IF Knowledge Base
David Fisher
something similar to the IFDB. This would be a collection of articles on
"how to do task X" in one or more different IF languages (or for general
topics, how to approach it in any language). Anyone would be able to
contribute a new article, and a classification scheme would enable people to
find topics easily. (The scheme would be editable as well).
(Without such a structure, it would be similar to what we already have --
searching through past RAIF posts using Google. You might find something,
but it's hard to know which keywords to use, you could easily miss something
important, and it isn't easy to compare the same solution in different IF
languages).
Anyway ... Mike Roberts created the IFDB, so I asked for his opinion, and he
had some additional ideas. I'm sure he won't mind me quoting them:
> Some random brainstorming ideas that come to mind:
>
> - users should be able to post questions that they're
> grappling with but don't know the answer to
>
> - other users should be able to see the list of open
> questions
>
> - maybe have a way of specifying your own areas of
> expertise, so that the system can show you open
> questions that you'd be most likely to be able to answer
>
> - let users rate article helpfulness
>
> - use helpfulness ratings as part of a reputation
> system, and also to re-open questions that were
> answered less than helpfully (that is, if an answer
> gets a low enough rating, the system puts the question
> back in the "open questions" list to encourage someone
> else to give it a shot)
>
> - maybe rather than a Wiki style where there's one
> overall article per topic/question, answers could
> be submitted on a per-user basis, analogously to
> reviews on IFDB; that way a given question could
> have several answers from different people, in case
> there are multiple ways to approach the problem or
> differing opinions on the right way
>
> - let users ask follow-up questions, maybe in a
> web-forum-style message list that can be attached
> to each answer
The above idea is quite different to what I was orginally imagining -- I was
mainly thinking of a kind of archive which would gradually be added to,
whose main purpose was to be a place where it is easy to find answers and
example code for various common implementation issues. (Lots of things have
been discussed on RAIF already, and it's a pity if they just get buried in
the mists of time ...)
Mike was suggesting a more active, forum-like thingy, where specific
questions could be asked as well.
So ... thoughts?
David Fisher
George Oliver
<davidfis...@australiaonline.net.au> wrote:
> In the 'RAIF "How to" topics' thread, the idea of an IFKB was mentioned --
> something similar to the IFDB.
>
> So ... thoughts?
>
> David Fisher
What I like about Mike's ideas is that they suggest a call and
response framework for adding to the database, which leads me to think
it would be more active than something with an archival function.
It reminds me of something unrelated in subject matter but somewhat
similar in framework:
Blank
I think that some of the ideas are good, especially the ratings idea,
but I think the post-a-question function is largely covered by RAIF
already. I'm not sure that duplicating that will be helpful - won't it
just mean two places to look with largely the same questions being asked?
I've heard that in some other ngs when a poster has asked a question and
got a lot of feedback it's considered courteous to post a summary with a
carefully worded subject header as a kind of thank-you. So I was
thinking about the new site as containing those sort of posts: "RAIF
Gems" really, I guess.
The thing I like about a wiki is that it's fairly low-maintenance in
terms of admin. I know we're lucky in having people like the archive and
'terp maintainers who've stuck around for a long time, but equally there
have been a lot of projects that died because the maintainer lost
interest, moved on to other things, or just got ambushed by RL.
just my 2c,
jayzee
Mike Roberts
> David Fisher wrote:
>> In the 'RAIF "How to" topics' thread, the idea of an IFKB was
>> mentioned -- something similar to the IFDB. This would be a collection
>> of articles on "how to do task X" in one or more different IF languages
>> (or for general topics, how to approach it in any language). Anyone would
>> be able to contribute a new article, and a classification scheme would
>> enable people to find topics easily. (The scheme would be editable as
>> well).
>>
>> he had some additional ideas. I'm sure he won't mind me quoting them:
>>
>>> grappling with but don't know the answer to
>
> I think the post-a-question function is largely covered by RAIF already.
> I'm not sure that duplicating that will be helpful - won't it just mean
> two places to look with largely the same questions being asked?
