Applied Networks - Challenges to consider for MeshForum

Considering a network-centric perspective is what MeshForum is intended to be all about. By bringing together experts from a wide range of fields, we're seeking to spark conversations, discussions, and hopefully new solutions to old problems.

Towards that end, I am proposing a set of "Big Challenges" which involve Network theory, might challenge pre-conceptions of one (or more) industries, and hopefully might be open to new solutions via the application of techniques, tools, and ways of thinking from other fields. All with, of course, a strong Networks aspect.

First - Commercial Air Travel.

Problem: Much of the commerical air industry even during a time of fairly heavy travel is suffering financially and in the stock market. I'm most aware of the situation in the US, but similar problems and challenges are true worldwide.

Historically the air travel industry has been bifurcated and heavily regulated. The same carriers carry both people and cargo (the exception being some newer airlines such as SouthWest and some alternative airtravel options such as FlexJets and the like which serve small airports mostly). Fares and schedules are the result of massively complex software with a long historical basis. Complex optimization schemes attempt to calculate the "best" solution to a vast network problem - which planes, crews, cargos, and passengers will combine to both "work" as a schedule (i.e. not require more planes, terminal slots, or crew hours than are physically available) and be profitable (i.e. ideally by some means of setting prices result in planes that collectively are profitable to opperate).

These systems are vast, complicated and embedded into many aspects of the travel industry. But clearly, the example of some of the newer carriers points out these are not the only ways to do business - flat fares, simplified seat classes (or even only one class) and a service emphasis can and seemingly do demonstrate a high degree of profitability as well.

So, the challenge is - Can network techniques from other fields help model this multi-varient problem, suggest new alternative solutions, and perhaps offer out of the box solutions to the problem facing the airline industry (and as a result the problems facing many travellers in these days of airline bankruptcy and uncertainty)?

See Doc Searls' post from Dec 6th for a related conversation.

Second - a problem that reoccurs across many areas of networks. How can networks be talked about (and used) when the network is changing more rapidly than it is possible to map and analyze it?

Whether trying to map out a social network made up of people who are interacting (and thus changing opinions of and relationships with each other) or more broadly looking at how a marketing message can spread or how to handle representing and optimizing a distribution network where demand and usage changes over time.

In general terms it can be thought of as a problem of dealing with networks over time - representing networks which change and which are also large enough and complex enough that the time to map and traverse them is less than the frequency at which they change (in at least some respect).

This may also imply that the network links (and perhaps nodes) are not of long duration and stable connections, but are of some form that changes, in at least some aspect, on a regular and probably somewhat frequent basis.

How is this problem dealt with in more controllable networks? Such as IP networks where nodes and/or connections fail on a regular basis and/or are added/modified and yet the whole network has to be navigable and useful at all times.

Third - a challenge related to the second challenge, but subtly different. How can "networks" be worked with, mapped, navigated, and predictions made about them when either connections between nodes are not uniform (i.e. can be represented via different degrees or perhaps as either one way or bi-directional) and when perhaps the "nodes" may be of very different characteristics. Many network problems whether biological, social, or otherwise are loosely talked about using a wide range of possible node types (in biological systems these might be some nodes that represent various chemicals while others that represent cells, elements of cells, drugs, viruses etc) in a social network nodes often are people but frequently groups, departments, locations, organizations, companies, schools, etc may also be included in the mapping.

Does this change the network theory in any important ways?

In upcomming days I'll post other possible "big" challenges for MeshForum to consider. Your feedback, comments, suggestions, and links are very welcome!