Tensegrity
en.wikipedia.org
en.wikipedia.org
Steel is equally strong in tension and compression, and once good steel became cheap, structural design was much less constrained. (Mass production of steel came later than is generally realized. Until about 1890, good steel was about as rare as titanium is now. Steel swords go back a long way, but I-beams don't.)
this is amazing. now I'm strongly resisting the urge to order a couple hundred tongue depressors.
Although these systems can be treated through classical analysis, there's something a bit more compelling about how they are truly faceted in ways that facilitate further nuanced analysis - and not the other way around, where the systems are a product of an analysis. It's a short-circuiting of the classical methods of scientific analysis and somewhat defies the limitation of models that we are so accustomed to.
Instead, they embody an underlying system with dynamic networked coherent responses. I've thought that there's something fundamentally different about them and a way to punch through some of the computational, analytical, and measurement barriers presented by classical analysis on traditional systems.
I've met one person, http://ti.arc.nasa.gov/profile/Vytas_SunSpiral/ who thought the same way I have with a requisite technical background to see the implications of this school of thought, and so far, that is all other than perhaps, Buckminster Fuller in his 1982 book, Synergetics - which deals with the same concept.
I worked on this a few years and then I put it down, about 12 years ago. I hit the end of my cognitive ability in the research and haven't returned. It's a redefinition of analytical engagement. The classical Aristotelian discreteness is a limiting factor in its development. Stepping outside this while maintaining a descriptive formality of what constitutes this form of analysis is exceedingly difficult.
I apologize that this is so vague but it is, for the most part, completely unexplored. I think there's a very real possibility of coming up with a divergent philosophy of science with a completely different disciplinary system - but with higher accuracy and fidelity of analysis and results of some of the problems which are for intents and purposes, currently intractable due to the compounding effects of being multi-faceted systemic problems.
Here we are, we can construct highly-coupled, dynamic, reactive, but ultimately stable systems - the same class of systems we see in our "unformed" natural world. That's the key insight - that this is a real thing and that there is a distinct and different way of dealing with it - and that it can be exploited generally. This I believe, is inherently revolutionary.
If you've read all this and don't think I'm nutso, then congratulations, we can work together maybe. Just respond below. I think this could honestly be the most productive and meaningful project that people could ever engage themselves in.
I agree with you that one of the most difficult things about tensegrity structures is predicting their behaviour under stress and failure - and since behaviour under exactly those circumstances is also one of the main strengths of the concept, that kinda sucks. It's like trying to sell a car which gets better gas mileage, but you can't really predict how much. Better but unpredictably better can be a hard sell and especially bad when trying to optimize a plan for.
I see a lot of parallels between the concept of tensegrity and certain concepts of resilient design of complex software. In both cases, testing (and having good test cases) is a vital component of testing. In both cases, even with testing it is hard to have a good predictive model of failure, because at a certain level the complexity of test cases can start to compete with the design itself.
What I see as the really cool thing about what Vytas is doing, is the addition of controls (e.g. motors etc) to tensegrity structures. This only adds to the complexity when it comes to testing, because now there are many more variables. However, if this is a huge issue, it is theoretically possible to set the controls in such a way that it actually reduces the space of possible behaviours, and thus makes analysis more tractable. Of course, I think it's much cooler to go the other way and make the darn things do crazy stuff, but it all depends upon what you need out of a design. It's somewhat the same issue with a robotic arm: it can move in so many ways, sometimes the easy or safest way to do a job is to restrict the possibilities (like not letting it move into a danger zone). But making it dance is more fun!
http://new.livestream.com/viewnow/NIAC2015/videos/75238510
at 33:03 into the video.