Not all computer-supported collaborative learning leads to cultural change, nor to sustained discussion, nor even to dialogue among students (Guzdial, 1997; Hmelo, Guzdial, & Turns, 1998). While we do know approaches to achieve sustained discussion (e.g., Guzdial & Turns, 1998), even that does not insure a cultural change in the classroom. An approach we (my collaborators at Georgia Tech and I) have taken is to reconceive CSCL as authoring, rather than as communications. We are using a new tool called a CoWEb in which both teachers and students can create collaborative activities (Guzdial, 1998; Guzdial, 1999), such as creating a shared Web address hotlist and developing a "Choose Your Own Adventure" game. By allowing students the same power and flexibility as the teacher, agency (Scardamalia & Bereiter, 1991) shifts so that teachers become participants in the students' activities and students become critical consumers of the teacher's activities.

In this paper, I briefly introduce the CoWeb, describe activities in the CoWeb, then describe a survey-based study of students' attitudes in three different classes. The activities are interesting because there is a sense in which agency is shifting from the teacher to the students. Students and teachers have equal powers in the CoWeb, and students can be the instigators of collaborative activities themselves. The survey highlights student attitudes in the CoWeb, and how they might be explained in terms of features of the environment and how the features have been used.

A CoWeb is a collaborative website. The software supporting the CoWeb is built on the Pluggable WebServer (PWS) implemented in the Squeak programming language (http://squeak.cs.uiuc.edu). Squeak, the PWS, and the CoWeb tool, have been run on a wide variety of platforms including Macintosh, Windows-95, Windows NT, and SunOS operating systems. CoWeb (sometimes called Swiki) pages look like any other page on the Internet, except that every page can be edited by anyone and that there is a simple mechanism for creating new pages. There are no security or synchronization checks; if something goes wrong it must be fixed by an administrator, or another user (since anyone can edit anything), or restored from the last saved copy. In the computer world that is increasingly being troubled by unauthorized access, firewalls, and passwords, Squeak is accessible to anyone and so, is extremely vulnerable. But that vulnerability is a trade off for extreme flexibility with no special privileges afforded to faculty vs. students.

Once set up, the CoWeb requires knowledge of just a few commands for operation by students. When editing page on the server, users send formatting commands simply by typing them in the body of the page, alongside regular text. For example, if any text string is put between asterisks (for example, *ABC*), a new page will be created on the server with that name (for example, a page called "ABC"). Subsequently, a link to that page will be inserted into the current page. Graphics and hyperlinks can similarly be created using simple commands. Any HTML that the user does know can be utilized, but none is necessary. Simply separating paragraphs with a carriage return (like writing email) is enough to create well-formatted web pages.

In the below paragraphs, I describe several of the applications of the CoWeb that faculty have implemented with this tool.

• Information source. The first use for many faculty was simply a course website. The CoWeb lends itself to being a course website particularly for those faculty who are uncomfortable with traditional methods of managing a website (e.g., creating and editing files). Further, other students and teaching assistants can easily add information to a course CoWeb. In early surveys with students, "very helpful information" and "easy to access" were among the most often cited benefits of the CoWeb.
  
  There is the obvious problem that information could be changed on the CoWeb by someone not in authority. For example, a student might edit a teacher’s announcement to have it say something different than what was intended. While we received no reports of malicious editing in the CoWeb, there was a question whether the students would trust the information.
  
• Student assignment hand-in. Another common first use for the CoWeb was a "hand-in" site, where students would post their homework assignments when they were ready for grading. While perhaps not appropriate for every class, a public hand-in system provides the students an opportunity to see one another’s work and even comment upon it. In at least one class, serious and contentful class discussions developed out of students reading and posting comments about one another’s essays being handed in via the CoWeb.
  
• Collaborative Writing. In a human anatomy class, students were asked to do collaborative writing projects on the CoWeb. The teaching assistant created a page for each topic that groups could choose from. On each of the pages, the assistant created four or five spaces for signing up for the given topic. Students in a group would edit the same page to enter their text for the group project. In this structure, students were provided a good bit of scaffolding for their collaborative writing. The teaching assistant defined writing topics, a way to sign up for a topic, a place to write, and an easy way to read collaborators’ writing.
  
