing.

Basically, I want to somehow "quote" text from some other URL, and have it included. I'm not sure of the scheme, although a few come to mind.

It would be really cool if I could just hit some magic "quote" button, then give it a URL, and a little browser would pop up with the URL. Then I could select the thing I wanted to quote (be it a sentence, paragraph, whatever) and the tool would magically "figure out" what I meant, and represent it.

Later, when browsing a page with the quote, it (the quote) would be retrieved from the source, and (maybe) the source would be informed that it had been quoted. And, naturally, there would be a nice link to the original.

When is this useful? When you're quoting a small part of a larger dynamic page, perhaps... like most things in wiki.

Given that we only have a stupid edit-box, HTML, and perl, how could we do it?

Maybe something like

 |quote http://c2.com/cgi/wiki?CommentOnTransclusion from "I had envisioned something" to "-- AustinDavid"
which would render something like:


I had envisioned something more hyper-texty, more groovy and powerful. Something that would let me quote some exotic page (Wiki, HTML, who cares?) here without transcluding the whole thing... it would also be nice if the quoted text (like some paragraph in a news article somewhere) could somehow magically "know" that it had been quoted. Maybe this is MoreSophisticatedReferencing? that ought to be discussed elsewhere...

-- AustinDavid


Then the quoting-text doesn't duplicate the info. Better, the quoted text can change (within the given bounds) and it won't break the ref. Given intelligent servers on both ends, we could go FURTHER and the page CommentOnTransclusion would know about the quote, and could even track it, so if someone later were to "break" the quote bounds (by removing "AustinDavid" or something) the editor could inform the author, who could either reassign or invalidate the quote.

The above text contains an embedded hyperlink - it could be slurped from the original, of course. And interesting regexp, but definitely possible.

Assuming stupid pages (eg, the WorldWideWeb) we could stick to the address/index/offset scheme... there are ways to get around URLs and name Kwiki-spaces more directly, so that could provide the local knowledge of an intelligent remote server (the one with the quoted text).

Hmmm..... what's that Xanadu thing? -- AustinDavid

http://www.xanadu.net/


Yeah, I wanted transclusion for (a later version of) SpikeWiki. SpikeWiki allows arbitrary words to be made into links by preceding them with '?' (which is also how SpikeWiki displays links to non-existent pages), and I was going to use '!' for a transcluded link in the same way. Currently I think the system should be able to transclude any link without the user having to use a special symbol. This is part of a general scheme where the system monitors the user's interests and adds links that the user might like to follow, as well as inlining those the user will probably want to follow.

But this is far from TheSimplestThing. -- DaveHarris


I'm not sure TheSimplestThing is even an option anymore - I think "lowest common denominator" (eg, WikiWiki-compatible) is about as close as we'll get.

What do you mean about the ability "to transclude any link without the user having to use a special symbol?" Do you mean that the "system" should monitor usage (easy, given that it all runs through the CGI), then supply another section (perhaps) including a handful of things "of interest?" What about the "inlining" thing you mention?

My goal with Kwiki [http://havoc.gtf.gatech.edu/cgi-bin/austin/kwiki] is to make as much as possible very simple, or natural (whatever that means). Harder things can be more difficult to do, but I've totally preserved all wiki-ness, because I like it.

My inclusion aims to give the user a simple way to quote a page (probably a wiki- or kwiki-page; hopefully any URL-specified page in the future) and faithfully represent it "inline" with the text, leaving the user a simple syntax (the above was pulled off the top of my head). Also, duplicating info for the sake of a quotation isn't very cool - the idea is that you provide both a link and some context to make the link meaningful. That's the idea in my head, anyway. -- AustinDavid

(PS: Dave, the only thing on Kwiki that cares about a login is the MetaPage?)


Yes, I mean the system monitors usage. It generates customized pages on the fly, to maximize the value for the individual reader. This may mean inlining the full text of a link or just a summary, or it may mean hiding stuff you've already read. "Hiding" is like reverse transclusion - it leaves a link and/or a 1-line summary. So transclusion is at the heart. -- DaveHarris


Interesting... what you really want, then, is some sort of browser on a SeaOfContent. We started getting into this in one of my software engineering classes back in skool, by lack of initiative/resources/intelligence kinda crushed the project. It was over the heads of undergrads, and probably longer-term than grads. But still a beautiful idea. -- AustinDavid


EditText of this page (last edited April 13, 2006) or FindPage with title or text search