The key functionality of a Coordinated Internet would be that the Internet actively watches what people do (analogous to search completion on desktops today), correlates these activities, and actively notifies people when and how their current tasks affect and are affected by the activities of other people. Such a Coordinated Internet would provide revolutionary functionality. Some solutions exist, notably in concurrent engineering such as the Redux solution, but foundational research remains to be done in Coordination Engineering.
Many technologies for the dynamic composition of web services have been
developed in the last 10 years. If we can discover and consume reusable services
with common descriptions, new processes may be dynamically constructed,
automatically, to achieve goals[
Given a goal, for example, to be reimbursed for travel expense, a new process can be created for this purpose, even just one time for one person and one set of expenses. The processes would take into account all of the currently expressed enterprise policies, such as, for instance, what amounts must be approved by managers and the conditions that qualify to be a manager2.
Individuals inside companies, and in cross-company projects, could synthesize workflows and other processes as needed. Companies could manage such processes by stating explicit policies and constraints that such processes would respect, instead of trusting that programmers would properly interpret these policies, and instead of waiting for new policies to be programmed. We call this Enterprise Physics: the universe doesn’t run on workflow, so why should enterprises?
Such dynamic processes should, and could, allow for change: contingencies,
conflicts, opportunities, policy revisions, and outright failure. This is a further
advantage over static workflows.There are supply chain problems, resulting from
a needed change in suppliers, that could not be handled by today’s workflows
[
This raises the question of how global dynamic processes could be managed. 3
Should we achieve the vision of dynamic processes, there is an even more radical future Internet that will then be necessary,: the “Coordinated Internet”. This is a vision of a pro-active Internet that not only facilitates sharing and collaboration, but which actively coordinates humans, as well as various programs, most notably services.
The key functionality functionality for coordination, beyond passive information sharing, is the notification of changes and their effects to the right people, and programs, at the right time. Example notifications are that a conflict has occurred and who is involved, safe and consistent solution options upon request, thrashing warnings, opportunities for synergy, and that some tasks are no longer necessary.
A Coordinated Internet must actively watch what people do (analogous to search completion on desktops today), correlate these activities, and actively notify people when and how their current tasks affect and are affected by the 2 “Web services”, as in the Dagstuhl definition, http://tinyurl.com/webservdef, need not have anything to do with the web but semantics are required even if the implementation is in, say, “apps”. activities of other people. The trick is to do this in a useful way, so that coordination is increased, but people are not annoyed by a “big brother paper clip”.
The effects of such future Internet could be world changing, revolutionizing
not only how companies are managed, but how any large enterprise is done,
especially ad hoc ones, such as global relief efforts for catastrophes[
The Coordinated Internet can be viewed as a the management of a configuration or planning problem worked on by distributed actors. When there are multiple objectives and no single objective function, as there is in the case of large projects with many engineers, the best we can do in terms of optimality is to ensure that no single objective solution can be improved without harming the solution of another: Pareto optimality .
The Redux model of design, as implemented in a subset called Redux’[
Notable characteristics of the Redux model include distinguishing between goals that need to be achieved, and constraints that should not be violated; distinguishing between conditions that affect the optimality of a design decision and its validity; and identifying opportunities, resulting from loss of pareto optimality, as well as conflicts. A set of notifications useful for active coordination is provided in this system. The simplest example is when a subtask has become redundant because the method of achievement of the supertask has changed.
Redux also guarantees no thrashing and identifies safe solutions to conflicts and goal blocks in a distributed environment by management of decision rationales and the justifications produced by conflict resolution. 5
There are other coordination models besides Redux, as well as outstanding
issues not solved by Redux. There is a need for much more research in the field
of coordination engineering. Topics include the proper technologies for detecting
and understanding tasks and filtering notifications so that they are more
useful to people, than not. Understanding how to combine constraint satisfaction
techniques with DDB is a major issue[
Better algorithms for underlying justification graph control as well as better representations of committed actions and sunk costs, and providing appropriate transparency of information in supply chains, are known research issues.
Internet implementation questions involve what part of the coordination
functionality to embed in what Internet layers. For instance, could the message
notification could be handled by smart routers and distributed servers? How
should the “watching” function be implemented in browser plug-ins and mobile
device apps? In Next-Link[
The Coordinated Internet also has deep issues in common with Internet of Things, such authorization to change descriptions of products and services.
The Coordinated Internet provides a rich new topic of research for computer science as well as the potential to radically improves mankind’s ability to manage complex projects.