• Design Review. In an architecture class, students were asked to create two CoWeb pages for each of their projects. On one page, students were asked to post descriptions of their projects ("pin-ups") with scanned images of their drawings. On another page, students were asked to identify research questions that they needed to answer in order to complete their designs, such as the optimal size of hallways for a given kind of building and kind of use. The goal of this structure was to provide students with an opportunity to review each others’ projects and to help one another in answering their research questions. While both of these activities did occur, the collaboration was not as broad-based as the faculty had hoped. On two occasions during the term for this class, expert architects were invited to tour the students’ pin-ups and comment on the projects. For each expert architect, a "tour page" was set up with the architect’s name on it. The architect was invited to visit each of the pin-up pages listed on his or her tour, and comment on the pin-ups either directly on the student’s page or on the tour page. This activity was perceived as being more successful in supporting the students’ learning. The experts wrote a surprising amount of commentary. They wrote sometimes left comments on students’ pin-up pages with particular advice, and sometimes they wrote on the tour page with general advice that the expert felt that the group needed. Students took the reviews quite seriously and worked to respond to the experts’ comments.
  
• Anchored Collaboration. One of common uses for collaboration spaces at Georgia Tech is anchored collaboration, where something to be commented upon (e.g., a paper, a list of exam review problems, an assignment description) is directly linked to a space for commenting on that paper, problem, or assignment. An anchored collaboration is a good structure to use for review activities (such as the design review listed above), but is also useful for supporting focused discussions, such as students studying for a final exam by posting and critiquing answers to sample questions, or such as students asking questions about an (anchor) assignment. Anchored collaboration was particularly simple to implement in the CoWeb, since the collaboration space could be literally the same space as the anchor or a separate page.
  
• Focused Discussions: Students did use the CoWeb for focused discussions. Students used a mechanism of writing their comments at the end of an anchor or comment page, usually signed. While there was no explicit support for tracking "threads" of comments (i.e., when one note comments upon another note, which comments upon another note), students sometimes spread their discussion across several pages, so that each page served as the marker of a given thread, or they would separate threads of discussion with horizontal lines. An interesting artifact of the structure of the CoWeb is that discussions did not always progress linearly down the page. Particularly if the page contained source code for comment, student additions were often placed at the relevant point in the code. Sometimes notes were removed if they were no longer relevant. In a sense, the discussion itself was a collaborative artifact for modification and improvement.
  
• Project case library. In some classes, students were invited to post their homework assignments after grading, particularly if the grade was high. The CoWeb became a project case library for exemplary projects . Students used these projects as positive examples of projects, as sources for ideas (particularly when two or more students posted their unique solutions to the same problem), and, in programming classes, as sources for code that could be re-used in new projects. In a class that I taught in computer science, the project case library became a mechanism for communication across classes. I used the same CoWeb for successive terms of the same class. Students in the second term using the CoWeb reviewed the project cases from the first term and left notes on them. Students from the first term revisited the CoWeb during the second term, answering questions and sometimes changing and improving their cases. The CoWeb thus facilitated a kind of apprenticeship exchange, where more senior students came back to help younger students . In some cases, senior students literally came back to update their pages and engage in discussions with students.
  
• Cross-Class Projects. In one application of the CoWeb, the involvement of junior and senior students was the explicit goal. Two classes in Chemical Engineering were paired using the CoWeb. The Senior level course had students designing a chemical system then constructing a simulation of the system. The Sophomore level course was on analyzing exactly that kind of simulation. Because of curriculum paths, it was possible for the Seniors never to have taken the Sophomore level course. The two Chemical Engineering faculty teaching the classes decided to require a cross-class project where Seniors would create the simulation, pass the data to the Sophomores who would analyze the simulation and return the results to the Seniors, who would use the results to complete the simulation. The CoWeb provided an open forum for sharing data, deciding on formats and other issues for such a technical collaboration, and working together on the solution.
  

Students did create their own pages and features within the CoWeb in a variety of classes. The below list describes several of the interesting pages and features created by students.

• One of the most useful navigation features on my class CoWeb was a "Hot List" at the top of the CoWeb of pages which were particularly useful or on which there were active discussions. I did not create this set of links. One of the students did. While I have maintained the Hot List when I wanted to raise awareness of a page, students have changed it often themselves.
  
• Students used the CoWeb to create special interest pages that invite contributions from others. A page on our CoWeb welcomes other students who are running CoWebs to list their sites for others to visit. Our CoWeb also hosted the College list of players of the network game "TeamFortress," which appeared in the first few weeks of the first term using the CoWeb. One TeamFortress player in the class created the page, with a formatted table listing key player information like the player’s "clan." Other students then filled in their information in the table.
  
• Perhaps the more striking examples of student initiated collaborative activities on the CoWeb are the ones that had little to do with meeting course objectives. In two different courses using CoWebs this last term, students wrote and posted songs about their courses, completely unsolicited. In my class CoWeb, students created an adventure game about one of their assignments, like a "Choose Your Path" book. A student created a situation (based on the simulation assignment that they were working on, started two nights before the assignment was due) with a set of links representing choices that the reader might select from. Other students added to the set of choices and created a variety of pages in the adventure game. Almost three dozen pages were created in this adventure .

