Since there seems to be a general fielding of ideas for the push towards 1.4, let me once again reiterate my deep concerns of the way SMW handles n-ary data. I am not a programmer, so I'm sure I'm asking a lot without knowing what it costs, but I think I speak for a growing number of SMW users and their desire for functionality.
On my wiki, I have an enormous about of data is simple, two-dimensions: a word and it's form. But if I want to list all the forms a word is used in across all pages, I have to either hand write it, or write regex loops to parse the n-ary data, using vast amounts of server resources.
I hope those of you who code will consider revisiting n-ary in queries and make it more functional. If I win the Lotto, the second thing I would do would be to pay a developer to work on this!