As I am always on the lookout for user interfaces it was delightful to see the bridge and the engine room of the Queen Mary with truly tangible man-machine-interfaces. Large wheels and all sorts of knobs, cranks, and levers are there to operate the many functions. At a first glance they look nice, intuitive, and easy to use - but are they? Looking closer it becomes clear that many user interface elements were built primarily to enable mechanical control and not because of ergonomic considerations. The size of physical controls (e.g. the length of a lever) is often due to the fact that certain forces must be created to control the machine. The size and visibility of the mechanical properties are helpful for the user to understand the operation and to some extent foresee the impact of the operation. At the same time many of these user interface elements make the operation very visible to co-workers - you do not have to tell that you change the speed as everyone around you can clearly see it from your actions. It seems that ergonomic properties are of lesser importance in many of these controls, e.g. some controls require strong forces or large physical movements. All in all I would conclude that these tangible controls (that are designed due to physical constraints) are helping with the understandability of a user interface but are not necessarily a good model for creating ergonomic controls.
In some recent examples of tangible user interfaces I feel that people took the worst from both worlds - they argue for the physical controls - and all they get are less ergonomic UIs (e.g. needs more forces, movements, etc.) which are less understandable as the physical constraints do not map to the constraints of the digital system… Hence I think it is important to keep this in mind: the physicality of the controls should be used to make interaction understandable and the design should not compromise ergonomics (as we do not live in the mechanic era anymore).
One further thing that can be learned for these mechanical interfaces is the beauty of the design and implementation. Some are in shinny and polished metal, others are nicely decorated, and on others it is just pleasant to touch the wood. Beauty and esthetical qualities play a major role - and we know this for screen based UIs from Noam Tractinsky's work [1,2] as well as from the market success of devices like the iPhone.
 Noam Tractinsky. 1997. Aesthetics and apparent usability: empirically assessing cultural and methodological issues. In Proceedings of the SIGCHI conference on Human factors in computing systems (CHI '97), Steven Pemberton (Ed.). ACM, New York, NY, USA, 115-122. DOI=10.1145/258549.258626 http://doi.acm.org/10.1145/258549.258626
 Tractinsky, N., Shoval-Katz A. and Ikar, D. (2000) What is Beautiful is Usable. Interacting with Computers, 13(2): 127-145.