Re: [Bug-gnupedia] Design proposal -> Wiki

[Bryce Harrington]

On Mon, 22 Jan 2001, Jimmy Wales wrote:
> > 1.  Switch to TWiki.
> > 
> > Presently, wikipedia appears to be running the original Wiki, which is
> > fine, but is deficient of some rather handy features that we'll want,
> > such as accounts, revision control, allowing of formats other than
> > WikiText (e.g., XML or HTML), automatic indexes, etc.  See www.twiki.org
> > for more info.
> 
> Actually, we are running usemod's wiki code, which has accounts (but? do they
> work properly?  not in my opinion), revision control (which seems to work just
> fine), and I'm not sure about automatic indexes, etc.
> 
> I'm happy to change to any codebase that looks better, and I am also thinking
> that there are so many different wiki codebases that we might just make our 
> own
> variant.  (But, I'm loathe to fork for no good reason. :-))

*Nod*  There are _tons_ of wiki varients out there.  We use KeheiWiki at
WorldForge, which is a good high-performance, full featured,
security-focused varient.  At FreeBooks we chose TWiki because it
was well documented and the features were well "exposed".  E.g., it has
quick links to view several previous revs of the page.  

I think for digital signatures to work at all, at least some form of
accounting should be active, obviously.  None of the wiki's I've played
with have had DS in them, so unless there is already a good one out
there for us, this could justify creation of a fork.  
 
> The nice thing about having revision history and no true delete function is
> that it creates some resistance to maliciousness, without having to get all
> "permission-y".  But yeah, if somehow restricted so that casual jerkoffs can't
> just delete everything, it could be a good thing.

*Nod*  Sticky issue.  On WorldForge this has been handy when we wish to
heavily move around and reorganize.  
 
> I have had a lot of ideas (vapor-ware alert!) of creating a "karma points" 
> system
> to enable frequent contributors to somehow "spend points" to protect their own
> articles from random edits by random strangers.  This only becomes necessary, 
> I
> think, after you grow large and popular enough to lose a positive sense of 
> wiki-community.

Maybe, although I once read an article that RMS put into the emacs
distro about point systems - in fact there is a whole book on this topic
called "Punished by Rewards".  Scoring systems sometimes have the effect
of turning fun into work.  

(This reminds me of a cute anecdote I read somewhere...  Old man notices
that the kids in the neighborhood are spraying paint grafitti on the
shops and homes.  He goes out and gives each kid a dollar per wall they
deface and tells them to come back tomorrow and he'll pay again.  They
figure he's crazy but accept.  Next day a *bunch show up* and spray
grafitti everywhere; but this time he gives only 90 cents per wall
("Supply and demand").  The next day 80 cents, then 70, 60, 50, 40, 30,
20.  When it gets to 10, only one kid shows up.  "Yeah, the others
figured it wasn't worth it anymore."  So the old man says, "Huh, okay,
well how would you like to help me paint the stores clean?"  "How much
will you pay for that?"  "Nothing, I just figured it'd be fun.  You can
stop whenever you want."  "Hmm, okay, I like painting.")

> I think it is very impressive how many wikis do so well, without a lot
> of spamming or maliciousness.  But surely there is a scalability
> limit.  I don't think CNN.com should convert to a wiki.  :-)

The original wiki (the Portland Pattern Repository) is *huge* and
has spawned several books.  The problem is less scalability and more
organizational.  They are only as good as the people who put work into
them.  

There's a bunch of fundamental reasons why spamming and maliciousness
occur only very infrequently; it's like scrawling profanity on a
whiteboard - what's the point, when someone can just come along later
and wipe it off?  With a revision control system in place and easily
accessable by users, it becomes pointless even to deface someone else's
work.  And with accounting, doing something bad to the site merely hurts
everyone's opinion of you.  

But yeah, CNN probably should stick with whatever high end software
they've got.  ;-)

Bryce