I have another site to share with regard to the Hanson reading. It was my company's Intranet. While it is not exactly like css-d, it did have news articles with comments feature, much like this class website.

One problem often faced was off-topic discussions, and moderating it was difficult because of the desire to encourage internal communication (ie community participation). What usually happens is the discussion is left to fade on its own (because articles only appear for a week on the homepage by default), and once it's not on the homepage, # comments fall considerably. While there were attempts to direct the offtopic discussion to another thread, it wasnt successful. We could have deleted the postings or censored the posters, but the 7 day limit was enough control. -- Now, the wiki offtopic page sounds like a good idea to implement to offload offtopic discussions, though it seems better suited to Q&A sites.

On the flaming aspect, indeed these people or these behavior exists almost everywhere (forums). In the company's Intranet, we "talk" to the offender directly - identifiable by their machine's IP and login ID.

I guess I can now know how to identify websites with threaded conversations. But I certainly would like to know the answers to the questions raised in the workshop. Or are we the ones who're suppose to answer those Qs? I dont quite get the "hidden intention" of what we need to get out of this article.

I still doubt how much the wiki can really fix these problems: overwhelming, free-ride, low-contribution, leaving of old-timers...I even doubt whether we should take the wiki as the repository of the community, given the fact we consider the mailing-list makes the community itself. To me, it would likely be that there are be different people participating on the two places eventually. They might be connected each other, but, they will be different.
The way of the mailing list is organized (open access) decided people would leave after a while: it is basically a way for a studying group to help each other. When one gets experienced enough she will leave. There is no incentive for one to keep there receiving boring questions. There is polite requirement that one should search before asking. However, everyone knows that it is easier if there will be a person(can be an expert) to answer and there is a considerable cost to search. In addition, newbies always don't know how to search so they come to ask.
I am not sure about the incentive mechanism for wiki things. For voluntary help, i think mailing list or a forum would provide more incentive for people to contribute, simply because the helper's name is seen by everyone. I went to the CSS wiki and saw, like I guessed, it is not so resourceful and organized as we hope (don't know why they put search box down the page). So for a user especially a newbie, is really easy to get lost and does not know where to start to look at, so she can just ask.

The paper on conversation pivots explains the use of software to establish semantic connections between disparate types of data objects.  If viewed in terms of Wenger's principles of cultivation, establishing these semantic connections enhances the focus on value and also evolves the function of the site to suit the [assumed] behaviors of the users.

By providing additional means to navigate the content, conversation pivots are a clear value-add.  The pivots are also being used to solve newly identified problems that have 'evolved' as the community has grown.  These problems include difficulty in searching the multitude of Perl modules currently available and a lack of referencing on the part of contributors.  Neither of these problems existed in the early stages of the community, when fewer modules would have been more easily searched.  As these problems arise, the software functions of the site are being adapted to meet the changing needs of users.  Interestingly, these adaptations are being contributed by users rather than official administrators of the site. 

 "Beyond Threaded Conversation" provides a brief but clear introduction on some of the current asynchronous communication systems that we usually refer to as "threaded discussion". While categorizing different systems, examples are also provided and research challenges in the area are pointed out. I just feel that the categorization seems not independent, by which I mean it seems to me there are overlapping in those categories and they are not from a same perspective, e.g. a system may be categorized both as threads and permanence at the same time or permanence and homogenous view. But anyway, it doesn't matter much I guess since it is only an idea in further research.

 The second article employs the pivot idea to explain the collaborative database and crosslinks. It mainly looks at how unique features of software modules and conversation threads can be used in inferring links between them and how these links can be used not only to help navigate from software modules to related conversation but also from software modules to related modules. Just noted that in Table1, there is no clear explanation what the number 0 and 1 stand for and therefore, while I first come across this table, it is a bit confusing, though quickly in the following paragraph, it is easily guessed out 0 means related while 1 is the opposite.  Overall, this short paper is easy to read and conclusion is clear that conversation pivots will make online collections more useful to users which is worth more research in future.

Beyond Threaded Conversation is a useful overview of threaded discussion features and notes forward thinking (circa 2005) modes of online conversation.

It's interesting to consider which features have become more or less popular. For example, while message permanence is achieved through diffs, version controls, and automatic logs, many current asynchronous communication systems allow rewriting. (blogs, wikis, threads)

The next version of AIM signals the shift toward synchronous communication because it supports real-time IM. While current versions let people see when their contact is typing, real-time IM lets people see what their contact is typing before he clicks "send".  (not a new feature, but it hasn't been used in mainstream IM clients)

More here: http://www.downloadsquad.com/2008/01/11/aim-6-8-beta-includes-real-time-...

