An interesting expansion of the "transport control" idea is motivated by some statements in this paper. The idea of "scaling" of something either in time or space could be a generalization of the control element I have been calling "speed". Also, the idea of the use of shortcut functions to facilitate ease of use such as "restart", "skip" etc.

The paper brings forward the idea of categorizing multimedia into primary and secondary depending upon weather or not the multimedia stream is modified by user input. The secondary stream can be dependent or independent of the primary one. If it is dependent then the secondary stream is either synchronized with the primary or events in the secondary occur before the primary. (Not sure what happens in the situation where the interaction between streams is more ad-hoc).

The control of the presentation by the user has some formalized operations:

1. Skipping events

2. Reverse presentation

3. Freezing and Restarting

4. Scaling of speed.

I think playing forward is assumed by the author to be a given. Conceptually, a random access directive can be accomplished by the skip event directive, however, the skip event requires we know how far from the current location is the desired event. This means we need to know the current event, the desired event and the distance between them. On the other hand, a random access feature only requires that we know where we want to go.

This brings up some useful discussions around "transport". While a minimal set of underlying primitives can be constructed that will allow the writing of any necessary control program, from the user's point of view, more intuitive higher-level directives need to be presented. What is useful are both ideas: a primitive set of control directives for programming, and a higher level set of user directives for creative authoring.

In order to accommodate user input into the function of a Petri Net, the author proposes an augmented version of Dynamic timed Petri Nets and Object Composition Petri Nets (OCPN). Some of the paper is devoted to making sure the new model is "safe" within the theory of Petri Nets, however, it seems that the "atomic" element of a Petri net is to large to support adequately the flow of multimedia within each of the network nodes. Instead, it seems like the transitions deal with the beginnings and ends of clips of media rather than the control of frames within that media. As such, as a presentation medium, it works, but as a media performer, it lacks sufficient fine resolution control (This can also be said of a cue list mechanism.)

This distinction between fine resolution and course resolution control is one of the reasons manipulating and coordinating multimedia is difficult. if dependencies between higher and lower levels exist, or between nodes, or through built up interactions, errors in presentation can occur. The recognition of these multilevel dependencies is the key to designing an adequate system.