My thinking was that it would be a good way to build out the database
organically, by letting people who have open questions that aren't already
answered in the database add them to the database, and by providing a way to
answer questions that directly captures the answers into the database.
But you're right - it would tend to fragment the discussion, which is
something I wasn't thinking about when I rattled off these ideas. It might
result in questions being duplicated in raif and IFKB, but I think it's more
likely to fragment the discussion between the two, which probably wouldn't
be good in terms of maintaining a critical mass in the newsgroup.
> I've heard that in some other ngs when a poster has asked a question and
> got a lot of feedback it's considered courteous to post a summary with a
> carefully worded subject header as a kind of thank-you. So I was thinking
> about the new site as containing those sort of posts: "RAIF Gems" really,
> I guess.
I think this is much more along the lines of what David had in mind - I
think his idea was that people would go and create IFKB articles to
summarize newsgroup how-to threads.
The trade-off is that there's not much incentive for the original
participants in the thread to do that - the questioner got his question
answered and will probably just go implement the solution, and the answerer
might not want to bother with entering the same information again into a
different system. So I think the database would largely end up being built
by volunteers who summarize raif threads into database articles.
> The thing I like about a wiki is that it's fairly low-maintenance in terms
> of admin. I know we're lucky in having people like the archive and 'terp
> maintainers who've stuck around for a long time, but equally there have
> been a lot of projects that died because the maintainer lost interest,
> moved on to other things, or just got ambushed by RL.
That's exactly where I was coming from with the Q&A format idea.
Initiatives like this are more likely to stay alive if the whole community
contributes to them rather than a smaller group of volunteers. With a Wiki,
the whole community *can* contribute; the question is whether most people
would. With a Q&A format, there's a natural incentive to add questions to
the database - it's a way to get them answered. And there's some incentive
to add answers, in the form of reputation. But maybe there's a way to
accomplish the same thing without the fragmentation.
--Mike Roberts
mjr underscore at hotmail dot com
ChicagoDave
> With a Wiki, the whole community *can* contribute; the question is whether most people would.
The IFWiki.Org website seems to be hanging in there. Of course without
David Welbourn, it wouldn't be anything resembling a maintained
website and if he gets hit by a bus, it would likely die. That said, I
think the idea of a database and logically functioning knowledge-base
is a great idea. It's something I tried to accomplish many moons ago
with the IFLibrary.Org website.
The primary concern of most community members seems to be the
fracturing of usenet conversations or even worse, moving a
"conversation" off of usenet entirely. I don't see any way around
this. Someone would have to identify the question and answer and enter
them into the IFKB. This really is no different than the way the
IFWiki is used. If someone asked the question on the IFDB however, we
_could_ automagically post the question and answer to usenet. The
answerer could even select which newsgroups to send it to and
rec.arts.int-fiction would always be a default selection.
Another issue I see is that some questions are platform specific and
some aren't. So we now get into a "Cloak of Darkness" scenario where
the "answer" really could have several implementations. I think we'd
want to capture the generic IF questions and allow multi-platform
answers.
Now I'm wondering if there's some way to educate usenet users to tag
questions in a certain manner, tag answers in a certain manner, and
let a spider extract the data into the IFKB. No, too many variables.
I like the idea of an IFKB, but I'm sure there will be strong
disapproval from others concerned over fracturing the community
surrounding rec.*.int-fiction.
David C.
David Fisher
news:5319cf2d-dbd4-411b...@d4g2000prg.googlegroups.com...
>
> Another issue I see is that some questions are platform specific and
> some aren't. So we now get into a "Cloak of Darkness" scenario where
> the "answer" really could have several implementations. I think we'd
> want to capture the generic IF questions and allow multi-platform
> answers.
Language-specific articles would be tagged as such; if you
wanted to see the solution to a problem in all available
languages, you just wouldn't supply a "language" tag when
searching, and it would return a list (one for each known
language, or just a single article if the topic isn't
language specific). A single article could include
multiple languages if that seemed appropriate as well
(and would be tagged with all of them).