Virtual Community Maintenance with a Collaborative Repository" studies the best practices and failures of the CSS-D community email list-serv and wiki. CSS-D was founded in January 2002 and its data was mined through May 2006. The article mostly explores how administrators handle social challenges such as anonymity, lack of social cues, motivating participation, keeping discussion on-topic, and avoiding offensive behaviors.

Of interest to me were the statistics about participation. I was surprised that only 4% are respondible for 50% of the messages. I guess even within the 80/20 percent rule, the 20 percent aren't necessarily equal contributors. Since 90% of the most prolific posters were also long-term members, I paid extra careful attention to methods that encourage member retention.

To handle off-topic posts, I learned politeness and links to where the conversation can appropriately take place will go a long way to encourage good relationships. Also, asynchronous modes of communication that don't automatically generate new content notifications are a good choice for discussions that may not be of interest to everyone.

A wiki acts as a persistent knowledgebase that helps attract new members through search engine results, acquaint them with policies, and educate them about the total level of community knowledge. Further, a wiki can help improve efficiency because frequently asked questions are documented and thus aren't asked as often.

While the continued success of css-d can be awarded to its active membership, the initial (and instant) success in attracting a sizeable community can be attributed to the notoriety of its founder. I'll be interested to read more about how to attract membership without well-known figures participating at the start.

As one of the authors of the "conversation pivots" paper, I'd like to provide more background info and intuitive explaination in order to help you better understand this piece of work.

The idea of "conversation pivots" comes from previous research work, in particular, the CHI2005 workshop "beyond threaded conversations" and the MovieLens project hosted by the CommunityLab. Originally, it's called "Slice and Dice", where the basic idea is to cut threaded conversation into small pieces, and then rearrange them into clusters based on some common properties such as "topic", and finally display the clusters to the audience. For example, suppose you are interested in the Drupal module CCK, then the ultimate goal of "conversation pivots" is to give you all conversation messages that are relevant to CCK while eliminating all off-topic messages. We used the term "pivots" here as a metaphor to illustrate the process -- from a message or item, people can "pivot" to other messages or items that share the same topic.

Our next step is to do some emperical evaluation on whether pivots are indeed useful to users. Then, by combining pivots, recommenders and trackers, we hope to provide another solution to reduce the tensions identified by Hansen et al (2007) in online conversations, such as on-topic vs off-topic, newcomers vs old-timers, etc.

I'm trying to wrap my head around the pivot concept, and from my understanding so far, the difference between pagerank and pivot is that while the former partially considers page links, a pivot considers what pages relate to a page.  Sort of like a trackback, but with less linking and more magic?

Pivots remind me of Amazon citations.  This feature allows people to find books that reference other books.  If someone wants to learn about another field, it may be beneficial to approach new material from a perspective they are already comfortable with.   

I find it interesting that the instructors chose to group the readings this time - in hopes to motivate us to post once, relating our thoughts on that group of readings perhaps?  And yet some people make separate posts, so the implicit intent may not have succeeded.

 I will limit myself to one post (for now).

I thought the css-d paper was an interesting case study.  It illustrates the complexity of this online community, especially those with a growing user base of many different types of users, and mentions a successful (imo) design adaption that gave them another feature of Wenger's CoPs- the shared stored knowledge base they were creating in the wiki.  It also shows that push and pull technologies used in conjunction can form a multi-faceted successful community.

Some things that caught my attention

Hansen:

Interesting that some conversations were offloaded to the wiki. Why did the community do that? I have seen communities where the wiki remained unused in exactly the same circumstances. It cannot be attributed to the ambiguous "value" or "efficiency" of the wiki. The piece does a good job of combining of quantitave and qualitative (with intracoder reliability). Also good to use the investigative approach of grounded theory.

The useful theoretical distinction was made of the wiki as a pull mode vs. the list as a push mode. This raises also the difference between the list archive as a repository (unused) vs. the wiki as a repository (used). Perhaps a collaborative repository is more useful than a threaded one?

Resnick:

Interesting peruse. Some questions:

how long did it take?
how did the names appear?
how did they pick a conference?
like to do one for panarchy

Since human conversation is threaded by definition, can we move "beyond" it? What would "beyond" really mean? (cf. William Burroughs, hypertext literature, etc.)

Zhou:

The use of the term "pivot" is interesting. Does it come from some other source?

--------------------------------------------------------
PHartzog@umich.edu
--------------------------------------------------------
The Universe is made up of stories, not atoms.
--Muriel Rukeyser

I think a theme common to all three readings is that although the Internet makes it easy to communicate, there are challenges to making that communication useful.  Conversation pivots seem like a promising way to find relevant information, and I'd be interested in learning if users of AnnoCPAN find the pivots useful. 