A survey of student attitudes and experiences was developed and distributed to three classes during the Winter 1998 quarter at Georgia Tech.

• A 40+ person Biology class that used the CoWeb for collaborative writing. 38 students completed the survey
• A 15 person Chemical Engineering class that used the CoWeb to coordinate cross-class collaboration. Three students completed the survey.
• A 100+ Computer Science class that used the CoWeb for a variety of activities including building a case library, reviewing for exams through an anchored collaboration structure, and discussing programming assignments. 52 students completed the survey.

The first few questions were about the usability of the software. The responses suggest that students found the CoWeb quite easy to use to read, edit, and create pages, and that the students used it often.

Students were asked to respond to the following questions on a five point scale, where 1 is Strongly Agree and 5 is Strongly Disagree.

I also split the data between those that reported that they had created CoWeb pages (49 students) and those that hadn't (44).

Using averages to analyze Likert scale data is certainly not the best way to study this data, but breaking out the distribution may be more data than necessary at this stage of the analysis. It is useful to point out that 33 of the creators agreed that they had their own CoWeb page, and 18 of those Strongly Agreed.

This is a preliminary survey, and one that's not even particularly well-designed. For example, there is often only one statement for each concept, rather than several statements to serve to verify or triangulate around a concept. However, there are some very interesting trends in these data that are worth exploring in future studies.

It's clear that, in general, students liked the CoWeb. Students found it useful, they liked it, they want to use it in other classes, they weren't frustrated by it, and they didn't find reading and writing a chore. (Biology students were almost ambivalent about whether reading and writing was a chore.) In general, they found it easier to share things with a class than email or newsgroups. The usability measures suggest that students found it simple to pick up and use. Students also (on average) found that the CoWeb helped in class task performance and in learning.

The beginning of a shift from a teacher-centered culture can be seen in these data. There was a strong indication that the teacher's involvement was important (i.e., students agreed with the statement "My main reason for using the CoWeb was to get information from my teacher") to the Computer Science and Chemical Engineering students, though the Biology students actually slightly disagreed with the statement on average. Students who created pages in the CoWeb only slightly agreed (2.9 on average) with that statement. Most surprising was that the Computer Science students are ambivalent (on average) about the statement that "I learned more from other students in the CoWeb than the teacher in the CoWeb." That's a shift in attitude from a lecture-based perspective where the teacher is the main source of information, to one where the teacher is valued, but others are valued too.

Students who created CoWeb pages seemed to agree that they "had their own page" in the CoWeb. That sense of ownership is interesting, because it can help to explain the students' motivation in creating the collaborative activities we see them engaging in. However, it's not clear what that "own pages" means. They were almost ambivalent (slight disagreement on average) about whether they felt motivated to maintain their pages. Students who created pages only slightly disagreed that they would come back, which suggests that some of them did in fact plan to revisit their page, as we had observed. We see that as an interesting and unusual cultural shift toward a more apprenticeship model of education.

The issue of the lack of security is an interesting one to explore here. Across classes, both the Biology and Chemical Engineering classes were concerned (on average) that anyone could edit any page, while the Computer Science students were less concerned. Yet all three classes generally found that the information in the CoWeb was trustworthy. Particularly interesting is that students who actually created pages (and thus knew firsthand just how easy it was to create information in this space) were on average slightly more trusting of the information in the CoWeb than students who did not report creating pages. It bothered page-creating students more that anyone could edit any page, perhaps because they had more to lose by their pages being modified.

While preliminary, the results suggest that an unusual tool can lead to some unusual attitudes by students. These data suggest that a tool that leads to no enforced distinctions between students and teachers can lead to students taking ownership of the collaborative space, finding value in the postings of their peer students, and not finding the information in the space less trustworthy. However, it is certainly not the tool alone that leads to this kind of shift. The teacher's attitude and involvement is critical – since so many students were in the CoWeb mostly to hear from the teacher, a missing teacher might lead to less student involvement. The activities encouraged in the CoWeb are also related to the cultural shift. Collaborative writing and case-building by students affords students the opportunity to see the value in others' work.

Thanks to the students and faculty involved in these studies, especially the students helping to build and analyze these tools–the Georgia Tech Squeakers. The figures below and some of the text describing the CoWeb are from a paper by David Craig, Saif ul-Haq, Craig Zimring, Sabir Kahn, and myself.

(FIGURES REMOVED DUE TO SIZE AND SPACE)