> Now I'm wondering if there's some way to educate usenet users to tag
> questions in a certain manner, tag answers in a certain manner, and
> let a spider extract the data into the IFKB. No, too many variables.
Interesting you should mention that ...
Mike came up with a similar idea:
> I wonder if there's an approach that would integrate the system
> with raif somehow. I don't have any specific ideas here - I'm
> just brainstorming. It seems to me that the blue-sky ideal
> goals would be (1) to keep raif as the primary discussion
> forum, and (2) make the discussion flow naturally into the
> IFKB database without a lot of volunteer work. Maybe IFKB
> would be an indexing system for raif threads, rather than a
> separate article database; maybe you could tag a thread for
> inclusion in the IFKB index by including some special keyword
> syntax within the article itself - like put a line at the end
> of the article under your signature with something like "<IFKB:
> Inform 7, Dark rooms, scope>" to specify the relevant topic
> areas; IFKB would act as a news client, pulling in articles
> from the wire and scanning them for the special <IFKB:> syntax,
> adding them into the database when it finds them.
... which sounds pretty good. If it wouldn't make it too
complex, there could be an additional tag <IFKB prev> or
something which someone could add in a follow-up to a post
that didn't include an <IFKB> tag itself.
Then again, maybe an <IFKB prev> tag isn't necessary, since
if someone has noticed that a particular post should go in the
IFKB they could just add it themselves immediately.
David Fisher
Juhana Leinonen
>I think that some of the ideas are good, especially the ratings idea,
>but I think the post-a-question function is largely covered by RAIF
>already. I'm not sure that duplicating that will be helpful - won't it
>just mean two places to look with largely the same questions being
>asked?
(note that I'm assuming here that IFKB would be "ask&answer" like Mike
Roberts suggested, not "discussion archive" that was the initial idea)
I believe this would not be the case but the people who ask the
questions would far more likely use either the new web site or RAIF, but
not both. The community is too small to maintain momentum in two places
so either IFKB would never gain enough audience and reach the critical
mass to be successful, or coding questions in RAIF would diminish to
nearly nothing because everyone used IFKB. (By this I don't mean that
other discussion would die out.)
Another thing is that Usenet is almost useless when it comes to storing
and searching old questions and answers. You just can't do it
efficiently. If you need an answer to a problem, you can either try
searching the group with Google (good luck with that), remember someone
asking about something similar in the past (you would have to be a
regular and have an elephant's memory) or post a new question, even if
it has been asked dozens of times before. There's no classification
system at all, no tags, and the only option to browse the old questions
is chronologically by posting date.
This could be a good time to discuss the identity of RAIF. Let's look at
the current recent threads.
* Little question about IFComp's schedule
* I7: Docs
* i7: i need an idea
* A random thought about NPCs
* Inform 7: Lock and Key Anomaly
* DreamPath composer 1.0.0.1 & Allocating features
* Displaying Thai in I7:5J39
* [I7] Carry out asking (using tables) build: 5J39
* I7 - problems saving/restoring with "Flexible Windows"
* Glulxe 0.4.4
* Inform 6 (or 7) custom text colors?
* first person narrative
(+ this thread, of course, which you may or may not count in)
Out of those 12 threads, 7 are questions about authoring in I7. Then
there are 2 announcements, one other (IFComp) question and 2
_discussion_ threads. By discussion I mean a real debate about the craft
and theory of IF. I don't consider language specific question-answer
-threads discussion.
So, what is the purpose of RAIF? Is it supposed to be and do we want it
to be a unofficial helpdesk for I7, with some theoretical discussion
thrown in from time to time? If the helpdesk questions would migrate to
a separate web site, would it be a good or a bad thing to the newsgroup?
And most importantly: are the I7 questions keeping the newsgroup alive
by bringing new people in? If there were no more of them, would the
group wither and die, or would the theoretical debate get more room to
breathe and bloom?
It should be noted that shortly after IFDB was launched, SPAG announced
that it has to change its direction because IFDB has replaced some of
its functions. The same might happen with IFKB/RAIF, and it might not be
a bad thing at all.