In the real world, we know that we can't control what other people say, so we tend simply to avoid objectionable conversations.  This is more difficult online. The css-d case demonstrates that some of the common problems in online conversation - going off-topic, holy wars, etc - require effort and innovation to remedy.

Now that we're all thinking about online conversation, I'm looking forward to hearing about how the communities we've selected keep conversation useful for their members.

I like how the first article brought up the neverending battle between old timers and noobs. One thing I've noticed in various communities is that noobs will often wear down the old timers until they leave, citing the "golden days" they once knew.

It really is hard for them to coexist together, but I think with good moderators, they can at least stop short of killing each other off. Some communities are intrinsically better at retaining old timers, too. I think a lot of discussions are often repeats of what has been said before. Freshness is definitely an issue for retaining the oldies.

Concerning the Hansen et al. reading, I just wanted to add that many online forums simply "sticky" FAQ or guideline threads to prevent unintentional newbie posts that may cluster up the forums. However, the problem is that many new users simply ignore these threads. As Sean pointed out above, if the purpose of the wiki is simply to prevent holy wars and divert attention to existing, collaborately constructed pages, one might think that not many users would be interested in having ongoing conversations on topics that are better left dead. I thought it was interesting how Hansen et al. describe the use of the wiki as a "pull" method. Given the fact that many stickied FAQs and "New Users Please Read" threads are simply ignored, I really don't see a link to a wiki in itself to be a much stronger "pull" technology. Of course, I also see that this perspective may be a little extreme, as at the end of the day different forums and online communities have extremely different cultures and beliefs.

    *Conversation Pivots and Double Pivots*
       This paper describes the application of Conversation Pivots and Double Pivots on Drupal.org and AnnoCPAN/Perlmonks, which would benefit users a lot from the perspective of the authors.
 I look forward to seeing their public released results. However, I am wondering how the authors know the need of major users as there are huge differences between the need of users and the interest of developers.

    *Virtual Community Maintenance with a Collaborative Repository*
     This case study reports the use of wiki to help meet those social maintenance on online communities.
     Although the authors claims that css-d is a typical help-based online community, it is not clear for me about its unique characteristics, such as the composition of users, compared to other online communities, which might be one reason for its success.
     Second, I am wondering about the drawbacks of e-maillist. If it could not perform as well as wiki, why we still use it with wiki?

    *Beyond Threaded Conversation*
This article opens up a discussion on the study of new phenomenon beyond the property of threaded conversation, which includes the development of LiveJournal, Slashdot and Wikipedia. What are those new phenomenon? How are they different with traditional threaded conversation, why are they quickly developed and attractive? Questions like these become another new research trend.
So, what are the most important current research results and implications? I wish to get answer from Paul’s class. 

 Having participated in a number of technical support communities (like css-d), I can easily sympathize with many of the experiences that are mentioned in Hansen, et al. The wiki provides a common repository for all types of information that are relevant to the community. This makes it much easier to direct individiuals - particularly newbies - to relevant sources to help answer questions.  This, of course, requires another community member to explicitly reply and link to the relevant page. The idea of a conversation pivot provides a new method for automatically creating a "scent" for the information forager. In my estimation, a pivot is nothing new and this seems to be acknowledged by Zhou, et. al. We have all seen pivots on Netflix, Amazon, and any other site that offers a recommender system. I do find it very intriguing to apply these concepts to conversations (threaded, or otherwise). In many of the technical support communities I have participated in, there are often numerous topics which all revolve around the same problem/issue. The problem for the searcher is that they must wade through all these and make some cognitive connection between the information sources. Pivots provide a recommender system for conversations and would greatly reduce this exercise that every searcher must go through when trying to understand a problem. In the case of CPAN, it seems like it was fairly simple to identify where to create conversation and double pivots due to the unique semantic nature of the PERL (:: as a hierarchical naming convention). It seems for this to be successful in other communities, there would have to be even better natural language processing to identify links. 

I think that the idea of mining content for the purpose of finding related article is generally a good idea. When I read this article though I wondered if this method was the best way. If I understand the article correctly the idea is that second order relationships are displayed along with the given article to help the reader find material that they may otherwise miss. One potential downside to this method is that it assumes that the terminology used in an article is the best indicator of the subject or genre that it falls under.

In many cases articles that use the same terminology are unrelated. Also, some terms are so common that everything has some relationship to it. In these cases machines perform poorly at making useful suggestions. The only way to truly discover which articles are related is by asking an expert.

