Re: Making progress

The point about a dark triple is that it does not have a corresponding 
assertion. i.e., does not appear the lbase translation of the graph and 
is transparent to satisfaction in the model theory. The reason for doing 
this is *to guarentee* that certain problematic assertions never really 
show up in the semantics, thereby ensuring that we avoid the Peter problem.

So, any mechanism that relies on users remembering to darken the right 
triples or not make statements they really shouldn't have made is a 
non-starter. Similarly, a triple has to be light or dark *everywhere*, 
otherwise we land up in trouble again.

Hope this helps.

guha