I've been to the site that proposes building the Analytical Engine.
I recall reading about Babbage's work many years ago and it was the
opinion of the author (I don't recall the source, sorry) that AE was
basically unbuildable using 19th century materials and methods. The
reason to the best of my recollection appeared not to be the scale,
but rather simple mechanical considerations; that is, frictional
losses, strength of materials, etc.
Does this view still hold any merit?
--Chuck