33
the Palimpsest model and system. These requirements for collaborative editing infrastruc-
ture, represent the key issues addressed in this dissertation. The overriding goal has been to
eliminate constraints on parallel and simultaneous work imposed in order to ensure system
and data consistency. A variety of studies of writing and collaboration have shown the ne-
cessity of collaboration tools to support authors in what they want to do, rather than to ex-
plicitly control the process that they follow (Beck and Bellotti 1993). A wide variety of dif-
ferent strategies for writing have been observed in behavioral studies, (Galegher and Kraut
1990; Beck 1993; Sharples 1993), suggesting that the best writing support tools are generic
ones that can support authors’ chosen strategies for writing but that do not compel particu-
lar strategies. Experience with structured methods of creating hypertext (Collier 1987; Conk-
lin and Begeman 1987; Conklin 1988; Marshall, Halasz et al. 1991), also suggests that it is
counterproductive to put a priorirestrictions on data and process structure(Marshall and
Shipman 1993; Marshall, Shipman et al. 1994).
Thus, considering only the issues of synchronization and concurrency control already dis-
cussed in this chapter, the following are all features of an ideal framework:
1. It should support flexible transitions from fully-synchronous to fully-asynchronous
work.Authors may need to make changes asynchronously because they may discover
the need to make a change at any time. Conversely, they may decide to collaborate
interactively at any time, if they are both online and have something to discuss. To
the extent that the underlying technology fails to accommodate such flexibility,
authors will be prevented from taking advantage of (or even finding out about) such
opportunities.
2. It should support both synchronous and asynchronous editing.Some approaches to
collaboration infrastructure are strongly oriented to either synchronous or asynchro-
nous collaboration. Since both synchronous and asynchronous work are usually in-
the Palimpsest model and system. These requirements for collaborative editing infrastruc-
ture, represent the key issues addressed in this dissertation. The overriding goal has been to
eliminate constraints on parallel and simultaneous work imposed in order to ensure system
and data consistency. A variety of studies of writing and collaboration have shown the ne-
cessity of collaboration tools to support authors in what they want to do, rather than to ex-
plicitly control the process that they follow (Beck and Bellotti 1993). A wide variety of dif-
ferent strategies for writing have been observed in behavioral studies, (Galegher and Kraut
1990; Beck 1993; Sharples 1993), suggesting that the best writing support tools are generic
ones that can support authors’ chosen strategies for writing but that do not compel particu-
lar strategies. Experience with structured methods of creating hypertext (Collier 1987; Conk-
lin and Begeman 1987; Conklin 1988; Marshall, Halasz et al. 1991), also suggests that it is
counterproductive to put a priorirestrictions on data and process structure(Marshall and
Shipman 1993; Marshall, Shipman et al. 1994).
Thus, considering only the issues of synchronization and concurrency control already dis-
cussed in this chapter, the following are all features of an ideal framework:
1. It should support flexible transitions from fully-synchronous to fully-asynchronous
work.Authors may need to make changes asynchronously because they may discover
the need to make a change at any time. Conversely, they may decide to collaborate
interactively at any time, if they are both online and have something to discuss. To
the extent that the underlying technology fails to accommodate such flexibility,
authors will be prevented from taking advantage of (or even finding out about) such
opportunities.
2. It should support both synchronous and asynchronous editing.Some approaches to
collaboration infrastructure are strongly oriented to either synchronous or asynchro-
nous collaboration. Since both synchronous and asynchronous work are usually in-