Dave: I think the answer is "search"
A reading list defines, among other things, a distributed authority network that asserts that certain pages are "relevant". One can derive a kind of "PageRank" from reading lists, a ranking where relevance is asserted explicitly and deliberately, a ranking that resists spam when outline editors route around junk by simply purging spammy entries.
Thus a search that relies on reading lists will help solve a large and pressing problem: spammy, irrelevant results. It can also attack the personalized search problem: just point your search tool to a different reading list. If done right, it can also vastly accelerate the rate at which good results assume the #1 spot. Is this what Dave's hinting at here?
I've believed this for some time and hope, some day, to work on systems for performing this kind of search. (Phil, this is what I mentioned in your comments here.)
Of course such a scheme works best when both outline writers and search vendors are free to participate. A closed system ala dmoz cannot succeed. So the wider exposure for this idea, the better. Please discuss!
Oh, and thanks to those who have entertained my thoughts on this last autumn, among them Rich Skrenta, Fred Oliveira, Dave Winer, Mike Arrington, Keith Teare, and Gary Flake.
Update: Dave meant feed synchronization, and calls the above "cerebral", which is fair. But that "reading list" searches can scale up to something like general search seems obvious enough to me, and importantly, obvious to someone skilled in the art. And best of all, the end product will be dead simple to users.