Asking experts to organize all of your articles would be costly and very time consuming. Another approach might be to analyze the articles accessed by certain users and then use that as the relationship to determine which articles are related

This obviously raises some privacy concerns, but on a sufficiently large site the behavior of a given user would become anonymous. This idea is along the lines of Amazon’s “also viewed” book recommendation system. To me it seems like a good way to leverage human intelligence to solve a tricky problem.

I have always imagined that Paradise will be a kind of library.

-Jorge Luis Borges

Most of communication happening online are text based. This naturally forces online communities to have a repositoty for storing content. I wonder how this will apply to activity based communities like gaming community that don't rely on text based communication.
Most of the online communities I have experienced have content at its core. From Facebook, to Ubuntu forums I visit the site primarily to read what others have written. Hence having a good tools and methods to maintain the repository is an important success criteria for online communities. Threaded conversations to wiki is a evolution of the same text based communication that has been happening.

The design of an information environment surely affects the
way in which a community maintains itself. 
As a newbie to world of website design I would be that newbie in the
Resnick article asking and writing something
relatively novice and completely out of place without a clue as to how I may be
disrupting the flow of exchange on said community.

I found this article very interesting in its perspective on
how interactions can truly deplete the structure of a community.  So, if for example, a moderator did not
exist within CSS-D it is fair to say that the community would cease to exist as
the more experienced users would just gravitate to for their own
community.  However, it is worth noting
that there is a fair amount of narcissism within these communities.  In fact, it is these traits that I would
imagine bring out balance.  Without that
narcissist/expert who is willing to help you and answer all of your questions
only after you hear his/her glory stories of how they were the ones that
actually architected the Net these communities would not be self sustaining.  I am not implying all experts behave in this
manner but it is worth noting that they have power and authority within their
respective community. 

As within any community, the measure of how healthy or
disenfranchised it is is reflected on the general adoption of social norms
pertaining to order and good will with an equal share of dysfunction. Without
these 2 elements it just wouldn’t be human.

Of the three papers in this cluster, the threaded conversation

piece interested me the most.  I first noticed it bothered me

when I set up my first RSS reader.  I quickly downloaded all the

news feeds for the topics and causes I care about, then the scholarship

and job sites I follow, then the political actions sites...I did it quickly because I knew exactly what kinds of conversations I wanted in on.  What bothered me about, and bothers me about threading in general, is that loss of stumbling on the news that you normally wouldn't have sought out.  I often learn the most from finding things I wasn't looking for.  In this the era of TMI, I frequently think about how insular our communities of interest have become.  I like the John Lubbock quote "what we see depends mainly on what we look for," I think there are downsides to that.  Threading has the same risks as specialization in a professional field has, the near-blindness to everything else that microcosm of expertise.

Lisa McLaughlin

It was interesting to read about the diverse ways to manage the excess (or irrelevant) information problem
in communities and guiding users to more useful information with greater efficiency and ease. One of the
biggest problems I've faced while visiting tech support forums or forums providing disucssions on programming
is that often the discussion seems to be so fragmented and sometimes you go through a lot of messages to realize that the discussion has reached nowhere.  I think the offloading some of the commonly disucssed things to the Wiki is an excellent way to manage at least part of this problem too apart from the other benefit of provding the "holy wars" content there.

The wikipedia model of information as a node is also interesting because with each response we are filterning or building knowledge. The important thing is that the edited node reaches a more refined (hopefully!) version each time it is edited. Thus it remains consistent while at the same time evolving into a more refined form of itself thus increasing the quality of the information and at the same time reducing the irrelevant information or making the information more relevant. The advantage for the end user is that he does not have to sort through a lot of the process and the intermediate dialogue that went into the formulation of that final knowledge (unless he wants to) but can directly reach the information he desires which is what is required in most cases.

John Blair

The one thing that occurred to me during these readings is that the varied means of online conversations within virtual communities (wiki, elists, listservs, topic/off topic, etc.), and the same method may work in one community and not another - just like in the real world.  It appears to me that all this conversation is based on the participates actually participating and finding some sort of personal / community value by doing so.  If they don't, they won't.  Additionally, I wonder if we should not consider that we simply need a new "language" in which to converse.  Researchers continue to develop algorithms and methodologies for sorting, filtering, organizing posts in a text based medium, when what we really need is a new language built to accommodate this type of interaction.  Just like other real languages where words don't exist and are created or some meanings are better defined as a result of the rules of the language itself.  As an example, the English language suffers terribly from a lake of reliable structures to it, thus allowing many different meanings from the same words and order of those words, which makes it difficult for non-English speaking people to learn it.