Amir Plivatsky

I propose a simpler algo: 1. For each word (like now). 2. For each multi-connector of a chosen disjunct (like now, but also `left` if needed). 3. Scan the link...

> Does this table currently exist, or would it have to be created? It doesn't exist. You can use `string_id()` on the connector name (with or without the `@`) to...

In any case I added this to my TODO list.

> Found 1836 linkages (1000 of 1000 random linkages had no P.P. violations) Random sampling here. Does it matter?

See also `lg_compute_disjunct_strings()`. It is my rewrite of the original one. It handles multi-connectors in a special way. This code should be rechecked. I think a debug flag can be...

> See also `lg_compute_disjunct_strings()` I haven't tried yet to add an option to directly use the chosen disjuncts, but the concept of the current code seems to me solid for...

Can we just ignore the count overflow? As each linkage is a path in the parse-set/parse-choice graph, maybe we can apply a heuristic algo to find the shortest (lower metric...

Here is how I see it: In the counting stage, there are many dead-ends in which one of the multiplications terms is 0 and the other is not but its...

> So, clearly, I've been fooling myself, with what I thought I was doing here. I think so. > What am I doing wrong? A local sorting of disjunct costs...

BTW, I forgot to add that I looked at your `sorting-experiment` changes.