Juhana
--
Spamblock: remove all numbers from the e-mail address.
David Fisher
news:Wyrmj.462$tQ1...@read4.inet.fi...
>
> So, what is the purpose of RAIF? Is it supposed to be and
> do we want it to be a unofficial helpdesk for I7, with some
> theoretical discussion thrown in from time to time? If the
> helpdesk questions would migrate to a separate web site,
> would it be a good or a bad thing to the newsgroup?
Well, I think rec.arts.int-fiction just "is what it is" ...
people will use it the way they want to, and nobody can
(or should try to) enforce anything. Clearly lots of people
like having a newsgroup format, too. (But
http://www.intfiction.org/forum is also available for
people who like that style).
An advantage of posting questions to a newsgroup is that
all kinds of other interesting things get brought up as
well (which is a pain for archiving, but that's not the
point!) It would be terrible to lose that.
RAIF is a cool community to be part of, and for myself,
I'm pretty happy with it as it is ... it would just be
good to have a more searchable archive of answers to
problems, thus the idea for an IFKB.
(That said, it can still be good to question the status quo ...)
David Fisher
James Jolley
<davidfis...@australiaonline.net.au> wrote:
> An advantage of posting questions to a newsgroup is that
> all kinds of other interesting things get brought up as
> well (which is a pain for archiving, but that's not the
> point!) It would be terrible to lose that.
Agreed. RAIF has it's own culture so to speak and it would be a shame
to mess with that.
> RAIF is a cool community to be part of, and for myself,
> I'm pretty happy with it as it is ... it would just be
> good to have a more searchable archive of answers to
> problems, thus the idea for an IFKB.
Yes, IFKB as an idea sounds really useful. It would be rather
interesting to see how one would go about archiving all those older
posts though.
Best
-James-
Juhana Leinonen
>> do we want it to be a unofficial helpdesk for I7, with some
>> theoretical discussion thrown in from time to time? If the
>> helpdesk questions would migrate to a separate web site,
>> would it be a good or a bad thing to the newsgroup?
>
> Well, I think rec.arts.int-fiction just "is what it is" ...
> people will use it the way they want to, and nobody can
> (or should try to) enforce anything. Clearly lots of people
> like having a newsgroup format, too. (But
> http://www.intfiction.org/forum is also available for
> people who like that style).
I agree 100% and I'm not trying to enforce any changes; all I'm saying
is that if a website appears that could potentially mean 2/3 of all RAIF
messages would be posted there instead of here, it _will_ have an impact
on the newsgroup. I don't know if it's realistic or not, but I would
imagine that newbies prefer a web site to the Usenet nowadays if there
is an option. (I don't know how new people find RAIF but I would imagine
it's because googling reveals most questions are asked here, so they
come here to ask because they don't know any other place.)
(Then again, intfiction.org forum has not catched on, but that's
probably because it has the exact same purpose than the newsgroup but
not enough mass to keep it going.)
tayr...@yahoo.com
> is that if a website appears that could potentially mean 2/3 of all RAIF
> messages would be posted there instead of here, it _will_ have an impact
> on the newsgroup. I don't know if it's realistic or not, but I would
> imagine that newbies prefer a web site to the Usenet nowadays if there
> is an option. (I don't know how new people find RAIF but I would imagine
> it's because googling reveals most questions are asked here, so they
> come here to ask because they don't know any other place.)
Have you guys tried google groups? That's actually how I've been
viewing RAIF. It displays the newsgroup postings like a messageboard.
Take a look: http://groups.google.com/group/rec.arts.int-fiction/topics
James Jolley
> Have you guys tried google groups? That's actually how I've been
> viewing RAIF. It displays the newsgroup postings like a messageboard.
> Take a look: http://groups.google.com/group/rec.arts.int-fiction/topics
Yes, I am using google groups for my access now as I refuse to use
windows just for a newsgroup client that is accessible for the blind.
miket...@embarqmail.com
>(snip)
> Have you guys tried google groups? That's actually how I've been
> viewing RAIF. It displays the newsgroup postings like a messageboard.
