Symposium on Complex Systems Engineering/Schedule/Closing session comments
From CSWiki
From Sarah Sheard Here are the cards that were submitted by 3:15 pm, grouped into somewhat arbitrary types. ID Comment Type
5 Capabilities are benefits for the user, while requrements tend to focus on solution implementation. The capabilities are part of an invariant architecture, with requirements being variant architectures. Characterization
7 Complex systems are those that look different at different scales/levels. (self-dissimilarity) Characterization
12 Distinctions on the types of emergence were very interesting. Material/Objective unpredictability/Conceptual Characterization
15 The SE community has not yet bounded acceptable professional competence. This is a barrier to getting Complex Systems knowledge spread and used Characterization
21 Encouraging fact learned: There is an emerging area of Incentive engineering that is beginning to make progress in this direction. Characterization
26 Complex systems is a fuzzvy, controversial, immature field. But who knows where it will go? Characterization
31 Kreitman Provides interesting human organization example related to discovering high leveraged agility points in a complex system. Characterization
33 Norman stimulates the issue of the boundaries of "professional" performance in dealing with complex systems; relationships between economic culture and systems culture. Characterization
36 There is an awful lot of theorizing in this community, and not a lot of examples or experiments. There is especially a lack of large examples. (2 comments) Characterization.
24 There is parallel research in the NASA space program and in the academic system biology community using the software paradigm for "engineering" either physical or biological systems. Both efforts provide insight into architectural principles for complex system design. Connection
32 Maier's SEs taxonomy of complexity/relates to Horowitz taxonomy in integration of economies and technology. Connection
1 Platforms: Bowties and Hourglasses Interesting
8 Platforms that enable evolvability Interesting
3 Using swarm ontology for complex systems modeling to provide understanding not prediction for a "manager flight simulator" Method
10 If (or when) we have developed (and wrung out) a small toolkit of system development tools useful for complex systems, assuming we publicize these well across disciplines and the models created communicate back with the tool developer, then the commonalities across created models can be automatically determined by computer, and this can alert humans to potential commonalities across disciplines Possibility
11 There are more methods than top-down engineering and bottom-up development. In fact, complex systems require use of growth (evolutionary development) plus preferential attachment. We can help engineering by saying which of these methods are useful when, and introducing engineers to relevant modelling tools and examples. Possibility
4 Sensitivity to "emergent" behaviors in robotics and the desire to limit them. The difficulty of getting desired behavior with simple rule sets. Principle (descriptive)
16 Robustness and evolvability are not mutual tradeoffs, but are both achievable. Principle (descriptive)
18 Conservation law relating control over some regimens to lack of control in others. Principle (descriptive)
19 Insight learned: Mission effectiveness models are critical for evaluating progress in developing not-yet-developed systems. Increasing computational power are making these more powerful and cost-effective. More emphasis needs to be placed on developing and using these in complex systems acquisitions. Principle (descriptive)
22 Useful robot swarm programming of sequential operations can be done without sequential control in the programs/robots. Principle (descriptive)
23 Evolutionary computation design optimization usually meets/beats human designers, and does so very rapidly. Principle (descriptive)
28 Bowties and hourglasses are key patterns of architecting systems which make a good compromise between robustness in normal operation and manimizing fragility to evolving threats. Principle (descriptive)
29 What I have learned (in order of importance): Complex engineering strategies; an academic view of adaptation and emergence; The role of complexity in engineered systems. Principle (descriptive)
9 Standard architectures for feedback control across multiple layers that maintain a mix of robustness and fragility. Principle (prescriptive)
13 A variety of definitions of complexity exist, with different usages, all of which have their value. But beware: make sure everyone is on the same page in discussions. Principle (prescriptive)
17 Designing for flexibility, adaptiveness, and robustness Principle (prescriptive)
20 Confirmation: The highest leverage in achieving success in developing complex systems is in improving the fitness landscape or incentive structures for stakeholders. Principle (prescriptive)
27 Self-dissimilarity across multiple scales is a promising quantitative measure for complexity of a system and/or its environent. Anyone involved in analysis of complex systems should become familiar with the work of Wolpert & Macready Principle (prescriptive)
30 Ryan provides an interesting start to defining emergence/needed to provide boundaries that SEs can use to represent their work. Principle (prescriptive)
34 A number of papers discuss metrics for determining the goodness of a model. This is something I haven't considered enough in my own work. Principle (prescriptive)
37 Beware: Definitions of complexity legitimately vary. Make sure all are on the same page. Principle (prescriptive)
2 If the netcentric approach is applied in the military, how is the common structure going to adapt to this non-hierarchal structure? Question
35 We can measure many things/properties/characteristics about a complex system. How do we know if a metric is useful and good? What metrics have been useful in improving our abilit to engineer complex systems? Question
25 The network analysis work brought to the symposium seems to focus on connectivity factors to the neglect of the semantics of information flowing over the network. This is typical of the broader community as observed via Internet published papers. Suggestion for improvement
6 It is hard to say all that I learned. What people are calling SOS? New paradigms, New Collaborations. Suggest record discussions.
