A graph of related claims would be a nice feature.

A tree view would be nice. So, you can see what inspired it and back a little and then how it branches and what it inspired. And then you could click on different areas to either see more detail or move around to see further up/down the tree or more to one side (if the claim bushes out enough creating enough conversation and more claims).

i woudl actually like tree views based on other relationships, as well

Have you looked at the argument mapping stuff OJC was talking about? It should be possible to automatically generate argument type maps from jyte claims.

To make it even better, we'd need 'make a related claim', 'make a claim that depends on this one', 'make a claim that this claim depends on'.

but IMO that dependency is data, not meta data, and thus should be voted on.

Yup, you're right.

I can make a graph if I get a
data file, and place to upload
the resulting files.

Svend, check out DivShare. I haven't used it personally, so no endorsement is implied.

The only problem with doing this is that it just wouldn't look very good with the current data set. Most claims are stand-alone, and the most I remember seeing in a related group is four or five. Also, if you want them to be labeled, we'd either have to use a unique claim number (which is buried in the page source) or some or all of the claim text for each claim. So basically you'd have a very large number of dots with really long labels on them, making it impractical for printing or viewing in a 2-D system. (Since the inspired claim relationship is one-way and singly-linked, I don't see how a tree would work currently.)

There are 3-D-ish solutions, but these tend to require platform-dependent software. Many Eyes and Swivel have potential, but aren't really there yet in terms of the volume we'd need. Last I heard, Vynce couldn't see the ManyEyes applet on his Mac, although it's supposed to work under Safari (perhaps somebody's using non-standard or newly-implemented Java code, but the service is still in alpha, so we'll have to wait and see.)

What would be cool would be several layers of relationships among claims that could be turned on and off as desired, including arrows for directional relationships, or even surfaces mapped to related sets of claims.

The spatial layout of claims could be based on any user's or group's perspective. Claims that are closer to the user in opinion space would be nearer the origin, colored hotter, or bigger in size (like a dot radius). You'd want to throw out all the claims the user (or a quorum of the group) hadn't voted on, and make a million other tweaks, but it could be useful.

I'm fairly sure there have been a few longer trees. Plus, tree is more the right word, because sometimes we get something like:

X inspires Y and Z, Y inspires A which inspires B and C and Z inspires D... that sort of branching is very interesting to look at, but very hard to follow through inspired links.

However, it hasn't come up that much. But I still think being able to see it when it happens is nice, plus we might notice more of it if it is easier to see.

Oscar, I don't necessarily mean a tree view of the entire site at once. When looking at a partiular claim, having a listing -- perhaps like the directory listings in many modern OS's GUIs -- that shows the heirarchical relationships between this claim and all its related claims would be quite useful, IMO. with ajax and lazy population you could even make it pretty darn fast and light.

similarly, individual graphs of related claim nests, listed and browsable as the islands they are, would be ... well, better than not having it.

I work on an algoritm fore my masters to make
"subset graphs" in leak of better
words.

So if I can get the claim a long with
the tags. I can build one on the tag-structure.

A tab separated datafile will do.
Like
claim1 tag1 tag2 tag3
claim2 tag1 tag2 tag3 tag4
....
But with tags instead of spaces.

I have started to make some
argument maps of some of
the claims on this site.

http://shs.demosophia.net/~shs/arguments/

Svend, can't get to your URL, I get an empty document.

Robin: It crashed Firefox for me the first time, then worked again, after starting a second instance of the java VM. Check your Java settings, if available.

A screenshot from Svend's site for the Java-impaired:

Svend: Neato. I had been thinking about doing some jyte remixing in FreeMind, and I didn't know they had an applet.

Oh Java, that would explain it...

/me turns to gopher for some entertainment.

Thanks for the screenshot and explanation.

yes it is in java.
It is a little plugin that the
people's of FreeMind have made.

I would like some suggestions
to claims that would be good to make argument maps of.

i recommend most of the works of D'Archangel; also many claims by Rachel; and a few of my own (english is phonetic, for example). generally, if you can find a way tog et inspired claims, many of those would work well.

This GreaseMonkey script displays related claims as a hierarchical tree in the claim page.

i'd love some screenshots of that before i go installing foreign code (and reading up on greasemonkey, etc.)

Screen shot:

that's pretty cool, lala. but the javascript is so long... i fear long chunks of JS. my machine is so slow already... i'm really torn. but thanks for the hard work! you tempt me greatly.

Don't whine about it, optimize it.

how much JS cred do you think i have, oscar? 'cause *i* wouldn't give me any...

I don't think it's necessarily rational to fear long chunks more than short ones; 'while(true);' would have an infinite performance penalty, for instance (assuming I translated it into javascript).

yes, but i would be able to read it and i'd be more likely to understand it. beyond that, the knowledge that others will try to read it means that evil code is often somewhat disgused; the disguising process makes the code longer. thus,while short code is jsut as dangerous, long code is more frightening.