> Take a look: http://groups.google.com/group/rec.arts.int-fiction/topics
While I do not use a newsreader to view raif, I probably would if my
isp provided usenet access. My understanding is that it addition to
allowing for better organization of threads and messages, newsreaders
have nice filtering options. I suspect that many people here are glad
they never have to read about MI5, ten - year - old flame wars, or
the occasional lapse in a poster's grammar. Google will also eat
posts sometimes, although I don't know how it compares with actual
usenet providers. The ability to search is nice, but as David and
others have pointed out, can be hit or miss.
Skinny Mike (not Roberts)
George Oliver
<davidfis...@australiaonline.net.au> wrote:
> "ChicagoDave" <david.cornel...@gmail.com> wrote in message
Yeah, there would have to be some way for users to tag posts
retroactively -- but wouldn't they just have to tag their answer for
the bot to pick up the thread? I think ideally the person who answers
shouldn't have to -add- the thread manually, though people could
certainly tag and re-organize threads by going to the IFKB itself.
It seems a little self-defeating to set up an elaborate categorization
scheme and try to educate people to apply it -- better to let people
tag it with whatever they want and let the most popular tags rise to
the top, right? The community would self-educated eventually.
I really like this idea of a IFKB bot.
David Fisher
news:8b3fa3ce-45c1-4896...@e4g2000hsg.googlegroups.com...
> On Jan 24, 1:29 pm, "David Fisher"
> <davidfis...@australiaonline.net.au> wrote:
>> If it wouldn't make it too
>> complex, there could be an additional tag <IFKB prev> or
>> something which someone could add in a follow-up to a post
>> that didn't include an <IFKB> tag itself.
> Yeah, there would have to be some way for users to tag posts
> retroactively -- but wouldn't they just have to tag their answer for
> the bot to pick up the thread?
I was mainly thinking of the situation where someone posts
a great answer to something without an <IFKB> tag, so
someone else could post a follow up containing an <IFKB-prev>
tag ... but then I realised that they could easily just add
it directly to the IFKB themselves instead of cluttering the
newsgroup with <IFKB-prev> tags.
> It seems a little self-defeating to set up an elaborate categorization
> scheme and try to educate people to apply it -- better to let people
> tag it with whatever they want and let the most popular tags rise to
> the top, right? The community would self-educated eventually.
This works with something like IFDB, because there are a limited
number of tags that apply and they are not all that specific
("single room game", "", etc.), but it wouldn't work so well
with a collection of "how to" topics which get very specific ...
The closest equivalent I can think of is Past RAIF Topics
on the IF Wiki, which contains 127 different topics in the
first two levels, most of which could be divided into quite
a few more ("General IF Theory" would have at least 30):
http://www.ifwiki.org/index.php/Past_raif_topics
To put it another way, many of the IFDB tags are "orthogonal"
and could apply to many different games throughout any kind
of hierarchy that you could set up, but the kinds of tags that
would apply to the IFKB would mostly be very specific --
things like "implicit actions", "liquid mixing", "random NPC
movement" and "adding an adjacent room to the current scope".
There would still be some orthogonal tags as well
though, such as the language tag (I7, TADS 3, etc.)
> I really like this idea of a IFKB bot.
Me, too. Yay Mike Roberts and Dave Cornelson ...
David Fisher
George Oliver
<davidfis...@australiaonline.net.au> wrote:
>
> To put it another way, many of the IFDB tags are "orthogonal"
> and could apply to many different games throughout any kind
> of hierarchy that you could set up, but the kinds of tags that
> would apply to the IFKB would mostly be very specific --
> things like "implicit actions", "liquid mixing", "random NPC
> movement" and "adding an adjacent room to the current scope".
It seems like an IFKB would benefit then from some kind of tag manager
interface...where you could see all the tags, find out what they apply
to, and meta-sort them into something that makes sense. So if the
canonical tag is 'liquid mixing' and someone tags 'potion mixing' you
could revert that tag (or I guess easily see that you could re-tag
with liquid mixing).
Is it better to keep -all- tags or let people delete tags?