14 Seems a number of complex systems folks are doing this on the side, and something else pays the bills! The money system seems not yet to fully appreciate complex systems activities...those activities aren't useful enough yet to earn money, perhaps.
--Mikelkuras 08:29, 12 January 2007 (PST)
As Doug Norman so aptly put it, the symposium has been an exercise in intellectual speed dating. Or as George McConnell put it, the stalking has begun. Tyson demonstrated his mastery of multi scale conceptualization by offering us his atomic and molecular analogy for Programs. Linda has challenged everyone to come to grips with the proposition that the opposite of the complex is not the simple. And that a thing can be neither complex nor complicated – or that it can be one or the other – but it is usually both. Despite the nay sayers, it IS important to define your terms. George has produced the most succinct diagnosis yet of ALL that is ailing us. And on and on. I have my hit list now… But that list is not what I want to share with you “at the end of this day” …
Senior representatives from two system engineering FFRDCs have asserted that their organizations are using the wrong tools, or the wrong problem solving template, to tackle some major acquisition-developments. The shorthand for the tools or template that they are using is an acquisition-development methodology centered on Traditional System Engineering.
They say that they are here because they are looking for the right tools now.
But if TSE is the wrong tool or template for certain problems, why are the FFRDCs continuing to promote and to apply that tool to the problems that are of concern to them even as we are gathered here?
I think I know the answer.
But move past that for a moment.
Let’s just assume for the moment that they have cut the string twice and it’s still too short.
Is the continued use of the wrong tool benign?
FFRDCs are meant to function in the public’s interest.
Is the continued application of TSE to complex-systems benign?
I think that I know the answer to this question too.
But before the answers motivate those in this symposium (and elsewhere), the answers will have to be internalized. You can only tell people what they already know – what they have already figured out on their own. For a community such as this, action flows from internal motivation.
I have offered some strong hints for where to look for the answers to my questions. I will even help those interested in looking. But in the final analysis, motivation and then action will flow from internalized awareness.
--George McC 15:26, 12 January 2007 (PST)
I love the idea of "intellectual speed dating", I think I have missed part of that, but I am grateful for the opportunity to open dialogue with some new friends.
I was surprised to be invited to take part in this symposium. Thank you Sarah. Despite not making it "across the pond" I have learned much, and suspect that there is still more to be gleaned from the information collected here. I hope that I have given a little and at least made someone think in a new way about something.
When going through the papers that I have reviewed there were obvious 'disagreements' in some of the viewpoints, there were things that I wholeheartedly supported, there were things I was extremely wary of, there were things I wasn't sure about and, of course, there were things that I have yet to understand. As Mike's comment above suggests, there are many things I have yet to work out for myself.
One things for sure - we none of us know all the answers yet - but hopefully we all know that because assuming you do know all the answers tends to put you in entirely the wrong place. Complexity, funnily enough, is complex and probably too much for any individual to grasp in its entirety. We need to continually learn.
Complexity requires a multi-disciplinary approach and, historically, we don't do that very well - just look at the regular battle lines drawn in many businesses between engineering and finance and 'human resources' and procurement and .......
If this symposium has done a little to cross those boundaries then I think it can be counted as a success.
Thank you all for the intellectual stimulation.
