Here are the original Oldenburg notes ... what we did was adding priorities (subject to debates of course) and comments.

undoing Don's mistaken optimisations ... may be a lot of work but actually quite important for language processing

The following two items are probably easy to do.

tough and not clear ... basically our conclusion was that this area would need a workshop of its own to come to useable approaches and algorithms.

This is partly handled by writing absolute positions, so downgraded from priority 1.

While considered worthwhile what is covered in the Oldenburg notes we thought that this need some further details to be worked out.

A nice to have, probably useful and also easy to implement but not really important, as it can be faked with some effort on the macro level.