David Fisher
>
> It seems like an IFKB would benefit then from some kind of tag manager
> interface...where you could see all the tags, find out what they apply
> to, and meta-sort them into something that makes sense. So if the
> canonical tag is 'liquid mixing' and someone tags 'potion mixing' you
> could revert that tag (or I guess easily see that you could re-tag
> with liquid mixing).
>
> Is it better to keep -all- tags or let people delete tags?
It would be good for the tags themselves to be searchable, so I think
keeping all tags is a good idea (but to still be able to correct errors if
necessary). In thesauruses and things like that, there is often a "main
entry" with a canonical term that other terms redirect to, so something like
that might be useful.
(I'm working on an example structure so there is something concrete to look
at and discuss, but it might take a little while to get into a reasonable
state ...)
Another issue is the scope of the IFKB. I have been focusing on just "how
to" topics so far, since there is already a Past RAIF Topics page for
theory-type things, but potentially it could include both; as someone
mentioned, it can still be a bit of work to wade through a 200-post thread
to find the topic-relevant parts.
And just to reassure people who are nervous about the amount of work
involved in maintaining the IFKB ... nobody is obligated to do anything at
all for this. In particular, experts on Inform 7 / TADS 3 will not be
drafted to fill in the relevant topics! I'm still thinking of it as an
archival system, where people can look up answers to common (or obscure) IF
problems. Whatever happens to end up in there will be useful, and if some
things never make it in, that's OK too, because something is better than
nothing ... we'll just see what happens.
Another thought -- the system I am most familiar with at the moment is
Inform 6, and the I6 documentation is fairly fixed (the DM4 isn't about to
change). As well as browsing through recent posts on RAIF, I'm thinking of
going through the DM4 to create a list of basic topics and to help design
the initial classification scheme; each topic will probably just be a link
to the online version of the DM4 (which is very convenient to link to
individual sections), rather than a copy of the actual text (which would
mean that the text isn't searchable from inside the IFKB, but I think that's
OK for a start. It's more important for the topic names themselves to be
searchable). I'm not familiar enough with TADS 3 to try this with its
documentation, and Inform 7 is still in a bit of a state of flux (and
doesn't have reliable links to the online manual).
Thoughts?
David Fisher
Blank
For me, the show-stopper with forums is *no bloody kill-files*. We have
had fun with trolls here in the past [nobody say his name, nobody say
his name] and although irritating, it's nothing like the havoc they can
wreak in a forum.
--jz
Blank
> On Jan 26, 7:00 am, tayruh...@yahoo.com wrote:
>> (snip)
>> Have you guys tried google groups? That's actually how I've been
>> viewing RAIF. It displays the newsgroup postings like a messageboard.
>> Take a look: http://groups.google.com/group/rec.arts.int-fiction/topics
>
> While I do not use a newsreader to view raif, I probably would if my
> isp provided usenet access.
That's odd. All of the ISPs I've ever used (and I've used some pretty
cheap dial-ups in my time) provided usenet access to text groups. They
didn't advertise it though, except right down in the tiny print. The
format is usually news.[name of the isp] so verycheap.dialup.com would
be news.verycheap.dialup.com - have you tried just having a go and
seeing if it works?
I'm using Thunderbird as my newsreader, which is free and gets the job
done, but isn't a patch on Agent - which costs about the price of a
restaurant pizza.
hth
Jayzee
George Oliver
> Juhana Leinonen wrote:
> >
> > (Then again, intfiction.org forum has not catched on, but that's
> > probably because it has the exact same purpose than the newsgroup but
> > not enough mass to keep it going.)
>
> > Juhana
>
> For me, the show-stopper with forums is *no bloody kill-files*. We have
> had fun with trolls here in the past [nobody say his name, nobody say
> his name] and although irritating, it's nothing like the havoc they can
> wreak in a forum.
>
> --jz
I don't know, it depends on the forum you use. SMF for example allows
kill-files (kill list maybe is the specific term for that forum).
Anyway, I think Juhana hit the nail on the head regarding Intfiction
-- no critical mass.