NJ > MIT?

Whelp, the first iteration of the "Worse is Better" design philosophy is complete. That is to say, I have a working projective parser that *at the very least* parses my trivial five-word example sentence. While I put a lot of thought into the design of the classes and how the algorithm would proceed, during the coding of this I was surprised to find it working without some of the pieces I planned on using. For instance, a data structure to keep track of multiple dependency spans per set of chart indices, not to mention that my span concatenation algorithm only accounts for left and rightward covered concatenation between adjacent spans, but accidentally performs concatenation amongst spans centered at one point as well. The only problem with this is it introduces some duplicates into the chart, but I always wanted to make the chart cell a set-based data structure anyway, so that'll be fixed shortly. Now that something's working, I'm going to try and go through tonight, comment everything nicely, and commit my first real commit, before attempting to tackle the problem of arity by Monday, the discussion day. As it stands, integration with the current chart parsing interface looks hopeful.