The arrival of the World Wide Web has changed the way many people work. By giving an almost instantaneous access to information distributed all over the world, the Web lets people who have never met collaborate together.

One activity which can benefit from technology facilitating collaboration is collaborative writing, in which people must work together in order to produce a document which can be as short as a letter or as long as a book. The appearance of communication technologies such as the phone, fax, and email has simplified the life of groups trying to work on collaborative documents and has helped reduce the time necessary to produce these documents.

The goal of this thesis is to explore the possibility of using the Web to assist collaborative writing by offering users a Web-based application letting them work on the document and remain in contact. In order to achieve this goal, we begin with a literature review presenting research projects on collaborative writing; we then do a comparative study of the principal collaborative writing systems offered on LAN (Local Area Networks) and WAN (Wide Area Networks) as well as the systems offered on the Web. Finally, we present an empirical study on collaborative writing projects.

Collaborative writing is a complex task which can vary along several parameters. Apart from difficulties associated with the writing task itself, collaborative writing introduces all those problems associated with a group of people trying to work together: how to divide the work, how to understand other peoples' modifications, how to resolve interpersonal conflicts, etc. These types of projects are also extremely variable: the group itself can be modified during the project, as well as the work methods chosen by the group members.

In spite of these numerous difficulties, researchers have tried to develop systems that would support collaborative writing, with more or less success. Unfortunately, these systems are seldom used. There are several explanations for this lack of popularity. One important problem is that the majority of systems have not gone beyond the prototype stage. Other problems include the fact that these systems are developed for a single operating system, making them unusable by people with other types of computers. There are also problems with groupware in general, like the high cost of learning a new, unknown software, the increase in the workload without increased benefits, and the fact that groupware is only interesting if there is already an important number of people using it.

One popular tendency at the moment is developing Web-based applications (sometimes called Webplications) both for individual and group activities. Among individual activities are those associated with electronic banking, such as being able to make transactions on personal accounts, and e-commerce sites, such as buying on-line. Among group activities are sites associated with project management, which may offer tools like group calendars, on-line chat, document sharing, etc., and presentation sites, which let people do PowerPoint-style presentations on-line. Developers are turning more and more towards the Web because, among other advantages, it offers them easy access to an important number of potential clients.

There are several different ways of using the Web to support collaborative writing, and these ways can be classified according to two axes. In the first axis, the categories depend on the place where the document is kept and where the user works on that document. The categories are: (1) the Web as user interface: users work on the document on their personal computers where the document resides at all times and the Web is used to keep in contact with other members of the group; (2) the Web as document storage: the document resides on the Web server but the users import the document to their machine when they need to work on it; and (3) the Web as word processor: the document resides at all times on the Web server and the users work on it through their browser. The second axis divides projects according to where the modifications on the Web infrastructure are done. The categories are: (1) modifications to HTML; (2) modifications to the client machine; (3) modifications to the Web server; and (4) modifications to the HTTP protocol.

Some researchers have already developed Web solutions to support collaborative writing, and some of these solutions are still available at present. Despite the interesting possibilities offered by these solutions, none of them contain the functions that we wish to incorporate into our own system, which are the possibility of writing HTML documents directly on the Web and of presenting these documents in different ways according to the audience reading them. Therefore, our recommendation is to go on to the next step, the design and development of a new collaborative writing system based on the Web.

Before designing a new system, software or otherwise, it is always a good idea to study how potential users accomplish those tasks that the system is supposed to support. This information will help us better understand the real needs of these potential users. There exist many methods to extract this information: ethnographic studies, questionnaires, task analyses, etc. Because of the complexity of collaborative writing, we chose to do a survey, interviewing people who have participated in the past in collaborative writing projects.

The questionnaire that we developed was filled out by 41 people. It turns out that very few people use existing groupware in these types of projects. Instead, the work method that is most popular is to use an individual word processor (Microsoft Word being the most popular) and to maintain contact with the others by using principally email, face to face meetings, and the telephone.

Other results show that the majority of collaborative writing groups are small and that a good number of these groups lose or gain members during the project. The same group can persist over several projects. About two-thirds of groups have a leader, who is most often chosen by somebody outside of the group. This leader's main functions include taking care of the documents, establishing a project schedule, and solving conflicts between members.

The length of these projects varies greatly (from a few days to a year), and the time that each member spends on this project per week also varies greatly. The majority of projects have a due date and a good number of subjects noted that a collective schedule was established for the project.

The majority of groups create a work plan at the beginning of the project, although this plan is frequently modified during the project. People work mostly asynchronously on the document, although a few people reported having worked synchronously at some time or other during the project. Among the most frequently mentioned problems are Internet failures, version control, and those associated with a too large document.

People take part in a collaborative writing project mainly in order to get a better final document, to help with their task or to get social support. However, collaborative writing can make their task more difficult and social relationships can be difficult to handle.

When asked for advice, respondents mainly suggested having someone in charge, establishing and imposing a good plan, and choosing well the group members. As to the ideal tool, participants suggested having synchronous access to the document, good version control, and easy communication.

These answers and past considerations by researchers have helped us establish a list of functions which should be included in the ideal collaborative writing system. These functions can be subdivided into four themes: suggestions concerning the document, the group, the work method, and the technology.