I think what Nick is talking about is not so much mining an existing
forum but rather creating a forum whose policies and collective
intentions are to build a document (white paper, publishable paper,
etc.) collectively. I think Nick is frustrated that the existing tools
for such fora are syntactically gunky to the point that someone with his
modest skills at hashing text is insufficient.
As I remember it, Nick engaged a number of us in what he called
"noodling" based on a wiki structure but with unmoderated policies
regarding cross-linking "threads" or more aptly "Noodles"... I
remember it as an interesting exercise, well motivated, and possibly
sufficient at least for the "brainstorming" phase of a project... It
wasn't clear that the tool (wiki) or the policy (ubiquitous
cross-linking of threads-as-noodles) would necessarily take us to that
next level. The failure, as I remember it was a lack of
participation. I don't remember how many of us "threw down" in Nick's
Noodle-Bowl, but I'm afraid it probably wasn't enough. My own
participation was spotty... a brief surge of interest and participation
which died off (naturally) fairly quickly and was perhaps extinguished
by a lack of broader participation. Again, memory is weak, Nick's may
be better, but I think it was at most a half-dozen who "played" and then
it might have been only Nick and I who put much at all into it?
It was an interesting experiment which I think was useful but maybe at
best "a good start".
In my own opinion (and experience), I think that the problem of tech
support fora has already been identified fairly well here:
* most people participate only to "take" from the forum
* many who "give" tend to give anecdotal speculation rather than hard
facts or useful advice
* many who "use" the forum tend toward bashing around a little,
expressing their frustration that the straw/needle ratio is too low,
thereby throwing more straw in the pile.
I specifically have been using the "Honda Insight" enthusiast forum to
try to debug a few problems with my own 2001 Honda Insight
(wonderfully/horribly eccentric 0th generation hybrid vehicle) and have
at least faulted on the first point of offering little of my own
information. This is partly as a consequence of wanting to avoid the
second two problems. By the time I actually have enough evidence and
experience with a problem, I am mostly relieved (problem solved, or at
least understood) that I want to "forget the problems" as quickly as
possible, and trying to write a succinct summary of what I learned is a
major challenge.
When I do contribute, I try to contribute in the form of documenting my
troubleshooting and debugging whilst ubiquitously cross-linking to other
threads which have addressed the problem I'm currently chasing. I do
this mainly (despite the risk of raising straw/needle ratios) because
THAT is where *I* find the most help, when someone chasing a problem,
points at a resource that they either know has (or hopes will) help them
before falling off that horizon of "well, my problem is fixed now, seeya
later and thanks for all of the fish!"...
I agree with Robert who says the ability to talk to a real live
technician in real-time is ideal, up to the fact that too often "help
line" technicians cannot have the kind of experience a working mechanic
(or enthusiast) can have.
Another forum I have used in this manner is an Appliance Repair forum
which has helped me through two major Washer/Dryer repair phases. In
this case, there are several semi-retired appliance repairmen who
apparently enjoy hanging out on the forum, providing expert advice. It
seems to be a win/win/win.... they enjoy this labor of love, the company
that sponsors the site (an appliance parts supply company) and the users
of the forum like me all get something good from it. I made a point
in my own case, to summarize my experience from cradle to grave (from
first symptoms to final resolution) when I was done... it was easier for
me to do this with an Appliance than with a Vehicle or a
Computer/Software problem because for the most part, an washer or drier
is a somewhat simpler electromechanical device with fewer "hidden
variables" and "subsystem dependencies"... things happen in relative
real-time, relatively sequentiall. Timers trigger things in sequence,
an observer can usually hear/feel/see the results (solenoids, motors,
valves, etc.) and the logic is relatively simple (turn on the fan,
energize the heating element, if the current exceeds Z open a breaker,
if the temperature exceeds X, open the heating element circuit, etc.)
This forum (and WedTech) do not really qualify (for me) as a "problem
solving forum" though I do see we use it that way sometimes... but more
in the mode of a "call for ideas"... with a cascade of responses ranging
from flippant to acutely helpful with a modest amount of "idle
speculation" in the middle of that gaussian distribution. This forum
(and WedTech) is almost entirely (IMO) a forum for announcing
interesting things and discussing the implications of them. Of course,
what is interesting to one is often not interesting to another, and I
think we collectively have a fairly good idea of who is interested in
what, but wonderfully an erstwhile lurker or even a regular participant
throws down on a topic unexpectedly with significant utility (i.e.
insight, unique new information, directly useful experience or knowledge).
