Luca de Alfaro wrote:

For one thing, this would delegate spam fighting almost entirely to a cadre of editors: others, even though they are motivated contributors, would not bother manually checking the latest page for every page they read, and thus, they would not discover whether the latest page is altered. The "good samaritan" phenomenon of people casually landing on a page, and fixing it, would be much reduced.

This is a circular argument, though :-) , as the current last edit/default view policy is what makes things such as link-spamming effective. And while I agree discouraging casual contributions is a very important concern, one could also play devil's advocate and point out its drawbacks, particularly the accretion of small nibblets of information on an article until whatever style it once had is ruined.

But in any case I completely agree that more data is needed, and very likely different policies may be warranted across different Wikipedias (or within the same Wikipedia at different stages of its development). Here are some possible determinants of when an article's default view should be its last flagged revision, which lend themselves naturally to configuration options on the extension:

* after an article has been featured * after an article has accumulated X # of stable revisions * if an article is in a poorly covered category (this option should support recursive flagging down the category tree)

More determinants can probably be suggested or uncovered after the extension has been deployed.