in reply to Reading PerlMonks offline

There are some practical approaches you can try, but I see two basic difficulties with your core idea.

...I look to a node and discover that it needs to be studied. I give the ID to the supost program, and it fetches the node and show me,...

That suggests a solution based on heuristics about authors and titles and the tree of replies, but prevents decisions based on the content.

The larger problem is that you want machine evaluation of human expression -- understanding of speech. That holy grail has been drifting just out of reach for a very long time.

PM does have XML feeds, including 'newest nodes', so heuristics like I mentioned are possible in limited bandwidth.

After Compline,
Zaxo