I have worked on Decision Support Systems and Collaborative Tools which
I think indicates that I have some hope and some experience in this
domain. I think it will be a long while before the holy grail is
achieved of structuring unstructured process in arbitrary ways. One of
my projects in this domain (entitled Faceted Ontologies) *did* attempt
to get a handle on structuring unstructured information artifacts, or
perhaps more to the point, ingesting information artifacts with one type
of structure, normalizing the structure, then allowing individuals to
restructure or extract artifacts and correlations according to a
structure more relevant to them. That problem was uber-hard, we did
not do more than outline an approach in 2 years of work.
All that said, I think it is an interesting (but hard) problem and I
applaud the intentions of folks like Nick who seek to achieve even a
small win in this domain and I applaud those who build little tiny tools
as bridges to begin to approach the foothills of such problems.
- STeve
Nick,
That sounds like a cool "big data" research project. There are
solutions at the command-line level to some of your problem - scripts
and lexical analyzers can extract the "meat" of the medium, but that
just gives you a pile of random cuts of meat. Big data may be able to
recognized different cuts and sort them into a butcher shop display
case (like my local Sprouts or Kellers).
However, as Robert pointed out - the majority of comments are
frequently questions without answers in technical forums. A display
case of nothing but hamburger is not very useful - so there needs to
be something more. Frequently, the most frustrating aspect is that
one has to look at every possible thread to find the one thread where
the correct answer is given.
The question and answer section of Amazon's product pages seems like
a successful model - they simply ask the question of every person
whose ever bought the item in question.
Ray Parks
Consilient Heuristician/IDART Program Manager
V: 505-844-4024 M: 505-238-9359 P: 505-951-6084
NIPR: rcpa...@sandia.gov <mailto:rcpa...@sandia.gov>
SIPR: rcpar...@sandia.doe.sgov.gov
<mailto:rcpar...@sandia.doe.sgov.gov> (send NIPR reminder)
JWICS: dopa...@doe.ic.gov <mailto:dopa...@doe.ic.gov> (send NIPR reminder)
On Jul 17, 2014, at 7:43 AM, Nick Thompson wrote:
Arlo,
As you may remember, I have been frustrated by a possibly related
problem. I have thought or years that academic fora could be used
for the development of publishable text. When I was teaching, I
tried to move undergraduates down what I hoped was a slippery slope
from arguing a point of view in a forum to writing a paper,
collaboratively. Since any convincing presentation of an argument
includes thorough presentations of opposing arguments, your opponents
in an argument can help you generate a lot of your text. Later, when
I came to Santa Fe, I tried to get groups of us to write together by
pouring rich material into an internet forum and then gradually
turning it to drafts of text, and finally finished publishable work.
I have had two successes, but given the amount of rich material that
has been generated, that seems chump change.
For my purposes, the trouble is the FORM of internet correspondence.
It is terribly difficult to get an internet exchange into editable
form, partly because of all the headers, partly because of the
quotation, partly because of the order, and partly because it's hard
to get it into one file. Now, I am sure you will say (because that's
what I kept saying), "Oh, you old dummy, that's easy!". And if it is
easy for you, please tell me how. I have spent weeks at it, and
failed every time.
What I think I need is a forum that is designed to produce one
continuous, temporally ordered document. Now, the correspondence
utility in Research Gate seems to do that. But its drawback is that
only two people can correspond at a time. So, if you -- or anybody
else -- know of a forum utility that produces a continuous,
temporally ordered document for many authors, please let me know.
Nick
Nicholas S. Thompson
Emeritus Professor of Psychology and Biology
Clark University
http://home.earthlink.net/~nickthompson/naturaldesigns/
<http://home.earthlink.net/%7Enickthompson/naturaldesigns/>
*From:*Friam [mailto:friam-boun...@redfish.com]*On Behalf Of*Arlo Barnes
*Sent:*Thursday, July 17, 2014 1:23 AM
*To:*The Friday Morning Applied Complexity Coffee Group
*Subject:*Re: [FRIAM] Smart Forums
Well, I generally think of the improvement of forum interactions as a
community phenomenon, eased or impeded by the structure of
communication and interaction with the site infrastructure*, rather
than as a computational dilemma/per se/; but I would be interested
to hear what ways you think fora could be improved by AI.
A general forum for thinking about internet fora (or forums,
depending on your preference) ishere
<http://www.reddit.com/r/TheoryOfInternet/top/>.
-Arlo James Barnes
*For software support, it is that interaction with the site is very
limited - mostly, people are only there for the duration of the time
they need assistance, and during that time are more interested in
getting help than giving help. There is no time for the structure,
mores, and history of the site to become clear, and so communication
is weakened.
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
